United States v. Sabre Corp., 452 F. Supp. 3d 97 (2024)

C.A. No. 19-1548-LPS PUBLIC VERSION

04-07-2020

UNITED STATES of America, Plaintiff, v. SABRE CORP., Sabre GLBL Inc., Farelogix Inc., and Sandler Capital Partners V, L.P., Defendants.

David C. Weiss, Laura D. Hatcher, and Shamoor Anis, UNITED STATES DEPARTMENT OF JUSTICE, Wilmington, DE, Julie S. Elmer, Scott A. Westrich, Sarah P. McDonough, Robert A. Lepore, Aaron Comenetz, Brian E. Hanna, Craig W. Conrath, Dylan M. Carson, Rachel A. Flipse, Michael T. Nash, Jeremy P. Evans, Vittorio Cottafavi, Seth J. Wiener, John A. Holler, Grant A. Bermann, Erin L. Craig, J. Richard Doidge, and Katherine A. Celeste, UNITED STATES DEPARTMENT OF JUSTICE, Washington, DC, Attorneys for Plaintiff United States of America Joseph O. Larkin and Veronica A. Bartholomew, SKADDEN, ARPS, SLATE, MEAGHER & FLOM LLP, Wilmington, DE, Tara L. Reinhart and Steven C. Sunshine, SKADDEN, ARPS, SLATE, MEAGHER & FLOM LLP, Washington, DC, Matthew M. Martino, Michael H. Menitove, and Evan R. Kreiner, SKADDEN, ARPS, SLATE, MEAGHER & FLOM LLP, New York, NY, Attorneys for Defendants Sabre Corporation and Sabre GLBL Inc. Daniel A. Mason, PAUL, WEISS, RIFKIND, WHARTON & GARRISON LLP, Wilmington, DE, Kenneth A. Gallo, Jonathan S. Kanter, Joseph J. Bial, and Daniel J. Howley, PAUL, WEISS, RIFKIND, WHARTON & GARRISON LLP, Washington, DC, Attorneys for Defendants Farelogix Inc. and Sandler Capital Partners V, L.P.

STARK, U.S. District Judge

David C. Weiss, Laura D. Hatcher, and Shamoor Anis, UNITED STATES DEPARTMENT OF JUSTICE, Wilmington, DE, Julie S. Elmer, Scott A. Westrich, Sarah P. McDonough, Robert A. Lepore, Aaron Comenetz, Brian E. Hanna, Craig W. Conrath, Dylan M. Carson, Rachel A. Flipse, Michael T. Nash, Jeremy P. Evans, Vittorio Cottafavi, Seth J. Wiener, John A. Holler, Grant A. Bermann, Erin L. Craig, J. Richard Doidge, and Katherine A. Celeste, UNITED STATES DEPARTMENT OF JUSTICE, Washington, DC, Attorneys for Plaintiff United States of America

Joseph O. Larkin and Veronica A. Bartholomew, SKADDEN, ARPS, SLATE, MEAGHER & FLOM LLP, Wilmington, DE, Tara L. Reinhart and Steven C. Sunshine, SKADDEN, ARPS, SLATE, MEAGHER & FLOM LLP, Washington, DC, Matthew M. Martino, Michael H. Menitove, and Evan R. Kreiner, SKADDEN, ARPS, SLATE, MEAGHER & FLOM LLP, New York, NY, Attorneys for Defendants Sabre Corporation and Sabre GLBL Inc.

Daniel A. Mason, PAUL, WEISS, RIFKIND, WHARTON & GARRISON LLP, Wilmington, DE, Kenneth A. Gallo, Jonathan S. Kanter, Joseph J. Bial, and Daniel J. Howley, PAUL, WEISS, RIFKIND, WHARTON & GARRISON LLP, Washington, DC, Attorneys for Defendants Farelogix Inc. and Sandler Capital Partners V, L.P.

AMENDED ON APRIL 13, 2020 TO CORRECT ATTORNEY LISTING

OPINION

STARK, U.S. District Judge INTRODUCTION

The United States Department of Justice ("DOJ" or "government") filed this expedited antitrust action seeking to permanently enjoin the proposed acquisition by Defendants Sabre Corporation and Sabre GLBL Inc. (collectively, "Sabre") of Defendants Farelogix Inc. ("Farelogix") and Sandler Capital Partners V, L.P. ("Sandler") (collectively with Sabre, "Defendants"). Sabre and Farelogix both play roles, which are described in great detail below, in the airline travel industry. The government contends that allowing Sabre to acquire Farelogix, and eliminate Farelogix as an independent entity, would harm competition, and thereby violate Section 7 of the Clayton Act, 15 U.S.C. § 18. DOJ contends Farelogix is an innovative disruptor in the market for "booking services," a market historically dominated by just three global distribution systems ("GDSs"), including Sabre, who have tried to stifle innovation in a market in which they earn billions of dollars annually.

The Court held an eight-day bench trial in January and February 2020. (See D.I. 251, 253, 254, 255, 256, 257, 258, 260, 261, 263, 264, 265, 266, 267) ("Tr.") After the trial, both sides submitted detailed proposed findings of fact (D.I. 234, 236) as well as opening and answering briefs (D.I. 233, 235, 241, 242).

"Citations to the trial transcript are in the form: "( [Witness last name] Tr. [page] )."

Pursuant to Federal Rule of Civil Procedure 52(a), and having carefully considered the entire record in this case, the arguments of the parties, and the applicable law, the Court concludes that DOJ has failed to meet its burden of proof. Therefore, the Court will enter judgment for Defendants and against the government. The Court will not enjoin Sabre's proposed acquisition of Farelogix.

PROCEDURAL BACKGROUND

DOJ filed its complaint on August 20, 2019. (D.I. 1) On September 26, 2019, the Court scheduled a bench trial to begin on January 27, 2020. (D.I. 31) After consideration of various requests relating to the length of the trial (see, e.g. , D.I. 147 at 8, 15 (parties initially asking for two-week trial within three months of case being filed); D.I. 175 at 2 (parties requesting 30 hours per side)), the Court ultimately allocated each side up to 25 hours for its trial presentation (see D.I. 197 at 11).

The parties prepared expeditiously and efficiently for trial, raising only two discovery disputes. (See D.I. 127 at 2-3) Trial began, as scheduled, on January 27 and was completed, with a full day of closing arguments and questions to counsel, on February 6, 2020. The Court commends all of the many attorneys on both sides for consistently outstanding performances throughout this litigation and especially at trial.

It should be understood that the Court is issuing this Opinion in the midst of the global coronavirus (COVID-19) pandemic. As readers today (i.e., April 2020) well understand, we are living through a national emergency, in which courtrooms (although not Courts) are largely closed, and most people (including judges, law clerks, lawyers, and assistants) are working remotely from home.
As of this writing, it is generally known and can be accurately and readily determined from sources whose accuracy cannot reasonably be questioned, that the travel industry, and particularly air travel, has been hit particularly hard by the virus. See, e.g. , U.S. Department of the Treasury, Procedures and Minimum Requirements for Loans to Air Carriers ... under ... the Coronavirus Aid, Relief, and Economic Security Act (March 30, 2020), https://home.treasury.gov/system/files/136/Procedures%20and%20Minimum%20Requirements%20for%20Loans.pdf (last accessed April 7, 2020); Tr. Mar. 30, 2020 teleconference at 4-5; see generally Fed. R. Civ. P. 201. Understandably, however, no evidence was presented at trial about the impact of the coronavirus, or how its devastation might itself transform the air travel industry. Therefore, and necessarily, the Court has not considered the potential consequences of the virus in making its findings of fact or conclusions of law. To be clear, the Court's forward-looking analysis does not (and cannot) take into account the current crisis caused by the pandemic.

At trial, the government called the following witnesses, including many who were adverse to DOJ's case:

All witnesses testified live at trial, unless otherwise noted.

• Cory Garner, Vice President ("VP") for Distribution and Sales for American Airlines ("AA" or "American") (Garner Tr. 89)

• Michael Radcliffe, Director of Distribution for United Airlines ("United") (Radcliffe Tr. 169)

• Susan Carter, Senior VP of Marketing for Farelogix (Carter Tr. 235)

• Jim Davidson, President and Chief Executive Officer ("CEO") of Farelogix (Davidson Tr. 364)

• Chris Boyle, VP of Corporate Development and Mergers and Acquisitions for Sabre (Boyle Tr. 484)

• Theo Kruijssen, Chief Financial Officer ("CFO") of Farelogix (Kruijssen Tr. 585)

• Sean Menke, President and CEO of Sabre (Menke Tr. 664)

• Gregory Gilchrist, Senior VP Travel Solutions Group for Sabre (Gilchrist Tr. 748)

• Jorge Vilches, former Senior VP of Airline Business for Sabre (Vilches Tr. 785) (by deposition)

• Chris Wilding, Senior VP Airline Line of Business for Sabre (Wilding Tr. 824)

• Dr. Aviv Nevo, economics expert (Nevo Tr. 878)

• Tom Klein, former CEO of Sabre (Klein Tr. 1072) (deposition)

• Christina Larson, Managing Director of Sales Analytics, Distribution, and Planning for Hawaiian Airlines (Larson Tr. 1108) (deposition)

• Jeffrey Lobl, Managing Director of Distribution Strategy, Delta Airlines ("Delta") (Lobl Tr. 1148)

Defendants, in addition to essentially conducting direct examinations of many of the above-listed witnesses during the government's case-in-chief, called the following witnesses at trial:

• Kurt Ekert, President and CEO of CWT (Ekert Tr. 1178)

• Rose Stratford, Executive VP Global Supply Relations and Strategic Sourcing, BCD Travel (Stratford Tr. 1211) (deposition)

• Werner Kunz-Cho, CEO Fareportal (Kunz-Cho Tr. 1280) (deposition)

• Tim Reiz, Chief Technology Officer for Farelogix (Reiz Tr. 1320)

• Dr. Kevin Murphy, economics expert (Murphy Tr. 1419)

• David Shirk, President of Travel Solutions for Sabre (Shirk Tr. 1580)

• Rocky Wiggins, Senior VP and Chief Information Officer for Spirit Airlines ("Spirit") (Wiggins Tr. 1638) (deposition)

• Tom Gregorson, Chief Strategy Officer for the Airline Tariff Publishing

Company, better known as "ATPCO" (Gregorson Tr. 1667) (deposition)

• Shane Tackett, Executive VP of Planning and Strategy for Alaska Airlines (Tackett Tr. 1698)

• Derek Adair, Managing Director of Revenue Management Development for Delta Airlines (Adair Tr. 1717)

See Carter Tr. 289.

One of the notable features of the trial was the juxtaposition of invariably excellent legal skill demonstrated by the attorneys combined with numerous witnesses who were not credible and/or not persuasive on multiple key points, as further explained below in the Court's findings of fact.

FINDINGS OF FACT

This section contains the Court's findings of fact ("FF"). Certain findings of fact may also be provided in connection with the Court's legal analysis later in this Opinion.

A. The Parties And Proposed Transaction

1. The United States Department of Justice Antitrust Division ("DOJ") enforces the nation's antitrust laws, including by challenging proposed mergers of private companies that may be anticompetitive. See 15 U.S.C. § 25 ; D.I. 1 ¶ 17.

2. Sabre Corporation is a Delaware corporation headquartered in Southlake, Texas. (D.I. 184 (Pretrial Order ("PTO")) Ex. 1 ¶ 2) Sabre Corporation is the ultimate parent entity of Sabre GLBL Inc., Sabre's principal operating subsidiary and a signatory to the merger agreement with Farelogix. (Id. ¶ 3)

3. Sabre's Travel Solutions division contains its Travel Network and Airline Solutions business units. (Menke Tr. 679) Travel Network operates Sabre's global distribution system ("GDS"). (PTO Ex. 1 ¶ 5) Sabre's GDS is the largest in the United States. (PX389 at -437) Airline Solutions sells other information technology ("IT") products for airlines, including a passenger service system ("PSS"). (PTO Ex. 1 ¶ 7)

4. In 2018, Sabre's revenues were approximately $3.9 billion. (PTO Ex. 1 ¶ 9) Most of Sabre's revenues and profits come from GDS booking fees paid by Sabre's airline customers. (Menke Tr. 665)

5. Farelogix, Inc. is a Delaware corporation headquartered in Miami, Florida. (PTO Ex. 1 ¶ 11) Farelogix's majority owner is Sandler Capital Partners V, L.P., a private equity fund, which is also a signatory to Sabre's merger agreement with Farelogix. (Id. ¶ 12)

6. Farelogix is an IT provider to airlines and offers a range of products related to distributing and merchandising airline content. (Carter Tr. 261-63)

7. In 2018, Farelogix's revenues were approximately [Redacted] million. (DX145 at -006) Farelogix Open Connect, also commonly referred to as "FLX OC," generates more than half of Farelogix's revenues. (PTO Ex. 1 ¶ 19)

8. On November 14, 2018, Sabre agreed to purchase Farelogix in a transaction valued at approximately $360 million (the "acquisition," "merger," or "transaction"). (PTO Ex. 1 ¶ 1)

B. Airlines, Content, And Distribution Generally

9. Airlines sell "content" – tickets and ancillary products and services, such as early boarding and seat upgrades – directly to travelers through their websites, call centers, and airport kiosk ticket counters, and also indirectly through travel agencies. (PTO Ex. 1 ¶¶ 4, 22; Stratford Tr. 1215, 1218) "Ancillaries" are "anything beyond the fare." (Wiggins Tr. 1641)

10. Distribution through airlines' proprietary channels is referred to as direct distribution, the direct channel, or colloquially, "airline.com." (Carter Tr. 240; PX025 at -943)

11. Distribution through travel agencies is referred to as the indirect channel or indirect distribution. (Garner Tr. 91-92; Radcliffe Tr. 171)

12. The indirect channel accounts for about 40% of airline bookings today. (Murphy Tr. 1430-31) The vast majority of indirect channel sales – 95% by passenger volume – are made through a global distribution system ("GDS"). (Id. ) The remaining 5% are sales directly between an airline and a travel agency or corporation ("direct connections"). (Id. )

The specific numbers in this paragraph do not appear in the trial transcript; they were shown on Dr. Murphy's Demonstrative slide number 7, which was marked confidential at trial, and therefore alluded to but not discussed specifically in open Court. Instead, counsel, the witness, and the Court looked at their own copies of the demonstrative while the witness testified. Although the demonstrative was not admitted into evidence, the Court does not understand there to be a dispute about the specific percentages recited in this paragraph. If there is a dispute, it is not material to any issue in this case.

An important but confusing fact in this case is that a "direct connect" is part of the indirect channel of distribution, as it refers to a direct connection between an airline and a travel agency.

13. Travel agencies can be divided into online travel agencies ("OTAs") and traditional travel agencies ("TTAs"), including travel management companies ("TMCs"). OTAs such as Expedia and Priceline primarily operate through consumer-facing websites and target leisure customers. (Kunz-Cho Tr. 1281-83) TMCs such as Carlson Wagonlit Travel ("CWT") and BCD Travel ("BCD") manage business travel for corporations. (Ekert Tr. 1178)

14. For many airlines, travel agencies are a critical sales channel, accounting for a significant portion of their revenue. (See Garner Tr. 92 (stating American earns half its revenue through indirect channel); Tackett Tr. 1703-04)

15. Airlines that sell tickets through travel agencies need to communicate with travel agencies in a format compatible with the agencies' internal systems. (Garner Tr. 146-47)

16. When a travel agent creates a booking, the agent typically receives a record of the booking that is compatible with the agency's IT system, which allows the agent to easily manage the booking and provide post-booking services like invoicing and duty of care. (Stratford Tr. 1220-21) (explaining automated processes applied to records and limitations of "passive segments," that is, records of bookings created outside of the GDS)

17. Travel agencies serve their customers by using proprietary systems to shop for and book flights. (Ekert Tr. 1180-83) For example, a travel agency that serves corporate customers may use comparison shopping engines that allow the agency to pull up only those options that adhere to a corporation's travel policies. (Id. ) Generally, travel agencies cannot meet the needs of their traveler customers by searching for and booking flights on airlines' websites. (Stratford Tr. 1215-16 (explaining that booking through airline website does not allow BCD to provide benefits that their customers value); Ekert Tr. 1184-85 (explaining that CWT allows corporate travelers to access full array of airline content))

18. Airline ticket sales may also be characterized based on whether the passenger is traveling for business or leisure. (Kunz-Cho Tr. 1282-83)

19. Airlines typically sell to leisure travelers through the direct channel. (Garner Tr. 101, 123-25) Leisure travelers tend to have less-complicated itineraries and are more price conscious. (Ekert Tr. 1184) Thus, leisure travelers typically purchase tickets through an airline's website, an OTA, or a metasearch engine (e.g., Kayak or Google Flights). (Id. ; Murphy Tr. 1439)

20. Business travelers tend to purchase more expensive tickets with more complicated itineraries (often booking at the last minute and making changes, all of which usually results in higher fares and fees and, thus, more revenue for the airlines), and may be subject to employer-specific travel policies. (Radcliffe Tr. 173; Ekert Tr. 1181-82) For these reasons, business travelers often rely on TMCs to purchase tickets on their behalf. (Ekert Tr. 1183-85)

C. Direct Distribution Channel

21. Airlines prefer that customers search and pay for fares and ancillaries on their websites because – unlike when a traveler uses an aggregated search tool – an airline's website displays only that airline's content, airlines can better control the retail experience on their own websites, and airlines' costs are typically lower when using their own websites. [Redacted] Searching for airfares on the airline's website can, therefore, diminish price transparency and airlines' incentive to compete on price. (Ekert Tr. 1192-93)

22. In 2005, direct distribution accounted for approximately 50% of bookings by U.S. passenger volume. (Murphy Demonstrative at 26) In 2018, direct distribution accounted for 58.8% of bookings by U.S. passenger volume. (Murphy Demonstrative at 7) In recent years sales volume has shifted from GDSs to airline.com. (Adair Tr. 1726; Wilding Tr. 864) As a result, direct channel sales now account for about 50-70% of airline bookings. (Garner Tr. 91; Wiggins Tr. 1640; [Redacted]

The two Murphy Demonstrative slides relied on in this paragraph were not admitted into evidence, but the Court understands the data reported here are not in dispute. Any dispute on these points is not material to any issue in this case.

23. Airlines have several tools to encourage travelers to search for and book tickets through the direct channel. Airlines can make lower price fares available only on their websites. [Redacted] Airlines can also provide an "advantageous retail experience" on their own websites. [Redacted]

24. Airlines may also drive bookings to their websites through metasearch engines such as Kayak and Google Flights. (Tackett Tr. 1701, 1711-12) Metasearch engines attract price-sensitive leisure travelers who are typically searching for the lowest fare regardless of airline, and airlines can require the metasearch engine to send the traveler to the airline's website rather than an OTA to book the ticket. (Tackett Tr. 1701-02, 1712-13)

25. In the early 2000s, the launch of airline websites led to a large volume of indirect bookings shifting to the direct channel, but over the last decade, the "[c]hannel shift to direct distribution has stabilized." (PX245 at -286)

D. Indirect Distribution Channel

26. In the indirect channel, unlike in the direct channel, airline content is sold through intermediaries such as travel agencies. [Redacted] Travel agencies offer customers many valuable services, including access to and the ability to book content from thousands of travel suppliers, including hundreds of airlines. (Ekert Tr. 1179, 1189) This access to content from many travel suppliers allows travel agencies to provide their customers with comparison shopping and fare transparency, which travel agencies consider one of their most significant value propositions. (Stratford Tr. 1243) Accordingly, airlines that use travel agencies can reach customers who seek to do comparison shopping that they cannot do through airline.com. (Lobl Tr. 1171-72; [Redacted]

27. The travel industry distinguishes between two types of travel agencies: online travel agencies ("OTAs") and traditional travel agencies ("TTAs"). (Garner Tr. 100-03; Klein Tr. 1078)

28. OTAs are travel agencies that sell travel primarily via the internet. (PTO Ex. 1 ¶ 23) OTAs cater primarily to cost-conscious leisure travelers who value comparison shopping. (PTO Ex. 1 ¶ 25) Examples of OTAs are Booking.com, Priceline, Expedia, and Fareportal. (PTO Ex. 1 ¶ 24)

29. OTAs are an important and lucrative distribution channel for airlines. (PX237 at -764; Radcliffe Tr. 172; [Redacted] OTAs enable airlines to reach customers they might not otherwise reach through airline.com by allowing travelers to comparison shop across airlines, hotels, cars, and other travel options, a service airline.com does not offer. (Garner Tr. 124; Radcliffe Tr. 172) OTAs also enable airlines to sell tickets to customers in areas where they "don't have a large presence," and to customers they might not reach through the direct channel based on brand loyalty alone. (Tackett Tr. 1703-04)

30. TTAs are comprised of travel management companies ("TMCs") and other brick-and-mortar travel agencies. (Garner Tr. 100-03) BCD Travel, CWT, and American Express Global Business Travel ("Amex GBT") are examples of TMCs. (PTO Ex. 1 ¶ 28)

31. TMCs primarily serve business travelers, which is the most profitable traveler segment for many airlines because they tend to book more expensive airline tickets than leisure travelers. (PTO Ex. 1 ¶¶ 27, 29-30) Indeed, business travelers are often required by their employer to book through a given TMC. (Radcliffe Tr. 172-73) TMCs ensure that business travelers comply with their employers' travel policies, help with duty of care, provide expense reporting and other back-office support, and manage changes to travel itineraries. (Stratford Tr. 1211-12, 1243-47)

32. A GDS is a transaction platform that connects a large number of travel suppliers, such as airlines, hotels, and car rental companies, to a large number of travel agencies. (Carter Tr. 266) For instance, Sabre's GDS connects more than 400,000 travel agencies (Radcliffe Tr. 206) to more than 400 airlines (PX253 at - 166) and "thousands of car companies or hotel companies" (Davidson Tr. 442). These connections allow travel suppliers to distribute their content to travel agencies and create and manage bookings. (D.I. 22 ¶ 22) Airlines rely on GDSs to sell tickets through both OTAs and TTAs. (Garner Tr. 92, 100-03)

33. The GDSs rely on a "traditional" payment model, under which airlines pay GDSs like Sabre a "booking fee" for each flight a passenger takes. (Garner Tr. 107-08) The GDS then keeps part of the fee and gives the rest to the travel agency as an "incentive payment." (Garner Tr. 108; Stratford Tr. 1209-10; see also Murphy Tr. 1435-36 (defining GDS "net fee" as what GDS retains after paying incentives to travel agencies)) Sabre's contracts with the hundreds of thousands of travel agencies within its network typically require Sabre to pay travel agencies an incentive per segment. (Stratford Tr. 1210) While the majority of Sabre's revenue is generated by GDS booking fees (Menke Tr. 663), Sabre also pays out millions of dollars in incentive payments (Shirk Tr. 1603).

34. GDSs provide various services, including "offer creation;" by using airlines' fare, scheduling, and availability information, GDSs assemble the flight options that an airline can provide in response to a travel agent's search. (PX096 at -098; Radcliffe Tr. 190) Offer creation is also called "constructing an offer" or "shopping." (Garner Tr. 97)

35. GDSs such as Sabre also provide "normalization" and "aggregation" services by combining flight options from various airlines, thereby allowing travel agencies to request and receive offers from multiple airlines. (PX237 at -752; Garner Tr. 92; Menke Tr. 718-19) As BCD's Stratford explained, the aggregation capabilities of GDSs are their "number one value proposition" because "customers want us to shop and compare." (Stratford Tr. 1243)

36. Although airlines cannot distribute their full range of products and services through Sabre's legacy GDS technology (Garner Tr. 93; Radcliffe Tr. 177-78; Adair Tr. 1719), GDSs allow airlines to send offers to travel agencies, create bookings when agencies select an offer, and manage any changes to those bookings (Lobl Tr. 1150; Adair Tr. 1721-22). GDSs also provide travel agencies additional services and technology, notably mid- and back-office software. (Radcliffe Tr. 203; Ekert Tr. 1193-94)

37. The reach and distribution capability of GDSs allows airlines to distribute their products and services globally in seconds, without spending millions of dollars on marketing. (Stratford Tr. 1241) It is critical for all travel agencies that GDSs can provide this aggregation function in sub-second response times. (Kunz-Cho Tr. 1299)

38. In addition to providing real value to airlines and end-user travelers, the intermediary players in the indirect distribution channel – especially GDSs and travel agencies – have strong financial incentives to preserve the travel distribution ecosystem and maintain it in its current form. GDSs earn billions of dollars each year by charging airlines for each passenger segment booked through their GDS. (Garner Tr. 107; Menke Tr. 665; PTO Ex. 1 ¶¶ 9-10) Travel agencies earn a sizable income each year from incentive payments from GDSs for booking airline tickets through the GDSs. (Stratford Tr. 1209-10; Ekert Tr. 1203-04; PX092 at - 622)

E. Sabre's Competitors, Including Other GDSs

39. Sabre's GDS is a two-sided transaction platform. (Nevo Tr. 964; Murphy Tr. 1422-23)

See also US Airways, Inc. v. Sabre Holdings Corp. , 938 F.3d 43, 58 (2d Cir. 2019).

40. As Dr. Murphy explained, "the value of the Sabre [GDS] ... occurs on two sides, where on the one side" travel agents value Sabre's GDS "because it gives them access to a wide range of travel suppliers in a single spot ... Because it's attractive to travel agencies and travel agencies want to use Sabre to do their side of the business, it's therefore attractive to airlines, because airlines after all want to sell their product and they want the customers." (Murphy Tr. 1423; see also Murphy Tr. 1425 (GDS "platform provides value to two sets of customers – the travel agencies and travelers on one side and the travel suppliers, including airlines, on the other")) 41. Sabre has controlled around 50 percent of the airline bookings made through travel agents in the United States. (Nevo Tr. 931, 947; PX389 at -437)

42. Because of the two-sided nature of Sabre's GDS, the primary competitors for Sabre's GDS "in the long run" are "the rival GDSs," Amadeus and Travelport. (Murphy Tr. 1433; see also Stratford Tr. 1250; PTO Ex. 1 ¶ 31)

43. Chris Wilding of Sabre testified that "travel agencies, when they're looking to a GDS, they want to make sure that the GDS can provide them with all the relevant content." (Wilding Tr. 860) Ekert of CWT similarly stated that if "Sabre was to lose content or enable content but on a noncompetitive basis ... then we have the ability to shift business away from them to another GDS." (Ekert Tr. 1186)

44. As Dr. Murphy explained, "[i]f [Sabre] do[es]n't provide [the airline content that] customers want, they'll lose travel agencies to Amadeus and Travelport, and if they don't provide what the airlines want and they don't get the airline content, that will cause them to lose on the travel agency side as well." (Murphy Tr. 1433)

45. Sabre's 2018 10-K filed with the U.S. Securities and Exchange Commission ("SEC") identifies "other GDSs" as competitors, but also lists "local distribution systems and travel marketplace providers primarily owned by airlines or government entities and direct distribution by travel suppliers." (PX251 at -160)

46. In addition to competitive pressures arising from other GDSs, airlines' direct distribution channels, including airline.com, constitute the "most important short-term constraint" on Sabre's GDS fees. (Murphy Tr. 1433) While airline.com is "relevant to both TTA and OTA bookings ... it's especially important for OTA bookings" because "direct channel is a mouse click away" (Murphy Tr. 1434; see also id. 1438-39; Adair Tr. at 1726 (testifying that Delta.com is responsible for roughly 50 percent of Delta's distribution and "volume has gone away directly from GDSs").

47. Chris Wilding, the Sabre employee responsible for GDS contract negotiations with airlines, testified: "Airline.com is one of the primary competitors that we face as a GDS." (Wilding Tr. 860)

48. Because airline.com, including the use of airline.com fostered by metasearch engines, competes for bookings with travel agencies, it poses a significant constraint on Sabre's GDS fees, as airlines can undermine the Sabre GDS by withholding content that the airline will only provide through its own direct distribution channel. (Murphy Tr. 1437-38)

49. The competition Sabre faces from both airlines' direct channels and the other GDSs is apparent from the decline in Sabre's "net fee" – i.e., the difference between the GDS fee charged to airlines and incentives paid to travel agencies. (Murphy Tr. 1435-36) GDS fees have declined since 2000. (Murphy Tr. 1435-36; see also Stratford Tr. 1250 (testifying that GDSs compete for BCD's bookings through incentives))

50. A sizable portion of travel agencies' revenue, sometimes around 10%, comes from GDS payments. (Ekert Tr. 1203; DX306 at 5; see also DX306 at 1)

51. Among the GDSs, Amadeus offers a New Distribution Capability ("NDC") application programming interface ("API") solution (Carter Tr. 270; Shirk Tr. 1629-30), but Travelport does not (Davidson Tr. 475).

52. To switch GDSs, travel agencies may have to retrain their agents on how to use the new GDS platform. (Stratford Tr. 1216-17) Travel agencies also often rely on mid- and back-office systems provided by their GDS, so moving away from their GDS would require a significant adjustment to their workflows. (Ekert Tr. 1193-94)

53. Travel agencies also face significant switching costs in transitioning customers from one GDS to another. (Ekert Tr. 1185-86, 1205-06, 1381 (explaining it took CWT [Redacted] months and cost [Redacted] to migrate business between GDSs); Stratford Tr. 1216-18)

54. Sabre's GDS contracts with airlines may limit how well airline direct connects can compete with it. For example, Sabre has had contracts with [Redacted] Likewise, Sabre's agency contracts also have other financial elements, like bonuses or technology grants, that incentivize agencies to book through Sabre. (Ekert Tr. 1381-83)

55. Due to these incentive structures and switching costs, most travel agencies use a single GDS in a given geography or for a particular corporate client. (Garner Tr. 119; Ekert Tr. 1185; Stratford Tr. 1216)

56. As a result of these market dynamics, airlines must distribute through all three GDSs to reach the entire universe of travelers booking through travel agencies. (PTO Ex. 1 ¶ 32)

57. Thus, while GDSs compete with one another, and their competition places some constraints on Sabre's ability to raise prices, that competition is also constrained.

F. New Distribution Capability ("NDC")

58. The New Distribution Capability ("NDC") application programming interface ("API") was initially developed at Farelogix by its Chief Technology Officer, Tim Reiz. (Reiz Tr. 1324-25)

59. An NDC API is an API that communicates using the NDC schema to enable other technology systems or third parties – such as GDSs, travel agents, or an airline's own public-facing website – to access an airline's passenger services system ("PSS"). (Carter Tr. 264-65; Reiz Tr. 1325-27) A PSS houses IT infrastructure critical to an airline's operations, including its reservations and inventory systems. (PTO Ex. 1 ¶ 8) Hence, an NDC API is one type of API that enables airlines to communicate offers and orders between the airline's PSS and third parties. (Stratford Tr. 1253-54)

60. In or around 2012, Farelogix donated the schema it had developed to the International Air Transport Association ("IATA"). (Carter Tr. 277-78; Reiz Tr. 1325; PX096 at - 097; PTO Ex. 1 ¶ 58) Later in 2012, IATA approved the new standard and labeled it NDC. (Carter Tr. 277)

61. Since 2012, IATA alone has controlled the development of NDC as a public and open standard, releasing two new versions on its website every year since it was first published in 2015. (Garner Tr. 150; Carter Tr. 278-79) As a result, NDC today "looks nothing like" the original schema donated by Farelogix in 2012, and Farelogix must now expend additional resources to "keep up" with NDC. (Carter Tr. 278)

62. NDC is not patent-protected; instead, the NDC schemas are open source and are available on IATA's website to any third-party company to implement and use. (Carter Tr. 271; Reiz Tr. 1327) According to IATA's website, more than 160 companies are NDC certified, and it is realistic for any programmer to become certified. (Reiz Tr. 1327-28)

63. In late 2017, IATA released version 17.2 of the NDC schema, which was the first "commercially viable" version of NDC. (Garner Tr. 150; see also Radcliffe Tr. 208-09; Davidson Tr. 433) This led in 2018 to a significant increase in the number of airlines looking to adopt an NDC API. (Radcliffe Tr. 209; Carter Tr. 287-88)

64. In mid-2018, American Airlines announced it would offer an NDC-enabled corporate bundle – essentially, a package of fares and ancillaries customized to a particular company's needs with pricing consideration. (Carter Tr. 286-87; Davidson Tr. 424) United Airlines announced a similar bundle shortly thereafter. (Carter Tr. 286-87)

65. NDC is an XML-based messaging schema designed to facilitate electronic communication between airlines and third parties for the purpose of distributing airline content. (PTO Ex. 1 ¶ 15; Reiz Tr. 1324-26) As an XML-based schema, NDC differs from EDIFACT – the schema principally used to distribute airline content in the indirect channel today – because it can accommodate varied and complex airline content. (PX237 at 80-81; Davidson Tr. 444-45; Reiz Tr. 1324-25; PX246 at - 981 (NDC "allows for enhanced communications between airlines and travel agents"))

66. NDC allows airlines to present more ancillary products in a more attractive way, including the potential to enable more personalized offers that bundle fare price with other products, which can increase airline revenue and consumer choice. (Wiggins Tr. 1648-49; Carter Tr. 261-62, 276) NDC gives travelers greater choice and enables airlines to increase their sale of ancillary products. (PX104 at -707 to -708; Radcliffe Tr. 178; Wiggins Tr. 1648) Hence, NDC allows "increased airline product innovation and consumer choice." (Davidson Tr. 395; see also PX104 at -707; PX197 at -917, -919 to -920 (outlining new types of offers available through next-generation technology)) In this way, greater use of NDC promises to increase revenues for airlines and travel agents.

67. NDC also shifts two key functions away from the GDS. First, unlike legacy GDS technology, NDC allows an airline to create its own offer instead of relying on the GDS for offer creation. (Radcliffe Tr. 190; Carter Tr. 249 ("The whole point of NDC is for the airline to have more control over its offer."); Davidson Tr. 385; Lobl Tr. 1151; Adair Tr. 1719) By shifting offer creation from the GDS to the airline, NDC gives an airline greater control over the distribution of its products. (Davidson Tr. 385; Adair Tr. 1719; PX300 at -260; PX025 at -953 (Farelogix marketing materials stating NDC API gives airline "[f]ull control over distribution and channel management, with less dependence on GDS/PSS")) Second, NDC enables airlines to decouple order management from the GDS bundle, allowing airlines to reduce their distribution costs. (Garner Tr. 107; Radcliffe Tr. 175; PX300 at -253 [Redacted] ) In these ways, greater use of NDC promises to decrease airlines' costs.

68. Although some airlines still tout the costs savings generated by use of NDC, more airlines view NDC technology as being beneficial primarily because of its ability to increase revenues, not cut costs. (See Carter Tr. 265-66; [Redacted] Wiggins Tr. 1648) [Redacted]

69. By enabling airlines to totally or partially disintermediate the GDS, NDC poses a threat to Sabre's traditional business model. (Adair Tr. 1719, 1736-37; PX002 at -885 ("NGD [Next Generation Distribution] shifts control to airlines and weakens the GDS value proposition;" "as an airline gains control of offers, the value of the intermediated channel will decrease"))

G. Farelogix's FLX OC And Its Uses

70. Farelogix sells a suite of IT solutions for airlines. (PTO Ex. 1 ¶ 13) 71. Farelogix has no travel agency customers and no commercial relationship with travel agencies. (Carter Tr. 266-67; PTO Ex. 1 ¶¶ 48-49; PX096 at -097 ("Airlines are our only customers; No travel agency subscribers"))

72. Farelogix sells an NDC API product to airlines that it calls "Open Connect," which is also referred to as FLX OC. (Carter Tr. 236, 264, 266)

73. Farelogix refers to FLX OC, which is at the "core" of its business (Davidson Tr. 366, 382), and its NDC API as its "order management" or "order delivery" product (Carter Tr. 238; Davidson Tr. 405; see also PX072 at -232; PX025 at -949).

74. FLX OC consists of two main components: an NDC API, which provides the "pipe" that carries messages between the airline and the travel agency or other third party; and an orchestration layer, which standardizes and normalizes the content transmitted between the airline's and third party's internal systems. (PTO Ex. 1 ¶ 17; PX025 at -948, -949, -951; Carter Tr. 238-39, 243, 246-47; Reiz Tr. 1348-49) Farelogix always sells Open Connect and its NDC API together. (Carter Tr. 264)

75. Since donating the NDC API schema to IATA, Farelogix must now retrofit legacy technology to match the latest NDC versions. (Carter Tr. 279; Davidson Tr. 420) Moreover, Farelogix must maintain several different aging NDC API connections for legacy customers, going back to versions of NDC from 2015, which is costly. (Davidson Tr. 421; Reiz Tr. 1331-32) Because support for these connections is "much more labor intensive," Farelogix must devote more employees to FLX OC. (Reiz Tr. 1339)

76. Farelogix holds no patents and has no trade secret protection covering FLX OC. (Carter Tr. 271; Davidson Tr. 417)

77. In the direct channel, airlines can use FLX OC to distribute directly to travel consumers, including through the airline's own website or a mobile app. (PX081 at FLX-000704047; Carter Tr. 283) Two of Farelogix's current customers use FLX OC for their own websites. (Carter Tr. 283)

78. Airlines can also use FLX OC in the indirect channel, to reach travel agencies in three ways. (PX081 at - 046; PX025 at -943; Radcliffe Tr. 178-79) Through each of the three booking paths, Farelogix allows an airline to send offers, receive bookings, and make changes to those bookings. (Nevo Tr. 886-87; Radcliffe Tr. 179-80)

79. First, FLX OC enables airlines to establish "direct connects" with travel agencies. (PX081 at -047; Carter Tr. 283) A direct connect is a link directly between an airline and a travel agency without an intermediary such as a GDS (and, hence, without paying GDS fees). (Radcliffe Tr. 175; Tackett Tr. 1705-06; Adair Tr. 1724; PX025 at -953) Thus, a direct connect is part of the "indirect channel" of distribution and is a form of "GDS bypass." (Garner Tr. 106-07; Carter Tr. 240; PX084 at -887)

A travel agency can access content from an airline's direct connect via a Farelogix proprietary user interface, known as SPRK, or the agency's own aggregation platform. (PX025 at -952; Reiz Tr. 1349-50; see also Garner Tr. 99-100)

80. Second, FLX OC enables airlines to connect to travel agencies via a non-GDS aggregator, such as Travelfusion. (Murphy Tr. 1494; see also, e.g. , Stratford Tr. 1221 (explaining that BCD uses Travelfusion to access Lufthansa content)) Distribution through a non-GDS aggregator is another form of GDS bypass. (Carter Tr. 241-42, 284; Murphy Tr. 1494) 81. Third, FLX OC enables airlines to connect to travel agencies who continue to use a GDS as an aggregator. This delivery path, referred to as "GDS passthrough," enables an airline to use its Farelogix NDC API to push NDC content to travel agencies using GDS aggregation. (Carter Tr. 284; Davidson Tr. 367-68; PX084 at -887)

82. Farelogix is agnostic as to whether an airline chooses the GDS bypass or GDS passthrough delivery path; either pathway increases the number of bookings made through Farelogix's NDC API and, therefore, increases Farelogix's revenue. (Radcliffe Tr. 202; Carter Tr. 265; Davidson Tr. 367-68) Farelogix does not distinguish between GDS bypass and GDS passthrough in how much it charges its Open Connect customers per ticket. (Davidson Tr. 368)

83. When an airline books through FLX OC, whether in GDS bypass or GDS passthrough, Farelogix provides the order management services. (Garner Tr. 118; Nevo Tr. 886-88; see also Carter Tr. 238, 244-45)

84. FLX OC is not a GDS. (Radcliffe Tr. 202; Carter Tr. 266) FLX OC "do[es]n't aggregate content," and offers only "one airline per connection." (Carter Tr. 266-67) FLX OC is "an input to the airline that the airline can then use to connect to customers or connect to travel agents .... Farelogix is not a platform. It doesn't bring the set of customers to any airline or other travel supplier." (Murphy Tr. 1425)

85. FLX OC enables airlines, like AA and United, to distribute differentiated content to Priceline, Orbitz, and other global OTAs (PX025 at -954), and enables these airlines to distribute corporate bundles through TMCs like AmTrav and TripActions (Radcliffe Tr. 195-96; PX033 at -340, -342). As United's director of distribution explained, Farelogix direct connects enable airlines to put "interesting content" into the market faster than is possible with the GDSs. (Radcliffe Tr. 174-76)

86. AA estimated that it is 80-90 percent cheaper to book through a Farelogix direct connect than through the Sabre GDS. (Garner Tr. 107; PX453 at -969)

87. Even in GDS passthrough, FLX OC enables airlines to push for lower booking fees because the airline creates the offers and Farelogix handles order management instead of the GDS. (Radcliffe Tr. 190-91; see also Wilding Tr. 848-49; PX496 (noting that airlines want to use NDC to "drive down cost and use as leverage with GDSs")) [Redacted] see also PX085 at -595 (over time, airlines' NDC APIs will "fully replace their existing legacy GDS connectivity," which may lead to lower GDS booking fees); PX300 at -260; Radcliffe Tr. 190-91)

88. Airlines using FLX OC can also gain leverage vis-a-vis Sabre by threatening to withhold content from Sabre by distributing it only through direct connect. (D.I. 228 ("Judicial Notice Order") ¶¶ 13, 24; see also Murphy Tr. 1570-71 (explaining that airlines can threaten to withhold content to negotiate better deal with GDSs)) Sabre's CEO acknowledged that the value of Sabre's GDS decreases when an airline withdraws content from the GDS. (Menke Tr. 672; see also PX156 at -953 ("Loss of content devalues GDS and causes uncertainty"))

89. In its SEC filings, Sabre acknowledged that airlines with "direct connect initiatives" can use the threat of direct connects to "apply pricing pressure" and negotiate GDS contracts that are "less favorable" to Sabre. (PX251 at -162; see also Menke Tr. 667-68 (discussing this risk and admitting that "pricing pressure is competition"); PX173 at -232; PX251 at -162)

90. U.S. airlines first used Farelogix and other GDS alternatives as leverage during the 2005-06 round of GDS negotiations. (Nevo Tr. 939-40, 1062) Dr. Nevo determined that airlines were able to obtain better prices from GDSs in 2006 than in 2003, which he attributed (at least in part) to leverage gained from GDS new entrants ("GNEs"), including Farelogix. (Nevo Tr. 938-40) Defendants' economic expert, Dr. Murphy, wrote in a 2012 white paper submitted to the Department of Justice on behalf of Sabre that [Redacted] had "leveraged the introduction of non-Sabre distribution channels in 2006 to reduce its booking fees." (Judicial Notice Order ¶ 4) By the next round of negotiations, in 2011-12, Farelogix was the only new entrant remaining in the market, but airlines maintained their gains from the 2005-06 negotiations. (Nevo Tr. 942-43; see also Radcliffe Tr. 170-71 (noting that G2 Switchworks was "purchased by Travelport and shut down")) Sabre's former CEO testified that airlines put Farelogix and direct connects on the table in their 2013 negotiations with Sabre as "something that they would trade" for things that "would benefit them." (Klein Tr. 1080-82)

H. Other Airline Travel Industry IT: PSS And FLX-M

91. A PSS is an airline's system for managing its operations, including inventory, departure control, check-in, and other functions, (Davidson Tr. 450)

92. An airline's PSS ordinarily includes three interrelated systems: an airline reservation system, which controls the sale of seats, scheduling, passenger name records ("PNR"), and the issuance of tickets; an airline inventory system, which provides information on available seats; and a departure control system, which is used to check-in passengers at the airport. (Davidson Tr. 450)

93. Sabre's main competitor for PSS modules is Amadeus. (Menke Tr. 717; Gilchrist Tr. 770)

94. Sabre provides non-core PSS modules that are PSS dependent and can, therefore, only be used with the core Sabre PSS. (Gilcrhist Tr. 780; Shirk Tr. 1632)

95. Farelogix provides only non-core PSS modules. Unlike Sabre's PSS modules, "the Farelogix solution is PSS agnostic, meaning it can attach to pretty much anybody's PSS system." (Gilcrhist Tr. 780-81)

96. Farelogix also sells four non-core PSS "offer management" modules (also referred to as "offer engines") that help airlines create combinations of itineraries, fares, and ancillary products to offer travelers. (PTO Ex. 1 ¶ 20; PX025 at -945, -956, -974, -979, -985; PX086; PX100 at -054 (offer engines help drive greater adoption of Open Connect); Carter Tr. 239; Davidson Tr. 479-80) Farelogix's offer management products are distinct from FLX OC. (Carter Tr. 239; Davidson Tr. 383)

97. Farelogix's primary offer management product, FLX Merchandise (also known as FLX M), is a tool that helps airlines create offers for ancillary products. (PTO Ex. 1 ¶¶ 4, 21; Carter Tr. 277; Reiz Tr. 1337-38) Farelogix's other non-core PSS modules are FLX Shop & Price, FLX Schedule Builder, and FLX Availability Calculator. (PX025 at 5; Carter Tr. 239)

98. FLX M is currently market leading and can drive hundreds of millions of dollars in additional revenue to airlines that use the product. (Tackett Tr. 1710; [Redacted] An airline can choose to use FLX M in the direct channel (see, e.g. , Tackett Tr. 1705) – that is, airline.com – and also in the indirect channel, if the airline has a content distribution API (see, e.g. , [Redacted]. 99. FLX M is sold separately from FLX OC, although airlines can buy multiple Farelogix products together. (Carter Tr. 281-82)

100. Sabre's merchandising solution is not competitive. (Kruijssen Tr. at 629 (Farelogix's CFO stating that FLX M is "better than Sabre's merchandising engine"); Boyle Tr. at 552-53 (explaining that Sabre "didn't have a compelling product" that could "create an offer with a flexible rules engine"); Gilchrist Tr. at 771 ("[Redacted] )

I. Farelogix Has Historically Been An Innovator While Sabre Has Resisted Change

101. AA's VP of sales and distribution strategy has described Farelogix as "the GDSs' leading competitor/agitator for years." (PX452; see also Garner Tr. 130 (describing Farelogix as "lone disruptor" and "lone alternative to GDS distribution")) Similarly, a Delta executive described Farelogix as "a disruptor in the industry" that is "forcing the GDSs to innovate and be more responsive to airline and agency commercial needs." (Adair Tr. 1720-21, 1735-36; see also [Redacted] United's director of distribution testified that Farelogix is [Redacted], which "keeps GDSs on their toes relative to innovating to keep up" (PX299 at -770).

102. In 2005, Farelogix introduced an aggregation platform for travel agencies that managed content from both GDSs and direct connects. (Reiz Tr. 1356-57) As this platform began to gain traction, Sabre launched an initiative to "shut down" Farelogix. (PX096 at -108, -119; see also Davidson Tr. 400)

103. In 2009, Sabre and Amadeus canceled their developer agreements with Farelogix, cutting off Farelogix's feed to GDS content. (Reiz Tr. 1357-58) This forced Farelogix to shut down its travel agency-facing business. (Reiz Tr. 1359)

104. In and around 2011, Sabre and the other GDSs tried to stop travel agencies from establishing direct connects with airlines using Farelogix's technology. (Davidson Tr. 388-89)

105. In the early 2010s, Farelogix pioneered the development of NDC. (Carter Tr. 255)

106. The GDSs tried to undermine the development of the NDC standard. (Davidson Tr. 401; PX096 at -109) Between 2011 and 2013, Sabre and the other GDSs lobbied against IATA's petition to approve the NDC standard. (Davidson Tr. 401; PX096 at -109; PX104 at -718 (Sabre's critiques of NDC were "consistent with a desire to ensure that the status quo stays in place"))

107. As recently as January 2018, Farelogix complained to the European Commission that the "GDSs consistently seek to block new non-GDS technology solutions that deliver what consumers need." (PX096 at -107; see also Davidson Tr. 397-98) Farelogix further told the Commission that the GDSs had engaged in "A Decade of Resistance and Changing Tactics" designed to "[u]ndermine and delay" the industry's effort to move toward NDC. (PX096 at -109; see also Davidson Tr. 401-02)

108. Also in 2018, Farelogix told a prospective buyer "the slow adoption" of Farelogix's order delivery products, like FLX OC, "was solely and unarguably due to the blocking and pressure GDSs put on Farelogix, airlines and travel agencies to not adopt NDC." (PX072 at -243; see also Davidson Tr. 412)

J. Travel Agencies Prefer GDSs To Direct Connects And Non-GDS Aggregators

109. Building individual direct connections to numerous airlines requires travel agencies to spend millions of dollars in upfront capital. (Ekert Tr. 1196-98; Kunz-Cho Tr. 1311-12) Ms. Stratford testified that the cost to BCD Travel of building a new agent point of sale system would be greater than [Redacted] (Stratford Tr. 1257)

110. Maintaining direct connects raises operating costs "exponentially" for travel agencies, because servicing bookings made with a direct connect is much costlier than servicing bookings made using the GDSs. (Ekert Tr. 1196)

111. Two of the largest TMCs in the world, CWT and BCD Travel, have estimated that if they were to move transactions away from GDSs and to direct connect, costs per booking would rise approximately $15. (Ekert Tr. 1196-97; Stratford Tr. 1256) Even when a booking is made outside of the GDS, the travel agency must then bring it back into the GDS to store the booking information. (Stratford Tr. 1255) This type of booking is "static" (Ekert Tr. 1191; Stratford Tr. 1255-56); so if, after a booking is completed, the customer needs to make a change, the travel agency is unable to make that change itself but must instead reach out directly to the airline (Ekert Tr. 1191; Stratford Tr. 1256). The increased costs of static bookings are ultimately passed on to the customer in the form of additional fees. (Stratford Tr. 1264-65)

112. Direct connects negatively impact travel agency operations by slowing response times, in part due to lengthier search processes. (Stratford Tr. 1255 ("[W]e prefer to use the GDS. Just from an efficiency standpoint, we don't want to go to multiple places to assess content."); Kunz-Cho Tr. 1298-99 ("[A] request to a large number of providers naturally has a tendency to cause delays and latencies.")) Removing content, and therefore bookings, from the GDS channel requires agents to search multiple platforms for travel content, thereby reducing efficiency and resulting in a "degraded user experience" compared to the GDS. (Ekert Tr. 1191-92; see also Stratford Tr. 1263; Kunz-Cho Tr. 1300) "[H]aving access to a wider array of airfare content and ancillaries through a single pipe is far more efficient." (Kunz-Cho Tr. 1288)

113. [Redacted]

114. Use of a non-GDS aggregator presents the same added costs and inefficiencies that make direct connects unattractive to travel agencies. As Mr. Ekert of CWT testified, aggregators are "highly inefficient for us, and the user experience and quality of what we do is degraded when we ... access inventory through Travelfusion as compared to the GDSs." (Ekert Tr. 1190) Rather than the sub-second response times of the GDS, aggregators typically have response times of 15 to 20 seconds. (Id. ) This time delay for every transaction, when multiplied by tens of thousands of searches for travel content, would result in an "unreal" increase in the labor costs of travel agencies. (Ekert Tr. 1191-92)

115. In addition, non-GDS aggregators do not provide the services – e.g., mid- and back-office support and facilitating duty of care – that the GDSs provide, and present similar added expenses as direct connect transactions. (Ekert Tr. 1198-99 ("Whether that is through Travelfusion or a, quote, ‘non-GDS’ direct connect, it is basically indifferent to us."); Stratford Tr. 1223-24 ("I think the challenge of even using a third-party aggregator today ... is that, again, you're still searching in multiple places. So I still have to go to the GDS to search."))

K. Sabre And Farelogix View Each Other As Competitors

116. Notwithstanding Defendants' repeated denials at trial (see, e.g. , Kruijssen Tr. 624-25; Menke Tr. 737), a preponderance of the evidence shows that Sabre and Farelogix do view each other as competitors, although only in a limited fashion.

117. Sabre considers Farelogix a competitor in developing NDC technology for direct connects. (PX011 at 6 (Sabre document stating, "NDC Connect is similar in concept to our AS Direct Connect, though priced lower"); PX246 at -989 (listing Farelogix as among "the competition" for NDC); PX197 at -938 (describing Farelogix as "non-GDS competitor[ ]" in next-generation offer and order management))

118. The record reflects competition between Sabre's and Farelogix's direct connect solutions for major airlines.

119. For example, between 2017 and 2019, [Redacted]

120. Sabre and Farelogix also competed to provide an NDC direct connect platform to [Redacted] (PX316 at -534) Sabre viewed Farelogix as its "main competitor" for the [Redacted] opportunity. (PX316 at -534; Gilchrist Tr. 761)

121. Some of the services Sabre and Farelogix offer airlines overlap with each other. (Garner Tr. 94, 118; Murphy 1495-96) Sabre and Farelogix each allow airlines to send their offers to travel agencies, process orders or bookings, and service those orders. (Lobl Tr. 1150; Adair Tr. 1721-22)

122. Farelogix identified Sabre as a "key competitor" in order delivery and offer management. (PX072 at -219)

123. Some airlines view Sabre's GDS and Farelogix's Open Connect as partial substitutes.

124. AA has described Farelogix's direct connect technology as "providing a low cost substitute for GDSs." (PX453 at -968, -972; see also Judicial Notice Order ¶ 5 (AA describing GDSs and Farelogix as "Competitive Booking Sources"); PX452 (describing Farelogix as "the GDSs' leading competitor")

125. Similarly, a United executive testified that Farelogix offers United its only alternative way to reach U.S. travel agencies, other than going through a GDS. (Radcliffe Tr. 174-76)

126. A Delta executive testified that Farelogix provides airlines with an alternative to the GDS for distributing content to travel agencies. (Lobl Tr. 1150-53)

127. However, there are also great differences between Sabre's GDS – which provides services to travel agencies as well as airlines – and FLX OC, which is only an input for airlines. (Radcliffe Tr. 202-03; Carter Tr. 267, 273)

128. Unlike the Sabre GDS, FLX OC does not create offers, it only delivers them. (Davidson Tr. 385-86; Vilches Tr. 802-03)

129. In part because Farelogix provides fewer services to airlines, Farelogix's prices are significantly lower than Sabre's. (Radcliffe Tr. 175; Garner Tr. 107) The cost to airlines for FLX OC is "roughly a tenth" of Sabre's GDS fee. (Murphy Tr. 1453-54) Dr. Murphy explained that this disparity is attributable to "comparing the price of an input to the price of the final product." (Murphy Tr. 1453)

130. FLX OC is sold on a per-ticket basis and requires airlines to pay a subscription fee. (Kruijssen Tr. 603; Nevo Tr. 993) By contrast, Sabre's GDS is sold on a per-segment basis. (Garner Tr. 107-08; Nevo Tr. 990)

L. Farelogix Has Transformed Its Business Model Several Times

131. Farelogix's business strategy has changed dramatically multiple times over the last 15 years. (Carter Tr. 272-73; Davidson Tr. 435-36)

132. Prior to 2009, Farelogix tried – and failed – to be a GDS (Davidson Tr. 442-43), and tried – and failed – to sell a product to travel agencies that aggregated content from multiple GDSs (Carter Tr. 272-73). Farelogix abandoned this business model by 2009. (Carter Tr. 272-73)

133. In 2009, Farelogix repositioned itself as a provider of technology solely to airlines. (Carter Tr. 273) The company focused on a product called Direct Connect, which was an XML API that airlines could use to create direct connections with travel agencies. (Carter Tr. 273) This period is referred to by Farelogix's head of marketing as "ancient history" when she trains new Farelogix employees. (Carter Tr. 274)

134. Farelogix's Direct Connect failed because it did not attract a sufficient volume of transactions for Farelogix's airline customers, as travel agencies rejected its use. (Carter Tr. 274)

135. By approximately 2011, responding to the airlines' "ancillary awake[ning]," Farelogix began aggressively marketing FLX M, a software product that had been under development since 2008. (Carter Tr. 276-77; Reiz 1337-38)

136. Farelogix also developed three other offer engines to assist airlines in creating richer offers: FLX Shop & Price, FLX Schedule Builder, and FLX Availability Calculator. (PX025 at 5; Carter Tr. 239)

137. Farelogix's marketing message in 2020 "very, very rarely" addresses the use of NDC as a basis for costs savings because now "the message that resonates more with our airlines ... is all about revenue and how much money they can make selling ancillaries through their NDC pipe." (Carter Tr. 265)

M. Farelogix Is A Successful Competitor, Despite Facing Challenges, And Is Likely To Create Great Value For Sabre

138. Sabre's and Farelogix's portrayal of Farelogix at trial as a non-unique company floundering in the NDC API marketplace was unpersuasive.

139. Defendants emphasize there are many other providers of NDC API's, and this is true.

140. But no other provider of NDC API enjoys Farelogix's unique combination of a history of innovation, experience servicing the largest global carriers, strong reputation, presence the United States, and seeming financial stability. [Redacted] Wiggins Tr. 1645) For over 15 years, Farelogix has built expertise working with large airline customers. (Davidson Tr. 409, 478; PX072 at -223; PX033 at -336-41 (outlining NDC case studies from Farelogix's work with American, United, Qantas, and Lufthansa))

141. Farelogix told a prospective buyer in a June 2018 presentation that it is the "NDC market leader," the "only company" that does NDC order delivery "at scale," and overall "Holds a Unique and Deeply Rooted Position in a Market with High Entry Barriers." (PX072 at -223, -226; see also PX025 at -953 (in 2016, Farelogix was "the only provider in the marketplace delivering NDC offer and order management with production-proven, PSS-agnostic connectivity, comprehensive functionality, orchestration and support – all fully under the airline's control"); Davidson Tr. 408-11)

142. Farelogix has augmented its NDC APIs with proprietary schemas that fill the gaps in the basic NDC schema. (Davidson Tr. 411) Farelogix also offers dedicated teams for most airline customers and has a group that assists airlines in connecting to travel agencies or third-party aggregators. (Carter Tr. 247, 253; Davidson Tr. 370; PX025 at -951; PX037 at -938)

143. No other third-party NDC API provider offers the additional capabilities offered by Open Connect. (Adair Tr. 1721; PX072 at -226 (no other NDC provider supports travel agency interface like SPRK); see also Garner Tr. 152-53)

144. Large U.S. airlines would not rely on IATA certifications to assess a potential NDC API provider's suitability (see Garner Tr. 121-22) and do not view IATA certifications as a reliable measure of IT providers' capabilities (see Radcliffe Tr. 230).

145. Datalex, an Irish travel software company, is the only non-GDS order management provider other than Farelogix that currently serves a U.S. airline on the NDC leaderboard (Jetblue). (Nevo Tr. 1022; D.I. 182 (PTO) Ex. 1 ¶¶ 64-65) [Redacted] Davidson Tr. 476; Menke Tr. 681; see also Murphy Tr. 1556)

146. OpenJaw, which is owned by a Chinese governmental entity, is unlikely to expand in the U.S. market, as its relationship with China presents airlines with data security concerns. [Redacted] Davidson Tr. 656)

147. Other NDC providers, like JR Technologies and TP Connects, are unattractive because they lack experience supporting an airline with global operations. *[Redacted]

148. Hewlett Packard Enterprise ("HPE"), also known as DXC, is not a viable option even for United Airlines, which uses HPE as its PSS provider. [Redacted] (PX300 at -257; see also Radcliffe Tr. 213 (citing DXC's lack of support services and "lack of experience with travel agencies and airlines"))

149. The ATPCO NDC Exchange ("NDC Exchange") is an airline-owned entity that converts non-NDC APIs into standardized NDC APIs. (DX210 at -101520; Davidson Tr. 448-49; Reiz Tr. 1344-45; Gregorson Tr. 1667, 1671) Recently Southwest – a large, complex U.S. airline, which accounts for a significant portion of the volume of direct connect transactions in the United States – chose to use the NDC Exchange, despite initially engaging in discussions with Farelogix. (Davidson Tr. 448-49; Reiz Tr. 1345; Nevo. Tr. 1008) ATPCO was able to develop NDC connections to Southwest's API connection in two weeks. (Gregorson Tr. 1686)

150. But ATPCO's NDC Exchange does not provide an NDC API for airlines and lacks other key functionality. (Garner Tr. 122; Gregorson Tr. 1691 (explaining that NDC Exchange lacks ticketing capabilities and does not facilitate offer or order management)) ATPCO's chief strategy officer confirmed that NDC Exchange does not compete with any products offered by Sabre or Farelogix; nor do any of ATPCO's other services. (Gregorson Tr. 1689-1690, 1693-1694; PX141 at 1-2)

151. Airlines themselves are also self-supplying their own NDC APIs, which has been made easier by the updates to the IATA NDC schema. (Davidson Tr. 433) British Airways and Delta Air Lines (both of which are considered top 20 airlines worldwide) have self-supplied NDC APIs instead of purchasing a third-party's product such as FLX OC. (Radcliffe Tr. 217-18; Carter Tr. 271; Davidson Tr. 419-20, 446-47) Air France has also built its own NDC API. (Radcliffe Tr. 217; Carter Tr. 271)

152. Both American and [Redacted] have told Farelogix during negotiations that they could replace Farelogix by self-supplying their own NDC API or choosing an alternate vendor. (Garner Tr. 128-29; [Redacted]

153. Farelogix considers airline self-builds to be a significant competitive threat. (Carter 270-71; Davidson Tr. 446-47, 635-36) 154. Farelogix has had to lower its price for FLX OC in each of its negotiations with [Redacted]

155. But airline own-build solutions do not appear to be a feasible, cost-effective alternative for distribution, even for large U.S. full-service carriers. AA estimated that building its own NDC API would cost about $40 million and require annual ongoing maintenance and troubleshooting costs of $20-25 million. (Garner Tr. 127; see also Wiggins Tr. 1645-46 (building systems for airlines requires "a lot of money")) Smaller U.S. airlines also do not have plans or the resources to build NDC APIs in-house; they would prefer to license third-party IT solutions. [Redacted] Wiggins Tr. 1644-45; DX306 at 20-21 [Redacted] )

156. Moreover, it would take four to five years for AA or [Redacted] just to replace Farelogix's existing capabilities. (Garner Tr. 127-28, 159-60; [Redacted] see also Wiggins Tr. 1645-46 (building systems for airlines "takes many years"))

157. NDC is an open standard (Carter Tr. 271; Reiz Tr. 1326-27) and, as a result, NDC has enabled many other IT companies – including many new entrants – to compete for and win NDC API bids in the past five years (Carter Tr. 269, 270-71; Reiz Tr. 1327; Gregorson Tr. 1687)). Mr. Reiz, the only technologist who testified at trial and the individual who created the NDC standard, testified that any company who can build an NDC API could do so for a full service airline. (See Reiz Tr. 1329)

158. Defendants emphasize that in 2018 and 2019, after the introduction of schema version 17.2, Farelogix has submitted [Redacted] bids for FLX OC in response to RFPs, and won only [Redacted] of those bids. (Carter Tr. 269; Davison Tr. 639-40; Murphy Tr. 1471-72)

159. Since 2015, Farelogix has submitted a total of [Redacted] bids for its NDC API product, FLX OC. Of those [Redacted] bids, Farelogix won [Redacted] – a [Redacted] success rate. (Carter Tr. 271; Davidson Tr. 639-40) In that same period, Farelogix has lost NDC API bids to competitors such as Amadeus, Datalex, OpenJaw, and TPConnects. (Davidson Tr. 270)

160. Since 2015, Amadeus has won at least [Redacted] NDC API RFPs. (Davidson Tr. 648) Amadeus also provides NDC API services to Spirit Airlines. (Wiggins Tr. 1646)

161. Since 2018, Datalex has won [Redacted] NDC API RFP, the same as Farelogix. (Davidson Demonstrative 1; see also Davidson Tr. 467 (admitting Davidson Demonstrative 1) Datalex currently pr[Redacted]ovides NDC API to Scandinavian Airlines and JetBlue. (Nevo Tr. 1022)

162. Likewise, since 2018, OpenJaw and TPConnects have both won [Redacted] NDC API RFP, the same number as Farelogix. (See Davidson Demonstrative 1; see also Davidson Tr. 467 (admitting Davidson Demonstrative 1))

163. DXC – formerly Hewlett-Packard – won [Redacted] to provide NDC API services in 2017. (Radcliffe Tr. 213) Although DXC provides an NDC API to a smaller airline (Flybe), IATA's schema standardization makes connections built for smaller airlines no different than those built for larger airlines. (Reiz Tr. 1329) SAP, one of the largest technology providers in Europe, currently provides the NDC API for Easyjet, which is a top 20 airline worldwide. (Davidson Tr. 641)

164. But these facts, and Defendants' characterization of them, merely mask the reality that Farelogix has won more RFPs than any other NDC API provider (Davidson Tr. 648) and has actually been quite successful in recent years.

165. In April 2018, IATA announced a leaderboard of airlines committed to making at least 20 percent of their indirect channel bookings via an NDC API by December 2020. (PTO Ex. 1 ¶ 63) Farelogix provides the NDC API for nearly half of these airlines, including two of the three U.S. airlines on the list. (PTO Ex. 1 ¶¶ 64-65; PX094 at -566, -569; Davidson Tr. 373-74,651)

166. Defendants count as Farelogix "losses" RFPs that remain open or may have been withdrawn. (Davidson Tr. 653-55) Defendants exclude [Redacted] whose contract Farelogix just renewed, and its active opportunity with [Redacted] (Davidson Tr. 649, 655-56) Defendants also wrongly equate every RFP with every other, without accounting for the size and importance of any particular airline's business – despite the uncontestable fact that Farelogix is working for two of the largest U.S. carriers: AA and United. (Davidson Tr. 649; Davidson Demonstrative 1) Nor does Defendants' argument attribute any weight to the fact that Farelogix's ability to win recent RFPs may have been impacted by the prospect that, in the near future, it may be owned by a GDS. (See generally Davidson Tr. 652-53) (Farelogix CEO acknowledging that some airlines are taking "wait-and-see" approach while merger is pending)

167. Today, Farelogix processes more NDC bookings than any other order management services provider. (Davidson Tr. 374; PX094 at -569) Farelogix has established a larger base of airline customers for NDC services than any other provider in the U.S. market. (Menke Tr. 731-32; PX072 at -223)

168. Farelogix is doing GDS bypass implementations for AA, United, and Lufthansa, as well as other airlines. (Davidson Tr. 375) AA and United already use Farelogix direct connects to distribute through OTAs, like Priceline and Orbitz, and TMCs, including AmTrav and TripActions. (Carter Tr. 249-50, 259, 261; Radcliffe Tr. 191-92; PX025 at -954; PX033 at -340, -342)

169. AA and Lufthansa also use Farelogix's NDC API to connect to third-party aggregators. (Garner Tr. 95, 112, 139; Carter Tr. 259-60) For example, CWT and BCD use Travelfusion to access the Lufthansa Group content distributed through Farelogix's NDC API. (Reiz Tr. 1352-54)

170. Farelogix is now handling 26 NDC GDS integration projects, as compared to just six at the start of 2018. (Carter Tr. 287-88; Davidson Tr. 375-76, 381-82, 439, 441)

N. NDC Will Likely Be Used More For GDS Passthrough Than GDS Bypass

171. To some extent, this case requires the Court to predict whether widespread adoption of NDC technology is going to lead to greater use of GDS bypass, essentially eliminating the role of Sabre's GDS, or whether it is more likely to foster the expansion of GDS passthrough, by which NDC allows for the creation, delivery, and servicing of richer airline content through GDSs, including Sabre. The Court finds that NDC will likely be used more for GDS passthrough than GDS bypass. (See generally Murphy Tr. 1461) ("I think we're going to end up with the GDS integration. It just makes too much economic sense.")

172. Farelogix has enabled airlines to establish direct connects with some of the largest OTAs, thereby bypassing the GDSs. (Garner Tr. 103-04 (about 10-11% of AA's OTA bookings are made using American's Farelogix NDC API); Radcliffe Tr. 175-76 (describing United direct connect with Priceline); PX453 at -970)

173. American estimates it has achieved annual cost savings of $35 million from shifting OTAs to direct connects. (PX453 at -970) 174. One large OTA, Fareportal, established Farelogix-enabled direct connects with [Redacted] to access certain fare and ancillary content not available on a GDS; it is also in talks with [Redacted] for a direct connect. (DX306 at 5-7)

175. However, since 2016, FLX OC ticket volumes resulting from direct connects for American have declined and for United have stagnated. (Reiz Tr. 1335-36)

176. Industry participants predict that the volume of NDC use for GDS bypass – that is, direct connects and transactions through non-GDS aggregators – is unlikely to grow significantly in the future. (See, e.g. , Ekert Tr. 1186-87; Stratford Tr. 1245-47)

177. Today, and for the foreseeable future, airlines expect a major portion of NDC bookings eventually to come from GDS passthrough. [Redacted]

178. [Redacted]

179. Farelogix expects most of its growth to come from GDS passthrough. (Davidson Tr. 375-76; PX072 at -241)

180. Farelogix CEO Davidson expects "pretty much all" future growth in FLX OC volumes to "come from GDS passthrough." (Davidson Tr. 442) Farelogix's Carter agreed that "the number one thing that [Farelogix] hear[s] from airlines now is about plugging into a GDS." (Carter Tr. 285)

181. Chris Wilding of Sabre reinforced this view, testifying that "every airline I talk with very much wants [Sabre] to implement NDC .... [E]veryone that is working with an NDC solution wants to have it integrated into the Sabre GDS because they believe that is the best way to distribute their content." (Wilding Tr. 857)

182. Rose Stratford of BCD Travel believes "broad delivery of systemwide NDC solutions will be best delivered by the GDS" because "it really is about the infrastructure that TMCs have and ... servicing the workflows. The GDSs ... know what those are and understand the complexities." (Stratford Tr. 1265-66)

183. Travel agencies have informed United that they prefer using the GDSs to direct connections. (DX246 at -084) ("Big 3 TMCs will wait for the GDSs to deploy NDC capabilities.")

184. GDS passthrough, not direct connects or transactions through non-GDS aggregators, has led to the recent increased interest in NDC. (Carter Tr. 287-88) Farelogix is currently working on 26 GDS passthrough implementations. (Davidson Tr. 381-82)

185. If a particular transaction is processed via GDS bypass, the GDS is cut out and does not collect any fee. Even when a transaction is conducted via GDS passthrough, GDS fees are likely to be reduced, since the GDS performs fewer functions than it does when airlines use non-NDC legacy technology through the GDS.

186. United's director of distribution expects to pay lower booking fees for GDS passthrough bookings because United will "take the heavy lifting" of creating and pushing out offers. (Radcliffe Tr. 190-91; see also [Redacted] Nevo Tr. 1066-67; PX300 at -260 (United noting that using Farelogix NDC for "GDS pass through replaces existing legacy technology" and may result in lower GDS segment fees))

187. As AA's Garner put it, with GDS passthrough, Sabre's role is reduced to being just "an aggregator for travel agents." (Garner Tr. 118)

188. When Sabre established its GDS passthrough connection with American, Sabre's then- president of Airline Solutions, Shirk, observed that Sabre "need[ed] to be careful to hot have this over enable Farelogix only to hurt us in all our accounts." (PX308 at -982; see also Shirk Tr. 1599)

189. Using NDC technology like FLX OC in a GDS passthrough implementation has also helped airlines persuade GDSs to switch to a "wholesale model" in the OTA market. (Garner Tr. 108; PX453 at -970 (indicating AA will not pay GDS fees for OTA bookings where it offers direct connect option)) Under the wholesale model, the, payment flow shifts: the airline, not the GDS, pays the travel agency an incentive, and the travel agency pays the GDS a technology fee for each booking made through the GDS. (Garner Tr. 108-09)

190. Under the wholesale model, airlines pay less for distribution. (Garner Tr. 109; Nevo Tr. 943-44) For example, American estimates that it has achieved cost savings of $66 million per year from shifting OTAs to the wholesale model. (PX453 at -970)

191. Also under the wholesale model, GDSs will likely earn lower fees. (See Davidson Tr. 108-09; Nevo Tr. 944)

O. Dr. Nevo's Analysis Is Unpersuasive

192. DOJ relies on the economic expert analysis of Professor Aviv Nevo of the University of Pennsylvania and its Wharton School of Business. (See Nevo. Tr. 876) Dr. Nevo is well-credentialed, including having served as the Deputy Assistant Attorney General for Economic Analysis at the Antitrust Division of the Department of Justice, and has testified on behalf of DOJ and the Federal Trade Commission in other merger review cases. (See Nevo. Tr. 875-78) In this case, however, his analysis was flawed and, ultimately, unpersuasive.

193. Dr. Nevo testified that he followed "a standard multistep approach," which he described as starting by "learning about the industry and the market realities because they were really kind of the key input into what I do later." (Nevo Tr. 880) Dr. Nevo emphasized the importance of learning about the particular market he was analyzing, stating "it is important to ground the analysis in the facts of the industry, to really understand what is going on.... This ultimately is a practical analysis that is aimed to answer a question, and for that I really have to understand the market and the reality." (Nevo. Tr. 881)

194. After he felt he understood the market realities, Dr. Nevo followed a "three-step approach:"

First, you define the relevant antitrust market ... [T]hen I go and evaluate the competitive effect. And then, finally, I look to see if there's any mitigating factors that could offset these competitive effects.

(Id. )

195. Unfortunately, Dr. Nevo did not instill confidence at even the first step of this process: gaining knowledge and familiarity with the airline industry.

196. Dr. Nevo opined that there are relevant product markets for "booking services." (Nevo Tr. 885)

197. According to Dr. Nevo, "booking services" include: (1) transmitting an airline offer to a travel agency or aggregator; (2) receiving or processing an order or booking; and (3) receiving or processing changes to the order. (Nevo Tr. 899-900)

198. Witnesses, including those with lengthy service in the airline industry ecosystem, consistently testified that "booking services" is not a term they use or have heard and, more importantly, that there is no standalone "booking services" product that either Sabre or Farelogix has ever offered for sale. (See, e.g. , Carter Tr. 268; Davidson Tr. 443; Wilding Tr. 848; Reiz Tr. 1348) 199. Dr. Nevo had to acknowledge that Sabre has not provided "booking services" in a commercial transaction in the United States; therefore, he is "separat[ing]" out "booking services" functionality from the services that Sabre actually sells through its GDS platform. (Nevo Tr. 985-87)

200. At trial, Dr. Nevo was unable to provide a clear answer as to which of Farelogix's products other than FLX OC (if any) comprise what he considers the "booking services" product. (See Nevo Tr. 959-61)

201. Dr. Nevo was unable to determine a value or price for either Sabre's or Farelogix's "booking services." (Nevo Tr. 987) When Dr. Nevo was asked the price attributable to the booking services functionality within Sabre's GDS platform, he explained that "Sabre has not offered it in the U.S. I believe there is no price." (Id. ) When asked the value of the "booking services" functionality within the Sabre GDS, Dr. Nevo testified that he "did not quantify what [the] value is" and that it "was not part of [his] analysis." (Id. ) When Dr. Nevo was asked whether he compared the value of the "booking services" functionality in the Sabre GDS to the value of the "booking services" in FLX OC, he testified: "That is not something that I've offered, no." (Id. )

202. In attempting to identify and confirm the relevant product market, Dr. Nevo applied the hypothetical monopolist test, which assumes that the hypothetical monopolist controls all the relevant products in the market for the given geography, then asks whether the hypothetical monopolist would profitably impose a small price increase (i.e., a small but significant and non-transitory increase in price, or "SSNIP") on those products. (Horizontal Merger Guidelines § 4.1.1 ("Guidelines"); Nevo Tr. 910) If so, the market is a relevant market.

203. Dr. Nevo calculated that a five percent SSNIP on OTA booking services would be $0.10. (Nevo Tr. 910) He then considered whether an airline would accept the SSNIP and adjust its fares to reflect the higher costs, or reject the SSNIP and stop using OTA booking services. (Nevo Tr. 910-12) Dr. Nevo concluded that an airline would accept a SSNIP because it would be more expensive for an airline to forgo distribution through OTAs than accept the SSNIP. (Nevo Tr. 910-14) To reject a SSNIP, the airline would need to be able to persuade travelers booking through OTAs to switch to other distribution channels, or sell additional tickets to different travelers through other channels. (Nevo Tr. 904, 906-12)

204. Dr. Nevo calculated that a SSNIP on TTA services would be $0.11. (Nevo Tr. 912-13) A SSNIP on TTA services is smaller relative to the average price of airline tickets booked through TTAs, and business travelers are relatively less price sensitive, so they are unlikely to shift in response to a small price increase. (Nevo Tr. 912-13) TTAs are a critical sales channel for airlines, who would rather pay a SSNIP than pull out of all TTAs. (Garner Tr. 101-02, 109-10; Radcliffe Tr. 172-74, 348-49)

205. The Court does not find Dr. Nevo's SSNIP analysis persuasive, for reasons explained elsewhere in this Opinion. (See, e.g. , infra FF 215-16)

P. The Relevant Product Market For OTAs Has To Include Airline.com

206. Dr. Nevo excluded from his relevant product market all the airline tickets that are sold directly by airlines to end-user travelers. (See Nevo Tr. 914-15) That is, Dr. Nevo excluded airline.com from the relevant market. The Court finds, however, that airline.com has to be included in the relevant market, at least with respect to the OTA market.

207. Airline.com accounts for approximately half of all airline tickets sold to leisure travelers in the United States. (Garner Tr. 91; Wiggins Tr. 1640; [Redacted]

208. Airlines believe they can succeed in shifting bookings from the indirect channel (which involves OTAs and TTAs) to the direct sales channel (i.e., airline.com). (Lobl Tr. 1165-66; DX287 at 1 (Delta distribution strategy documents state that "[d]riving customers to direct channels is core to our strategy"); see also Tackett Tr. 1702-03 (testifying Alaska Airlines is seeking to grow volumes in direct channel))

209. [Redacted]

210. Chris Wilding, who has negotiated as many as 100 GDS agreements with airlines on behalf of Sabre, testified that airline websites are "one of the primary competitors that we face as a GDS" and explained that Sabre continues to see "a point of [market] share shift from the GDS channel to the dot com," on average, every year. (Wilding Tr. 860, 864)

211. The competition between OTAs and airline websites has increased with the rise of metasearch sites that provide direct channel results alongside OTA results. [Redacted] Metasearch sites may direct the consumer to the OTA website or the airline's website – in the latter instance, the sale is a direct channel sale. (Tackett Tr. 1701-02) The metasearch site may even permit a consumer to complete a booking through the airline's website without leaving the metasearch site. (Tackett Tr. 1701-02)

212. Fareportal co-CEO Werner Kunz-Cho testified that there is competition between OTAs and the direct channel via metasearch sites like Google Flights. (Kunz-Cho Tr. 1300)

213. Defendants' economics expert, Dr. Kevin Murphy of the University of Chicago, opined persuasively, and consistent with the record, that airlines have recognized that much of their revenues derived through sales via OTAs can be replaced by sales through airline websites. (Murphy Tr. 1438-39, 1485) For an airline, the "closest alternative" to distribution through an OTA is distribution through its own website. (Murphy Tr. 1552)

Dr. Murphy, like Dr. Nevo, has impressive and pertinent qualifications. Dr. Murphy received his Ph.D. in economics from the University of Chicago, has published articles in around 70 journals, and has received both the John Bates Clerk medal and the MacArthur fellowship. (Murphy Tr. 1415-18)

214. Dr. Murphy opined that airline direct sales have exerted significant competitive pressure on GDS fees. (Murphy Tr. 1426) Aside from the Amadeus and Travelport GDSs, airline.com is the biggest constraint on Sabre's GDS fees. (Murphy Tr. 1433)

215. Even Dr. Nevo agreed that airline.com serves as a competitive constraint on Sabre's GDS. (Nevo Tr. 1009-10) But in conducting his SSNIP analysis, Dr. Nevo could not (and did not even attempt to) determine whether airline.com was a bigger competitive constraint on Sabre's GDS than is FLX OC. (Nevo Tr. 1006, 1026-27)

216. Dr. Nevo's SSNIP tests assume that an airline confronted with a hypothetical price increase has only two choices: pay the increase or walk away. (Murphy Tr. 1481-82) As Dr. Murphy explained, however, Dr. Nevo's assumption "ignores" that airlines also have the "ability to withhold content or not reach a deal" with a GDS and instead try to steer traffic from OTAs to airline.com. (Murphy Tr. 1481) Q. U.S. Point Of Sale Is Not The Relevant Geographic Market

217. Bookings made through travel agents located in the United States are referred to by Dr. Nevo as "U.S. point of sale." (Nevo Tr. 897, 1017; see also Wilding Tr. 836, 853) To Dr. Nevo, an OTA has a U.S. point of sale if its IP address has a U.S. address. (Nevo Tr. 897) A TTA has a U.S. point of sale if it is physically located in the U.S. (Id. )

218. Dr. Nevo opined that there is a relevant market for "booking services" with a U.S. point of sale because "[i]t's not practical for an airline to substitute away from a U.S. point of sale." (Nevo Tr. 897-98) Airlines cannot easily induce travelers to switch from booking through U.S. travel agencies to booking through travel agencies in other parts of the world, and they cannot easily replace sales to travelers in the United States with sales to travelers in other countries. (Id. )

219. Sabre's GDS business has a strategy for the U.S. market and appears to separately track its market share for U.S. point of sale and rest-of-world point of sale. (PX389 at -437) For certain airlines, Sabre's GDS charges a lower price for U.S. point-of-sale bookings as compared to rest-of-world point-of-sale bookings. (Wilding Tr. 852-53; PX389 at -437)

220. However, DOJ failed to persuade the Court that U.S. point of sale – that is, travel agencies located in the U.S. – is the relevant geographic market. The reasons Dr. Nevo gave for his opinion are unsupported by the record.

221. Dr. Nevo said that his geographic market is based on "who the customer for the product is." (Nevo Tr. 1016) But, as Dr. Nevo recognized, Farelogix's customers for FLX OC are airlines. (Id. ) Thirteen of 15 FLX OC customers are airlines based outside of the United States. (Nevo Tr. 1018) Yet the point of sale Dr. Nevo used for his relevant market is not where the airline using FLX OC is based, as one would assume if the market is based on "who the customer" is. Instead, Dr. Nevo's point of sale is where the travel agent – with whom Farelogix has no relationship, and who does not use FLX OC – is based.

222. Dr. Nevo further explained that the geographic market for technology products "depends on how they're priced," i.e., whether it is a product "that you can buy in the U.S. that has a different price than if you buy it, for example, in Israel." (Nevo Tr. 1017-18) But Dr. Nevo provided no evidence that the transaction fee for FLX OC varies by the location of the travel agency that purchases a ticket. (Nevo. Tr. 1070)

R. DOJ Did Not Prove It Is Entitled To A Presumption Of Competitive Harm

223. DOJ contends that Dr. Nevo's market analysis gives rise to a presumption that the proposed merger will lead to competitive harm. The Court disagrees.

224. Dr. Nevo measured market concentration using the Herfindahl-Hirschman Index ("HHI"). (Nevo Tr. 919-20) The HHI is a standard measure used in economic literature and is calculated by computing the share of each firm in the market, squaring the shares, and summing them. (Nevo Tr. 919-20; Guidelines § 5.3) An industry with an HHI over 2,500 is considered highly concentrated, and a merger that causes an increase in HHI of more than 200 points raises significant competitive concerns. (Nevo Tr. 915-16, 920-21; Guidelines § 5.3)

225. In calculating shares for his alleged market for booking services sold through OTAs, Dr. Nevo calculated that Farelogix had 3.9% market share in 2018, while Sabre had 48.0% market share. (Nevo Tr. 917) Using Sabre's projections for 2020, Dr. Nevo calculated that Farelogix would have 12.5% market share in 2020, while Sabre would have 43.7% market share. (Nevo Tr. 917) Using 2018 data, Dr. Nevo calculated a post-merger HHI level of 4,268, with a post-merger change in HHI of 371. (Nevo Tr. 920-21) Using Sabre's projections for 2020, he calculated a post-merger HHI of 4,465, with a change in HHI of 1,093. (Id. ) These calculations would support a presumption of competitive harm, as they all exceed the standards set out in the Guidelines.

226. In calculating shares for his alleged market for booking services sold through TTAs, Dr. Nevo calculated that Farelogix had 0.1% market share in 2018, while Sabre had 54.8% market share. (Nevo Tr. 922) Using Sabre's projections for 2020, Dr. Nevo calculated that Farelogix will increase to 6.4% market share in 2020, while Sabre would drop to 51.1% market share. (Id. ) Using 2018 data, Dr. Nevo calculated a post-merger HHI level of 3,895, with a post-merger change in HHI of 6. (Id. ) Using Sabre's projections for 2020, he calculated a post-merger HHI of 4,085, with a change in HHI of 657. (Nevo Tr. 922-23) These calculations would support a presumption of competitive harm, as all but the post-merger change based on 2018 data exceed the standards set out in the Guidelines.

Dr. Nevo's market share calculations in the TTA market included bookings made through Southwest's "self-build" API. (Nevo Tr. 921). Because Southwest does not compete to provide "booking services" for other airlines, its inclusion in these calculations is conservative, as it makes the reported market shares lower than they would otherwise be. (Nevo Tr. 921-22)
Some of specific data cited in this paragraph do not actually appear in the trial record, as neither Dr. Nevo nor Dr. Murphy expressly stated all the numbers during their testimony and their demonstratives were not admitted into evidence. The relevant numbers and calculations appear in the expert reports (which were also not admitted). The Court understands there to be no dispute between the parties as to how Dr. Nevo calculated his HHI numbers or the resulting numbers if the adjustments suggested by Dr. Murphy are made. While the parties disagree about the import of the HHI analyses, and whether Dr. Nevo's or Dr. Murphy's analysis is more pertinent, they do not disagree as to what figures result from each expert's inputs.

227. Dr. Murphy identified flaws in Dr. Nevo's HHI analyses.

228. With respect to the OTA market, Dr. Nevo excluded all sales made through airline.com. (Murphy Tr. 1480-81) If Dr. Nevo's calculations using 2018 data are corrected so that airline direct channel sales are included in his alleged market for booking services sold through OTAs, the result is a post-merger HHI level of 1115 and a change in HHI of 19. (Murphy Tr. 1485) If Dr. Nevo's calculations using Sabre's 2020 projections are corrected so that airline direct channel sales are included in his alleged market for booking services sold through OTAs, the result is a post-merger HHI level of 1127 and a change in HHI of 53. (Murphy Tr. 1485) All four of these numbers fall below the Guidelines criteria (i.e., 2500 and 200) for presumptive competitive harm.

229. With respect to the TTA market, Dr. Nevo attributed all sales from GDS passthrough to Farelogix, not to any GDS. (Nevo Tr. 919; Murphy Tr. 1483-86) However, when an NDC-enabled sale is made through a GDS in a passthrough, it is the GDS that maintains the connection as well as the commercial relationships with travel agencies and airlines; the NDC provider (such as Farelogix) does not replace the GDS but, instead, acts as an upstream, vertical complement to the GDS. (Carter Tr. 269; Murphy Tr. 1446-48, 1456-58) It is the GDS, not Farelogix, that gets paid for this passthrough transaction, as even Dr. Nevo recognizes. (Nevo Tr. 1068-69) Consequently, the sale through GDS passthrough is more properly understood as a sale by the GDS, not by Farelogix. (Murphy Tr. 1486)

230. If Dr. Nevo's calculations using 2018 data are corrected so that GDS passthrough sales are credited to the GDS, not the NDC API supplier, the result is an HHI level of 3,895 and the HHI change is just 6. If Dr. Nevo's calculations using Sabre's 2020 projections are corrected in the same manner, the result is an HHI level of 3,898 and an HHI change of 19. This is ambiguous support for a presumption of harm in the TTA market, as 3,895 and 3,898 exceed the Guidelines criteria, but 6 and 19 do not.

These specific figures are not in the trial record, but they are in Murphy's expert report. As with other data described above, the Court understands the parties to be in agreement as to these numbers but not as to their relevance and import.

S. Sabre's Story Is Not Credible

231. Sabre contends that it does not view NDC, even when used for GDS bypass, as a threat. Sabre further contends that the principal reason it wants to acquire Farelogix is for its FLX M product, not for FLX OC or to address the risk of GDS bypass and passthrough. The Court does not believe Sabre.

232. Sean Menke, Sabre's CEO, testified that GDS bypass is not a threat to Sabre. (Menke Tr. 736-41) This testimony is not credible as it does not comport with the record.

233. Sabre's 2018 10-K identified "direct connect initiatives ... bypassing the GDSs" as a risk factor for Sabre's business. (PX251 at -162; see also id. at -162-63 (also identifying direct distribution through airline.com as another risk factor))

234. The 10-K contains another section on competition, which notes that the "travel distribution market is highly competitive, and [Sabre is] subject to competition from other GDS providers, direct distribution by travel suppliers and new entrants or technologies that may challenge the GDS model " – a concept which includes at least GDS bypass – and that "[a]ny inability or failure [by Sabre] to adapt to technological developments or the evolving competitive landscape could harm [Sabre's] business operations and competitiveness." (PX251 at -166) (emphasis added)

235. PX343 at -188, which is a slide from Sabre's planning forecast for 2020, also shows that Sabre views direct connect as a forward-looking risk: it states that "risks center around ... [n]ew airline models and ability for the major carriers to shift share to direct connect." (See also Shirk Tr. at 1621-22 [Redacted]; PX156 at -953; PX497)

236. Defendants' expert, Dr. Murphy, also concluded that direct connects can be a competitive constraint on GDSs and that "GDS bypass is somewhat of a threat" to Sabre. (Murphy Tr. 1520, 1558-59)

237. Even Menke admitted (eventually) that Sabre would be impacted if even a portion of the market shifts to GDS bypass. (Menke Tr. 739, 746)

238. The Court does not find credible the Sabre witnesses' testimony that none of the uses of Farelogix technology is viewed as a threat to Sabre. It would be irresponsible for Sabre's leadership not to understand that GDS bypass, and even GDS integration and the wholesale model, are threats to Sabre's traditional revenue flow.

239. Because GDS bypass is a threat to Sabre, and because Farelogix is a major player in enabling GDS bypass, it is logical to conclude that part of Sabre's interest in acquiring Farelogix is to mitigate the risk from GDS bypass. Notwithstanding CEO Menke's unequivocal denial that there is "any connection" between GDS bypass and Sabre's "reason for buying Farelogix" (Menke Tr. 744), other evidence – including internal Sabre documents – prove that this is part of Sabre's motivation.

240. Chris Boyle, who led Sabre's evaluation of the proposed Farelogix acquisition, sent a presentation to Menke identifying as part of the "value" of acquiring Farelogix that it would "[m]itigate risk from potential GDS bypass." (PX011 at 3; PX012 at 7; see also Boyle Tr. 506-07) Menke then forwarded a version of this presentation containing the "mitigate risk" language to the chairman of Sabre's board of directors. (Menke Tr. 741)

241. Boyle testified the "mitigate the risk of GDS bypass" bullet point was inserted at the last minute into a July 2018 draft internal powerpoint deck, that it was done by two subordinates during a hectic period, that the bullet is inconsistent with the rest of the document as well as an earlier deal rationale document from February 2018, and that he deleted the bullet after he noticed it because it did not reflect his views. (Boyle Tr. 506, 516-17, 544-46, 553-57; see also DX086) The final presentation prepared for Sabre's board of directors to approve the Farelogix transaction did not include "mitigating the risk of GDS bypass" as a rationale for the deal. (DX145)

242. However, ordinary course documents indicate that, in fact, Boyle did participate in the meeting where the slide was created and suggest he likely deleted the bullet in September 2018 after speaking with Sabre's antitrust counsel. (Boyle Tr. 513-16, 519-35; PX024; PX014; PX492; PX493; PX494; PX495; PX439)

243. Menke testified that the principal reason Sabre is interested in acquiring Farelogix is to own the FLX M merchandising product. (Menke Tr. 744-45 (explaining that Sabre's primary focus was on Farelogix's "merchandising capabilities to help our online customers sell the products and services that they want"); see also Boyle Tr. 552-53 (Farelogix deal leader testifying that interest in acquiring Farelogix, an idea Sabre had abandoned in 2017, was "instigat[ed]" by interest in "their merchandising engine"))

244. FLX OC generates the majority of Farelogix's revenue. (Davidson Tr. 379) Approximately two-thirds of Farelogix's employees work on Open Connect and NDC API, while far fewer Farelogix employees (around 15) work on FLX M. (Davidson Tr. 379, 382-83)

245. Internal Sabre documents do show that Sabre is according substantial value to Farelogix's FLX M and the further value that could be created by integrating it into Sabre's technology. (DX145 at -005 (November 10, 2018 presentation made to Sabre's board of directors for approval of the Farelogix acquisition, indicating that transaction would enable airlines to "integrate[ ] PSS-agnostic merchandising engine with PSS"); see also Boyle Tr. 552-53 (explaining that Sabre viewed FLX M "as a key building block, an ability to create an offer with a flexible rules engine [that] we hadn't done ourselves"); Gilchrist Tr. 781-82 (acquiring FLX M "will give [Sabre] inroads in [to] new customers that we can't currently do business [with] today as far as merchandising is concerned" and "[f]or [Sabre's] existing PSS customers who have struggled without merchandising capabilities [offering Farelogix's merchandising solution] will give [Sabre] a much, much stronger level of capability"))

246. While the Court finds that integration of FLX-M into Sabre's offerings will promote innovation and enable Sabre to better compete against its GDS counterparts, this is only one motivation for Sabre's desire to acquire Farelogix. Other motivations are dealing with an entity Sabre views as a competitor and threat and transforming NDC technology from a risk to an opportunity for innovation and integration, including by obtaining control over FLX OC, the employees who work on it, and the revenues generated from it.

247. As the Court has already found (see supra FF Part K), Sabre and Farelogix view each other as competitors, both to supply NDC APIs and for a component of the traditional full-service GDS.

248. In a presentation created for a potential acquirer, Farelogix described Sabre and Amadeus as its primary "Order Delivery" competitors, both in their development of NDC APIs and in "traditional GDS distribution until NDC is fully adopted." (PX072 at -219)

249. Similarly, Sabre strategy documents identify Farelogix as one of Sabre's "most relevant threats" and as a competitor in next-generation distribution. (PX048 at -802; see also PX005 at -927; PX197 at -938 (describing Farelogix as a "non-GDS competitor[ ]" in next-generation order management))

250. In a February 2, 2018 deal discussion presentation, Sabre's Boyle noted that Farelogix's technology solutions would help Sabre "bring[ ] NDC-enabled distribution to scale in the marketplace." (Boyle Tr. 555) (referencing DX086 at 4) Sabre's November 10, 2018 presentation seeking approval from its board of directors detailed how combining with Farelogix would accelerate Sabre's "NDC-enabled strategy." (DX145 at 5) (describing several value propositions of possible merger with Farelogix, including that Farelogix "has key building blocks that complement our organic plans enabling at-scale NDC+," "NDC Offers integrated in the GDS," "Immediate access to NDC API," "Ability to scale NDC volume quickly," "Adds NDC platform to leading Sabre product offering") These benefits are not limited to acquisition of Farelogix's FLX M product but also will be derived from Sabre's acquisition of FLX OC.

T. Sabre's Promises About Post-Merger Pricing And Availability

251. On August 9, 2019, Sabre CEO Menke sent letters to current Sabre and Farelogix airline customers, making commitments concerning what Sabre would do following its acquisition of Farelogix. (Menke Tr. 720-22; DX225)

252. Specifically, Mr. Menke wrote that after the acquisition, Sabre will:

(a) Continue to offer and support Farelogix's NDC APIs and Open Connect capabilities to any third parties and all outlets that wish to use them to connect to Sabre, other GDSs, other distribution partners, or directly to travel agents;

(b) Continue to offer Farelogix's NDC APIs and Open Connect capabilities to airlines and other third parties for both GDS bypass and GDS pass-through;

(c) Continue to make Open Connect agnostic to any form of distribution of provider;

(d) Commit to make Farelogix's NDC APIs and Open Connect available at industry competitive rates that are no greater than they are today;

(e) Provide at least the current level of support (or more) for the capabilities;

(f) Continue to invest in the development of Farelogix's capabilities at

levels no less than current levels; and

(g) Offer to extend any existing Sabre GDS contract or Farelogix contract on the same terms for a period of at least three years past the current termination date.

(DX225)

253. A three-year extension of FLX OC contracts would provide airlines that currently use FLX OC time to find an alternative supplier before their next negotiation with Sabre or with other GDS providers. ( [Redacted] )

254. Notwithstanding that the Court does not believe Sabre's story about why it seeks to acquire Farelogix (see supra FF Part S), the Court does believe that Menke intends to abide by the commitments he has expressed to customers and the market. However, as the government correctly notes, Sabre has not memorialized its offer in any legally binding agreement, and no airline has accepted Sabre's offer. (Menke Tr. 730) Furthermore, CEOs – and even firm cultures – can change.

U. Most, But Not All, Players In The Airline Travel Ecosystem Support The Proposed Transaction

255. Most of the players in the airline travel ecosystem – including especially travel agencies and airlines – support the proposed transaction. Several of those who do not, American Airlines and United Airlines, have obvious interests in seeing the deal die.

256. Kurt Ekert, CEO of CWT, testified that the proposed acquisition "will enable Sabre to ... offer much better value to the airlines than they can today." (Ekert Tr. 1199) He added that "if Sabre at scale ... is able to make NDC a reality, well, that means we don't have to go and make fundamental changes to how we operate our business." (Ekert Tr. 1201)

257. Werner Kunz-Cho, CEO of Fareportal, stated, "Sabre with a Farelogix merger will be able to embrace a wider array of content, particularly in the low cost carrier domain which is very important to us." (Kunz-Cho Tr. 1316; see also id. 1290-91 (noting one benefit of merger will be enhanced ability to meet growing traveler demand for ancillaries))

258. BCD Travel noted in a letter to DOJ that the Sabre-Farelogix combination "will benefit the travel industry as a whole – by accelerating the adoption of the [NDC] protocol – as well as the travel agencies that use global distribution systems (GDSs) such as Sabre, and companies and travelers we serve." (DX284 at 1)

259. Rocky Wiggins, Chief Information Officer at Spirit Airlines, testified that the deal "is a great step" because "it would accelerate Sabre's capability in the NDC area." (Wiggins Tr. 1658; see also id. 1661-62 (agreeing with proposition that combining with Farelogix "is another example of Sabre['s] commitment to deliver the end to end NDC-enabled retailing, distribution and fulfillment technology required to accelerate ... profitability and growth"))

260. [Redacted]

261. As far as the record shows, the entities opposing the proposed acquisition arc two of the largest airlines in the United States, American and United Airlines.

Shortly after the acquisition was announced, a senior Delta executive told his CEO that [Redacted] see also Adair Tr. 1720-21 (Farelogix is "an airline-centric entity" that "think[s] about how airlines want to solve the problems"))

262. Both American and United have previously attempted to acquire Farelogix. Specifically, in 2015, United considered a joint venture with AA to acquire Farelogix and gain "strategic influence over product direction" at Farelogix. (PX299; see also Radcliffe Tr. at 223) Starting in 2017, United began considering another offer to acquire Farelogix. (Radcliffe Tr. 224-25) That effort ended in 2018. (Radcliffe Tr. 225-26)

263. [Redacted]

264. AA's opposition to the transaction is in tension with its offer to Sabre to publicly support the proposed merger if Sabre were to agree to grant significant concessions on commercial terms fo[Redacted] (DX202; see also Garner Tr. 155-56,158)

V. The Merger's Likely Impact On Pricing, Innovation, And Leverage

265. Based on the evidence, the Court predicts that the merger of Sabre and Farelogix will not increase prices nor deter innovation, even though it may reduce one source of airlines' leverage in negotiating with GDSs.

266. During negotiations leading to the Sabre-Farelogix transaction, some consideration was given to whether acquiring Farelogix would enable Sabre to raise prices. Sabre's and Farelogix's CEOs discussed whether the transaction could lead to a "possible uplift on GDS transaction fees." (Boyle Tr. 496-97; see also PX227 at -130 ("[T]his is something we discussed on the phone and you had a strong level of confidence surrounding the opportunity.")) Farelogix and its private equity owner, Sandler, believed Sabre would be able to increase Sabre's booking fees post-merger. (Boyle Tr. 486-87, 492-93, 495; PX006 at -279; PX009 at -037) In August 2018, Farelogix's CFO noted that, for Sabre, buying Farelogix could reduce the "price pressure in [the] market." (PX187 at -960; see also Kruijssen Tr. 615-16) On the day the acquisition was announced, a Sabre senior vice president wrote: AA's "FLX bill is going up big time." (PX140 at -639; see also Gilchrist Tr. 751)

267. Even though the Court rejects Sabre's story about not perceiving GDS bypass as a threat and about acquiring Farelogix principally for FLX M, the Court does find credible Sabre's representations to the market that it intends to hold or even lower prices for FLX OC if it succeeds in acquiring Farelogix. Doing so would be consistent with the internal financial modeling Sabre used in deciding to pursue the transaction.

268. As part of Sabre's due diligence, Chris Boyle created three sets of projections to model the value of a potential acquisition: the Farelogix "Management Case," "Sabre Base Case," and "Sabre Synergy Case" models. (PX015 at 59; see also Boyle Tr. 538-40) Sabre's Board relied on Boyle's projections to approve the deal. (Menke Tr. 717-18) The FLX OC projections in the "Sabre Synergy Case" mirror those in the "Sabre Base Case" and both models project FLX OC prices decrease over time. (Boyle Tr. 567-70) In all three cases, the projected revenue per ticket was lower than current FLX OC prices. (Boyle Tr. 569)

269. [Redacted]

270. [Redacted]

271. The record is not devoid of contrary evidence. In January 2019, after a particular airline asked for no change to its current Open Connect terms with Farelogix, Sabre quoted that airline a price nearly [Redacted] times higher than Farelogix's pricing. (Compare PX393 at -079; PX392 at -026-27; Wilding Tr. 828-30 [Redacted] segment) with Davidson Tr. 413-14; PX087 at -734; PX011 at 6 [Redacted] )

272. Still, on the whole, the Court is persuaded that the most likely impact on pricing is that prices will remain the same or be reduced following the transaction.

273. Bringing Farelogix's technology in-house will foster Sabre's integration of NDC into its GDS, allowing it to respond more effectively to market demand.

274. Since at least 2017, Sabre's GDS business has been "under urgent pressure from key airline carriers" – including [Redacted] – "to deliver NDC capabilities in the near-term." (PX241 at -318; see also Menke Tr. 685)

275. In 2017, AA began paying a $2 per booking incentive to travel agencies for any bookings made using NDC technology. (PX072 at -300) Farelogix's CEO, Davidson, told his board that AA's NDC incentive program would put pressure on GDSs to speed up their GDS passthrough implementation efforts. (Davidson Tr. 374-75; PX097 at -235) Likewise, Sabre viewed [Redacted] as a "commercial action to force GDSs to incorporate NDC connectivity." (PX241 at -318)

276. Faced with this pressure, in August 2017, Sabre laid out a strategy to catch up to and surpass Farelogix's next-generation offer and order management capabilities by 2020. (Menke Tr. 682-83; PX197 at -937 to -938; PX002 at -914 (Sabre recognizing an "opportunity ... to leapfrog" Farelogix's NDC API)) Sabre recognized that if it did not invest in next-generation distribution technology, there would be a "tipping point" at which large airlines' efforts to bypass the GDSs could lead to a steep fall in Sabre's GDS booking volume. (Boyle Tr. 504-06; PX005 at -046, -048-49 (listing FLX OC customers Lufthansa, AA, Qantas, and United as carriers with "GDS bypass motivation"))

277. Acquiring Farelogix, and integrating its FLX OC NDC API into its GDS, will allow Sabre to use new technology to deliver services airlines and agencies are demanding in a more efficient manner. (See, e.g. , Kunz-Cho Tr. 1307) (Fareportal CEO testifying that "this merger will likely help innovation and allow Sabre to provide better content in ways that airlines wish to see that content displayed")

278. Additionally, Farelogix's Tim Reiz, the inventor of the NDC API and the only technologist to testify at trial, has considered developing a new type of PSS, but could not do so due to Farelogix's limited resources. (Reiz Tr. 1371-73) Reiz is excited by the prospect of joining Sabre and the opportunity, potentially, to develop a next generation PSS and artificial intelligence technologies. (Reiz Tr. 1346)

279. At least some individuals within Sabre believe that acquiring Farelogix will help Sabre gain leverage over airlines in the next round of negotiations. One Sabre senior vice president predicted that AA's Garner would "hate" the Farelogix deal because "it entrenches us more," vividly describing an independent Farelogix as American's "Trojan horse to f--- us." (Gilchrist Tr. 750-51; PX140 at -639)

280. In 2017, United's director of distribution advised United's leadership that partnering with Farelogix for NDC distribution would "improve[ ] United's position" in upcoming contract negotiations with Sabre, Amadeus, and Travelport (PX300 at -253; see also Radcliffe Tr. 187) He added that, even in the most recent GDS negotiations, "[t]he existence of the direct connect alternative powered by Farelogix gave me the ability to command a lower price from the GDSs" and otherwise obtain "better terms." (Radcliffe Tr. 189; see also id. at 181-82 (United has "certain freedoms in the [GDS] agreement that are made possible because of the existence of a bypass opportunity"))

281. United's director of distribution wrote to colleagues that "[i]f a GDS owns Farelogix, they may ... remove a major threat that is out there in the industry that helps apply pressure to GDSs when we negotiate. Without that alternative in the market, we lose leverage." (PX299 at -770; see also Radcliffe Tr. 184-86) He described the acquisition as the "stuff of nightmares" and "the worst case scenario coming true." (PX301 at -261; see also Radcliffe Tr. 231-34)

282. Similarly, Delta's managing director of distribution strategy testified that having Farelogix as a GDS alternative improved Delta's bargaining position with the GDSs. (Lobl Tr. 1152-54)

283. It follows that the loss of an independent Farelogix will somewhat reduce airlines' leverage in the next round of negotiations with GDSs.

LEGAL STANDARDS

Section 7 of the Clayton Act prohibits any merger "in any line of commerce or in any activity affecting commerce in any section of the country" whose "effect[s] ... may be substantially to lessen competition, or to tend to create a monopoly." 15 U.S.C. § 18. To prevail on a claim under Section 7, the government must show a "reasonable probability that the merger will substantially lessen competition." Brown Shoe Co. v. United States , 370 U.S. 294, 325, 82 S.Ct. 1502, 8 L.Ed.2d 510 (1962). While the government need not prove anticompetitive effects "with ‘certainty,’ " FTC v. H.J. Heinz Co. , 246 F.3d 708, 719 (D.C. Cir. 2001), it is "not enough" to show "[t]he mere possibility of the prohibited restraint," FTC v. Consol. Foods Corp. , 380 U.S. 592, 598, 85 S.Ct. 1220, 14 L.Ed.2d 95 (1965) (internal quotation marks omitted).

Courts considering a Section 7 merger challenge follow a burden-shifting framework. First, the Court determines whether the government has established a prima facie case that the proposed merger is anticompetitive by (1) identifying the proper relevant market and (2) showing that the effects of the merger are likely to be anticompetitive. See FTC v. Penn State Hershey Med. Ctr. , 838 F.3d 327, 337-38 (3d Cir. 2016). If the government succeeds at this first step, the Court next determines whether the defendants have rebutted the government's prima facie case. See id. at 337. Finally, if defendants do successfully rebut the government's prima facie case, "the burden of production shifts back to the [g]overnment and merges with the ultimate burden of persuasion, which is incumbent on the [g]overnment at all times." Id. Ultimately, the government's burden is to prove, by a preponderance of the evidence, a reasonable probability that the merger it seeks to enjoin will substantially lessen competition. See United States v. Anthem, Inc. , 236 F. Supp. 3d 171, 192 (D.D.C. 2017), aff'd , 855 F.3d 345 (D.C. Cir. 2017).

"The relevant market is defined as the area of effective competition." Ohio v. American Express Co. , ––– U.S. ––––, 138 S. Ct. 2274, 2285, 201 L.Ed.2d 678 (2018) (internal quotation marks omitted) (" Amex "). It consists of two components: a "product market" and a "geographic market." Brown Shoe , 370 U.S. at 324, 82 S.Ct. 1502. A properly-identified relevant market "must correspond to the commercial realities of the industry." Amex , 138 S. Ct. at 2285 (internal quotation marks omitted).

"[D]etermination of the relevant market is a necessary predicate to a finding of a violation of the Clayton Act." Brown Shoe , 370 U.S. at 324, 82 S.Ct. 1502. Because "[p]laintiffs have the burden of defining the relevant market," the failure to properly define either a product or geographic market is fatal to plaintiffs' case. Queen City Pizza, Inc. v. Domino's Pizza, Inc. , 124 F.3d 430, 436-42 (3d Cir. 1997) ; see also FTC v. Tenet Health Care Corp. , 186 F.3d 1045, 1053 (8th Cir. 1999).

DISCUSSION

I. DOJ Failed To Establish A Prima Facie Case

Based on the Court's findings of fact, set out above, and as further explained below, the Court concludes that DOJ failed to establish a prima facie case that the merger of Sabre and Farelogix will violate Section 7 of the Clayton Act. Specifically, DOJ has not identified a proper relevant market. As a matter of antitrust law, Sabre, a two-sided transaction platform, only competes with other two-sided platforms, but Farelogix only operates on the airline side of Sabre's platform. Even if that were not the law, DOJ's market analysis fails because it does not relate to the relevant product market or the relevant geographic market. In the alternative, even if the Court were to assume that DOJ has identified a relevant product market, and were to assume that the record at least supports a prima facie case that the effects of the merger are likely to be anticompetitive, the Court further concludes that Defendants have rebutted the government's prima facie case. After making credibility determinations, drawing the reasonable inferences the Court deems most warranted, and weighing the evidence, the Court concludes that the government has failed to prove by a preponderance of the evidence that the Sabre-Farelogix transaction is reasonably probable to substantially lessen competition.

A. As A Matter of Antitrust Law, Sabre And Farelogix Do Not Compete In A Relevant Market

The first dispositive flaw in the government's case is that, as a matter of antitrust law, Sabre, which is a two-sided platform facilitating transactions between airlines and travel agencies, does not compete with Farelogix, which indisputably only interacts with airlines and is not a two-sided platform. Due to the combination of the Supreme Court's 2018 Amex decision, holding that "[o]nly other two-sided platforms can compete with a two-sided platform for transactions," 138 S. Ct. at 2287, and the Second Circuit's 2019 finding that the Sabre GDS is a two-sided transaction platform, Sabre and Farelogix do not compete in a relevant market.

In Amex , the United States and several states alleged that American Express violated Section 1 of the Sherman Act by requiring merchants to agree to "antisteering" contract provisions that prohibited merchants from discouraging the use of American Express cards. Id. at 2283. The Court first evaluated the relevant market and concluded that "the credit-card market is one market, not two." Id. at 2283, 2285-87. It noted that credit-card networks like American Express were what economists call "two-sided platforms:" entities that "offer[ ] different products or services to two different groups who both depend on the platform to intermediate between them." Id. at 2280, 2286. According to the Court, these platforms "cannot raise prices on one side without risking a feedback loop of declining demand;" thus, the Court held, "courts must include both sides of the platform" when defining a relevant market for purposes of antitrust analysis. Id. at 2285.

The Court emphasized that "it is not always necessary to consider both sides of a two-sided platform," such as when "the impacts of indirect network effects and relative pricing in that market are minor." Id. at 2286. However, the Court continued, it is always necessary to consider both sides of "two-sided transaction platforms" that "facilitate a single, simultaneous transaction between participants" and "cannot make a sale unless both sides of the platform simultaneously agree to use their services." Id. at 2286-87. These platforms, according to the Court, "suppl[y] only one product - transactions." Id. at 2286. Because these transactions necessarily involve both sides of the market, "[o]nly other two-sided platforms can compete with a two-sided platform for transactions." Id. at 2287. In other words, for two-sided transaction platforms, "competition cannot be accurately assessed by looking at only one side ... in isolation." Id.

DOJ argues that Amex does not govern here, but its arguments are unavailing. At trial DOJ suggested that Amex may be limited just to the credit card industry, but DOJ points to nothing in Amex to support such a conclusion. (See Tr. 1834-35) DOJ also observes that the record in this case, including from Defendants' expert Dr. Murphy, shows that "GDSs can face competition from one-sided competitors." (D.I. 233 at 13) (citing Murphy Tr. 1521-22) While the government has accurately characterized the record, the facts presented in the instant case cannot change the binding precedential law issued by the Supreme Court.

DOJ maintains that Sabre and Farelogix are competitors because "a firm can compete in more than one market [a]nd the booking services portion of the Sabre GDS competes in a one-sided booking service market." (Tr. 1831) However, Amex establishes that two-sided transaction platforms such as the Sabre GDS supply "only one product:" the transactions that link both sides of the market. Amex , 138 S. Ct. at 2286. Because Farelogix indisputably does not supply this product – it offers services to airlines, not to travel agencies – it does not compete with Sabre.

While relatively few courts have had an opportunity to apply Amex , the Second Circuit has, in a different case that happened to be all about Sabre's GDS. In US Airways v. Sabre Holdings Corp. , 938 F.3d 43, 48-49 (2d Cir. 2019) (" US Airways "), US Airways alleged that Sabre's restrictive contract provisions and "monopoliz[ation of] the distribution of system services to Sabre subscribers" violated Sections 1 and 2 of the Sherman Act, respectively. A jury found that the relevant market was one-sided – and also that, even if the market were two-sided, US Airways had proven competitive harm – and returned a verdict for US Airways. See id. at 53.

On appeal, the Second Circuit vacated the jury verdict, determining it was based on the jury's "erroneous" conclusion that "the relevant market was one-sided." Id. at 57-58. Applying Amex , the Second Circuit held that the Sabre GDS is a two-sided transaction platform because it "offer[s] different services to different groups of customers – to airlines, access to travel agents; to travel agents, flight and pricing information – and [it] connect[s] travel agents to airlines in simultaneous transactions." Id. at 58. Thus, according to the Second Circuit in US Airways , "the relevant market for such a platform must as a matter of law include both sides." Id.

DOJ is correct that the Second Circuit's US Airways decision was directed to a different legal question, was based on review of a different factual record than the one created here, and is, of course, not binding precedent in the Third Circuit. However, the Court agrees with Defendants that, for purposes of understanding Amex , there is no meaningful distinction between Section 1 Sherman Act claims and a Section 7 Clayton Act merger challenge. (See D.I. 235 at 11-12) This Court finds the Second Circuit's application of Amex to the very same Sabre GDS platform this Court is required to consider highly persuasive authority and chooses to follow it. Moreover, DOJ agrees that the Sabre GDS operates in a two-sided market. (Tr. 1831, 1836; see also D.I. 233 at 15 (acknowledging that at least Sabre's aggregation function "is properly classified as a two-sided service"))

As best as the Court can tell, the Third Circuit has not yet had occasion to consider Amex .

See also United States v. Grinnell Corp. , 384 U.S. 563, 86 S. Ct. 1698, 1705, 16 L.Ed.2d 778 (1966) ("We see no reason to differentiate between ‘line’ of commerce in the context of the Clayton Act and ‘part’ of commerce for purposes of the Sherman Act."); Hornsby Oil Co. v. Champion Spark Plug Co. , 714 F.2d 1384, 1393 n.9 (5th Cir. 1983) ; Kellam Energy, Inc. v. Duncan , 616 F. Supp. 215, 218 n.3 (D. Del. 1985).

From these conclusions it necessarily follows that DOJ cannot prevail on its claim as a matter of law. "Only other two-sided platforms can compete with a two-sided platform for transactions," Amex , 138 S. Ct. at 2287, and Farelogix is not a two-sided platform, as even DOJ concedes (see Tr. 1836). Even if "it is not always necessary to consider both sides of a two-sided transaction platform," it is necessary to do so where, as here, both sides of Sabre's GDS platform "facilitate a single, simultaneous transaction between participants." Id. at 2286-87. Airlines on one side of Sabre's GDS cannot make a sale to travel agencies on the other side of the GDS "unless both sides of the platform simultaneously agree to use" Sabre's GDS services. Id. at 2286. This is a requirement in order for Sabre's GDS to provide its product: "transactions." Id.

Even if all of this were not the case, and if it were possible as a matter of law for Farelogix on one side of the two-sided GDS platform to be found to compete in a relevant market, the evidence presented by DOJ does not adequately account for the fact that Farelogix is a one-sided player and Sabre is a two-sided transaction platform. That is, DOJ has failed to produce evidence that the anticompetitive impact of the merger on the airline side of the GDS platform would be so substantial that it would sufficiently reverberate throughout the Sabre GDS to such an extent as to make the two-sided GDS platform market, overall, less competitive. DOJ did not even try to meet this burden. Instead, DOJ (and its expert, Dr. Nevo) "look[ed] at only one side" of the Sabre GDS "in isolation," which "cannot" result in an "accurate[ ] assess[ment]" of competitive effects. Amex , 138 S. Ct. at 2287. Thus, again, given Amex and US Airways , DOJ cannot, as a matter of law, make out a prime facie case of a Section 7 Clayton Act violation.

Finally, the government warns that a conclusion that, as a matter of law, Sabre and Farelogix do not compete for purposes of antitrust law would give "any GDS ... carte blanche to buy any one-sided competitor, free from scrutiny under Section 7." (D.I. 233 at 13) The Court is not persuaded (although, even if it were, the Court would still be compelled to follow the Supreme Court's decision in Amex ). Rather, Amex provides that if the government seeks to stop a GDS from buying a "one-sided competitor," it must show that this purchase will harm competition on both sides of the two-sided market – i.e., the market for travel services to airlines and the market for travel services to travel agencies. Here, however, the government only attempted to demonstrate harm to the airlines side of the two-sided market. It has, thus, failed to meet its burden. Notwithstanding this dispositive ruling, the Court will proceed and, in the remainder of this Opinion, assess the government's attempt to meet its burden on the assumption, in the alternative, that Farelogix and Sabre could be found to compete in a relevant market, even though they are not both two-sided platforms. As explained below, the Court finds that DOJ has failed even when given the benefit of that assumption.

The Court does so for multiple reasons, including the present lack of guidance from the Third Circuit on applying Amex , the possibility of appellate review in this case, the expedited nature of this litigation, and the enormous resources the parties and the Court have devoted to this case. Additionally, as is clear from the findings of fact, the Court has found as a matter of real-world economic reality that Sabre and Farelogix do compete to a certain extent, so resting a decision in this case entirely on a determination of law that Sabre and Farelogix cannot compete in a relevant market is not a comfortable result. Further, the US Airways decision was issued after the government filed this case, so it also does not seem right to summarily terminate DOJ's case based on a non-binding subsequent decision from an appellate court in a case in which the government was not even involved.

B. As A Factual Matter, DOJ Failed To Identify A Relevant Product Market

"The outer boundaries of a product market are determined by the reasonable interchangeability of use or the cross-elasticity of demand between the product itself and substitutes for it." Brown Shoe , 370 U.S. at 325, 82 S.Ct. 1502. Thus, products comprising a relevant market "need not be identical, only reasonable substitutes." United States v. Energy Sols., Inc. , 265 F. Supp. 3d 415, 436 (D. Del. 2017). To determine the reasonable interchangeability of products, courts consider "price, use, and qualities." Queen City Pizza , 124 F.3d at 437.

DOJ argues that the relevant product market consists of "booking services" sold through online travel agencies ("OTA") and brick-and-mortar traditional travel agencies ("TTA"). (D.I. 233 at 9) In the government's view, "[t]hese markets encompass the nexus of competition that exists between Sabre and Farelogix and are the appropriate focus for analyzing the competitive effects of this merger." (Id. ) According to DOJ's economics expert, Dr. Nevo, "booking services" are a "component of what we saw in the GDS bundle ... of getting the offer from, after it was created using the airline data and getting it either through the travel agency or through the aggregator that then gets it to the travel agency." (Nevo Tr. 899) The Court concludes that DOJ failed to identify a relevant product market.

DOJ's product market is flawed for several reasons. Foremost among them is that DOJ selectively (without persuasive explanation) dissects Sabre's overall GDS services into what DOJ refers to as "booking services." In doing so, DOJ improperly excludes the services Sabre's GDS provides to travel agencies, inconsistent with Amex , as already discussed above. But even as a factual matter, DOJ's proposed "booking services" product market does not accurately correspond to what actually is transacted in a market relevant to the proposed transaction.

As Defendants aptly put it: "DOJ contends that a single technical functionality of Sabre's unitary GDS transaction platform, which is sold in a two-sided market, can be extracted from the GDS and placed in competition against a slice of FLX OC, which is sold in a different, one-sided market." (D.I. 235 at 12) "No party can expect to gerrymander its way to an antitrust victory without due regard for market realities." It's My Party, Inc. v. Live Nation, Inc. , 811 F.3d 676, 683 (4th Cir. 2016).

DOJ urges the Court not to "[f]ocus[ ] on the breadth of Sabre's entire bundle of GDS products" but instead to evaluate the one "submarket" of "booking services," which DOJ says is a "one-sided product." (D.I. 233 at 13-15) However, DOJ has not proven that the Sabre GDS's "booking services" is a "well-defined submarket [that] constitute[s] a relevant product market." Tunis Bros. Co. v. Ford Motor Co., Inc. , 952 F.2d 715, 723 (3d Cir. 1991). DOJ failed to show that the Sabre GDS "booking services" is a distinct product that generates separate demand. Thus, DOJ's proposed product market is at odds with the "commercial realities of the industry." Amex , 138 S. Ct. at 2285 (internal quotation marks omitted).

Both sides introduced evidence that the Sabre GDS platform performs several different functions, including aggregating and normalizing content, automating rules set by corporate travel policies, and creating offers for travel agencies. (See, e.g. , FF 34-36) The Court agrees with Defendants that the correct interpretation of the record is that the subset of Sabre services DOJ labels "booking services" are not a separate product but, instead, have "no independent economic significance." (D.I. 235 at 13)

Dr. Nevo's testimony failed to persuade the Court that there is separate demand for Sabre's "booking services." Dr. Nevo acknowledged that Sabre's "booking services" are not sold as a separate product in the market, but are only "sold as part of a bundle," so he himself "separate[d] them" for his analysis. (Nevo Tr. 985-86) Besides seeming to be an arbitrary exercise, Dr. Nevo's approach is unpersuasive also because he was unable to identify basic features of his purported product. For instance, Dr. Nevo could not identify a price of the Sabre GDS's "booking services" functionality because, as he testified, "Sabre has not offered it in the U.S. I believe there is no price." (Nevo Tr. 987) Dr. Nevo further testified that he "did not quantify what [the] value is" of the Sabre GDS "booking services," as this "was not part of my analysis." (Id. ) Fundamentally, Dr. Nevo gave the Court no solid basis to conclude that "booking services" is a product customers are demanding from Sabre.

An analogy can be made to tying cases, in which a plaintiff must prove "an agreement by a party to sell one product [or service] but only on the condition that the buyer also purchases a different (or tied) product [or service]." Avaya, Inc., RP v. Telecom Labs, Inc. , 838 F. 3d 354, 394 (3d Cir. 2016). In these cases, "the answer to the question whether one or two products are involved turns not on the functional relation between them, but rather on the character of the demand for the two items." Jefferson Parish Hosp. Dist. No. 2 v. Hyde , 466 U.S. 2, 19, 104 S.Ct. 1551, 80 L.Ed.2d 2 (1984), partially abrogated on other grounds by Illinois Tool Works, Inc. v. Independent Ink, Inc. , 547 U.S. 28, 126 S.Ct. 1281, 164 L.Ed.2d 26 (2006). Thus, "to identify a distinct product market" for a service that is included in a larger "package" of services, a plaintiff must show "sufficient demand for the purchase of [that] service[ ] separate from [the package]." Id. (emphasis added); see also Avaya , 838 F.3d at 397 ; Collins v. Associated Pathologists, Ltd. , 844 F.2d 473, 477-78 (7th Cir. 1988) (holding that because "no separate demand exists for pathological services that is sufficient to create a separate market [from hospital services]," pathological and hospital services "combine in the form of one product, not two tied products").

As part of its effort to show that Sabre offers "booking services," DOJ points to Sabre's "new NDC API" that it has tried to sell to two airlines (D.I. 233 at 15), as well as Dr. Nevo's testimony that "[i]n kind of the early 2000s, there was a case where the Brazilian Airline asked Sabre to help it in creating a kind of a direct connect product" (Nevo Tr. 937; see also D.I. 233 at 15). Even this evidence, however, fails to show that the Sabre GDS platform includes a distinct "booking services" component that generates separate demand; at best, it shows that Sabre has tried to market an entirely different product that provides booking services.

Likewise, while DOJ attempts to show "distinct demand for booking services and aggregation services" by suggesting that Farelogix and Travelfusion offer these services on a standalone basis (see D.I. 241 at 3), this evidence does not demonstrate that Sabre offers these products on a standalone basis; nor does it establish that Sabre participates in these product "submarkets."

Accordingly, DOJ's proposed product market – which assumes a Sabre "booking services" product exists (or at minimum can be meaningfully and non-arbitrarily separated out from Sabre's actual GDS functionality) and competes with Farelogix's FLX OC – fails to "correspond to the commercial realities of the industry." Amex , 138 S. Ct. at 2285 (internal quotation marks omitted). DOJ failed to show that a portion of Sabre's GDS and Farelogix's FLX OC are reasonable substitutes for one another.

There are other flaws in DOJ's efforts to identify a relevant product market. One is that the OTA booking services market improperly excludes sales made by airlines in the direct channel on their websites, at kiosks, and through call centers. That is, DOJ wrongly excludes airline.com from the OTA market.

Courts can determine the "boundaries" of a product market by "examining ... practical indicia" such as "industry or public recognition of the submarket as a separate economic entity, the product's peculiar characteristics and uses, unique production facilities, distinct customers, distinct prices, sensitivity to price changes, and specialized vendors." Brown Shoe , 370 U.S. at 325, 82 S.Ct. 1502. Hence, courts evaluating whether different products are in the same product market often examine real-world market conditions, like whether customers "switch back and forth between [the] products," or if firms "compete against each other" to offer these products. Energy Sols. , 265 F. Supp. 3d at 437-38.

Substantial evidence supports the Court's conclusion that airline.com is reasonably interchangeable with the travel services the Sabre GDS offers through OTAs. Airline.com and OTA both target leisure travelers. (See, e.g. , Garner Tr. 101, 123-25; Radcliffe Tr. 172; Kunz-Cho Tr. 1282-83) End-user travelers switch back and forth between OTAs – like Expedia and Priceline – and airlines' own websites. (See, e.g. , FF 46, 211) Defendants' expert, Dr. Murphy, persuasively explained how airline.com is the "closest alternative consumers can see" to OTAs. (Murphy Tr. 1553) He added that airline.com is the "most important short-term constraint" on GDS fees, because airline.com is only "a mouse click away" from OTAs. (Murphy Tr. 1433-34) Even Dr. Nevo testified that he "believe[s] that there is some competitive pressure [to GDSs] from airline.com" (although he could not say whether airline.com exerts more or less competitive pressure on Sabre's GDS than FLX OC does). (Nevo Tr. 1009-10)

Further, airlines view OTAs as competitors with airline.com and believe they can steer at least a substantial portion of OTA sales to their own direct sales channels. (See Murphy Tr. 1438-49; Lobl Tr. 1165 (Delta Director of Distribution testifying Delta is "hoping that customers will choose" Delta's direct channels); DX287) Airlines can switch from offering at least some of their leisure traveler tickets from OTAs to airline.com.

DOJ insists that Defendants are wrong, and that airline.com is properly excluded from the OTA market, because Dr. Murphy did not use the small but significant and non-transitory increase in price ("SSNIP") test to evaluate whether airline.com should be included in the relevant product market. (D.I. 233 at 17-18) As an initial matter, the burden is on the government to show that airline.com is not part of the relevant OTA market, and it has failed to meet that burden. In any case, while the SSNIP test is a "common method" that courts can use to define the relevant market, see Penn State Hershey , 838 F.3d at 338, it is not the only one. Here, the Court is evaluating economic "practical indicia" to assess whether products are "reasonably interchangeable," which is another permissible analysis. Brown Shoe , 370 U.S. at 325, 82 S.Ct. 1502 ; see also Energy Sols. , 265 F. Supp. 3d at 437-38. Doing so, the Court concludes that airline.com and OTAs are reasonably interchangeable.

DOJ also tries to justify excluding airline.com from the OTA booking services market on the grounds that airlines cannot "replace enough of their OTA bookings with airline.com to make forgoing OTA distribution economically feasible." (D.I. 233 at 18) But the test for a relevant product market is whether the "commodities [are] reasonably interchangeable by consumers for the same purposes" – not whether it is feasible to convert all users to only one of those commodities. United States v. E.I. du Pont de Nemours & Co. , 351 U.S. 377, 395, 76 S.Ct. 994, 100 L.Ed. 1264 (1956) ; see also Tunis Bros. , 952 F.2d at 722 ; FTC v. Arch Coal, Inc. , 329 F. Supp. 2d 109, 122 (D.D.C. 2004) ("In determining interchangeability, therefore, the Court must consider the degree to which buyers treat the products as interchangeable, but need not find that all buyers will substitute one commodity for another.").

It is not as clear whether airline.com must be included in the "TTA booking services" market. TTAs primarily serve business travelers and offer specific support services that these travelers request, which are not available from airline.com. (Radcliffe Tr. 172-73, 335; Ekert Tr. 1184) There is also a lack of evidence that airlines view TTAs as an alternative to airline.com or that airline.com exerts competitive pressure on TTAs. It may be, then, that DOJ's TTA market analysis is somewhat less flawed than its OTA market analysis.

For at least these reasons, DOJ has failed to identify a relevant product market.

The Court agrees with Defendants that, to the extent DOJ is now inviting the Court to "unilaterally change the defective market allegations if necessary to save its case," it would be wrong for the Court to do so under the circ*mstances here, which include that both parties (and the Court) have already devoted enormous resources to the case the government chose to bring. (D.I. 235 at 2; see also id. at 18-20; D.I. 242 at 2 n.3; see generally D.I. 233 at 30 n.6 (DOJ: "The Court is not limited to considering the markets that the Government alleged; if the Court finds harm in any relevant market based on the evidence presented, the Court may find a Section 7 violation."))

C. DOJ Failed To Identify A Relevant Geographic Market

The relevant geographic market is comprised of the area where customers look to buy a seller's products or services. See Tunis Bros. , 952 F.2d at 726 ("[T]he geographic market is not comprised of the region in which the seller attempts to sell its product, but rather is comprised of the area where his customers would look to buy such a product."). Here, the evidence shows the customers of the purported "booking services" are airlines. (See, e.g. , FF 70-71, 221) Therefore, the relevant geographic market must be based on where airlines look to purchase these services. (See Nevo Tr. 906, 1016)

DOJ relies on Dr. Nevo's analysis that the relevant geographic market is "U.S. point of sale," that is, for TTAs those travel agencies located in the U.S. and for OTAs those travel agencies having an IP address in the U.S. (See Nevo Tr. 836, 853, 897; FF 217-18) DOJ reasons that airlines looking to sell tickets in the U.S. will only use travel agencies operating in the U.S. because U.S. travelers will not use foreign travel agencies. (D.I. 233 at 19-21; D.I. 241 at 6-7)

The Court disagrees with the government. Instead, the Court agrees with Defendants that DOJ's geographic market is a "contortion," wrongly "focus[ing] not on FLX OC's actual customers in the upstream market (i.e., the airlines), but rather those customers' customers in the downstream market (i.e., the travel agencies or travelers)." (D.I. 242 at 3) The airlines' demand for travel agency services is not a useful starting point for the geographic market analysis because airlines that use "booking services" (assuming, for the sake of argument, this is an accurate and meaningful description of a relevant product) need not – and frequently do not – sell their tickets through travel agencies. Indeed, airlines that use "booking services" such as FLX OC can (and do) distribute their tickets through several means other than travel agencies, such as their own websites or mobile apps, or through non-GDS aggregators. (See, e.g. , PX081 at -047; Carter Tr. 283-84)

Moreover, the evidence shows that airlines do not look only within the United States to purchase "booking services." As Dr. Nevo testified, the vast majority of Farelogix's customers, 13 out of 15, are airlines located outside the United States. (Nevo Tr. 1018) Farelogix also competes with foreign competitors for bids. (See, e.g. , FF 145-46, 159-63) Likewise, when Sabre marketed a standalone direct connect product – what DOJ labels a Sabre "booking services" product – Sabre tried selling it to airlines located outside the United States. (Gilchrist Tr. 761; PX316 at -534; Nevo Tr. 937; see also D.I. 233 at 15) Thus, DOJ's proposed U.S.-based geographic market is at odds with commercial realities. Additionally, there is no evidence that the transaction fee for FLX OC varies by the location of the agency that purchase a ticket. (See FF 222)

For at least these reasons, DOJ has failed to identify a relevant geographic market. See Dicar, Inc. v. Stafford Corrugated Prods., Inc. , 2010 WL 988548, at *12 (D.N.J. Mar 12, 2010) ("Where, as here, there is no indication that a consumer would be unable to purchase a product abroad, the Court will not arbitrarily limit the geographical market to the U.S."); E & E Co., Ltd. v. Kam Hing Enters., Inc. , 2008 WL 3916256 at *3 (N.D. Cal. August 25, 2008) (finding plaintiff did not adequately allege relevant market when it "fail[ed] to indicate why similar bed coverings originating from other countries would not be reasonably interchangeable by consumers for the same purposes") (internal quotation marks omitted). This failing (like others already described by the Court) is dispositive.

II. Even Assuming DOJ Made Out A Prima Facie Case, The Record Does Not Demonstrate A Reasonable Probability That The Merger Will Substantially Lessen Competition

A. DOJ Is Not Entitled To A Presumption Of Anticompetitive Harm

Because DOJ has not identified the relevant market where competition occurs – and has offered evidence of likely harm only with respect to its flawed markets – it cannot prove the second step of its prima facie case: that "the effect of the merger in [the relevant] market is likely to be anticompetitive." Penn State Hershey , 838 F.3d at 337-38. However, even if DOJ were correct that the relevant market is "booking services" at the "U.S. point of sale," DOJ has still failed to prove likely harm to competition in this market.

At trial, DOJ presented Herfindahl-Hirschman Index ("HHI") measurements to show that Sabre's acquisition of Farelogix would harm competition in the relevant market. (D.I. 233 at 21-23) Even assuming that DOJ identified the correct market, this evidence was not persuasive.

HHI is a commonly-used measurement of market concentration. See Penn State Hershey , 838 F.3d at 346. As the Third Circuit has explained:

The HHI is calculated by summing the squares of the individual firms' market shares. In determining whether the HHI demonstrates a high market concentration, we consider both the post-merger HHI number and the increase in the HHI resulting from the merger. A post-merger market with a HHI above 2,500 is classified as "highly concentrated," and a merger that increases the HHI by more than 200 points is "presumed to be likely to enhance market power." The Government can establish a prima facie case simply by showing a high market concentration based on HHI numbers.

Id. at 346-47 (internal citations omitted).

DOJ offered HHI evidence for both of its proposed markets. To calculate HHI in the "OTA booking services" and "TTA booking services" markets, Dr. Nevo relied on 2018 market share data, Sabre's projections for Farelogix's growth, and Farelogix's own projections. (Nevo Tr. 916-21) Dr. Nevo testified that in both the "OTA bookings services" and "TTA booking services" markets, the post-merger HHI greatly exceeds the threshold for a highly concentrated market (2,500) and the increase in HHI exceeds the level at which enhanced market power is presumed (200). (Nevo Tr. 920-23)

But Dr. Nevo's HHI cannot be relied on because it ignores industry realities. In calculating market shares in the "OTA booking services" market, Dr. Nevo excluded airline.com – but Dr. Nevo's own testimony, and much other evidence besides, showed that airline.com exerts competitive pressure on the Sabre GDS. (See, e.g. , Nevo Tr. 1009-10; FF 206-16) Likewise, when calculating market shares in the "TTA booking services" market, Dr. Nevo attributed sales generated by GDS passthrough to Farelogix rather than to the GDS – even though the GDS controls the sale and the commercial relationship. (See Nevo Tr. 917-18; Murphy Tr. 1486; FF 229) When these errors are corrected, the HHI results look much different. As Dr. Murphy showed at trial, adding airline.com market shares to the "OTA booking services" market and attributing GDS passthrough sales to the GDS in the "TTA booking services" market yields post-merger HHI levels and increases that do not give rise to a presumption of harm to competition. (See Murphy Tr. 1485-86; FF 228, 230) DOJ's HHI analysis, therefore, does not satisfy DOJ's prima facie burden and does not entitle DOJ to a presumption of harm to competition.

B. DOJ Did Not Prove A Reasonable Probability Of Anticompetitive Harm

Because DOJ does not enjoy a presumption of anticompetitive harm, had DOJ proven a relevant product and geographic market, it would still be the government's burden to prove, considering all of the evidence – including that produced by Defendants to rebut DOJ's prima facie case – that the merger of Sabre and Farelogix is reasonably probable to substantially lessen competition. Even if the Court were to reach this part of the analysis (which it does solely as an alternative to the other deficiencies already identified in the government's case), the Court would conclude that DOJ has failed to meet its burden. Some of the reasons for this conclusion are discussed below.

1. DOJ Failed To Prove That Barriers To Entry Prevent Adequate Competition To FLX OC

"Barriers to entry are factors, such as regulatory requirements, high capital costs, or technological obstacles, that prevent new competition from entering a market in response to a monopolist's supracompetitive prices." Broadcom Corp. v. Qualcomm Inc. , 501 F.3d 297, 307 (3d Cir. 2007). While not a required element of a Section 7 claim, evidence of barriers to entry in the relevant market can "bolster[ ]" the government's case. FTC v. Univ. Health, Inc. , 938 F.2d 1206, 1220 (11th Cir. 1991) ; see also FTC v. H.J. Heinz Co. , 246 F.3d 708, 717 (D.C. Cir. 2001). In addition, defendants can rebut the government's prima facie case by showing that the relevant market exhibits low barriers to entry. See Energy Sols. , 265 F. Supp. 3d at 443.

DOJ identifies three barriers to entry in the relevant market: (1) the technological challenge of building an "NDC booking services solution;" (2) the "need for a good reputation and track record serving major airlines;" and (3) GDS contract provisions that discourage airlines and travel agencies from switching to "new entrants." (D.I. 233 at 32) DOJ has failed to persuade the Court that any one or a combination of these barriers significantly affects the analysis here.

The record does not establish that building an adequate "NDC booking services solution" is particularly difficult. As several witnesses testified, the schema for developing a marketable NDC API product is publicly available, which has allowed numerous companies to begin offering NDC API services. (Davidson Tr. 433, 439; Nevo Tr. 1038-39) For example, Farelogix Chief Technology Officer Tim Reiz testified that programmers can easily compete with Farelogix for NDC API services "[b]ecause they have web services that they can connect to the airline systems." (Reiz Tr. 1342-43) Marketplace evidence supports these witnesses. Datalex and Amadeus have successfully built and marketed NDC API services for several airlines. (FF 160-61) Likewise, Delta, British Airways, and Air France have built and supplied NDC APIs for themselves. (Carter Tr. 271; Davidson Tr. 446-47; Adair Tr. 1732; FF 151) While executives from American Airlines and United Airways believe only Farelogix can provide adequate NDC API services (Garner Tr. 121; Radcliffe Tr. 229-30), the totality of the evidence persuades the Court that these are personal (or company) preferences, not market realities.

Farelogix's vigorous competition with rival NDC API providers further undermines DOJ's remaining arguments. Farelogix has competed for bids to supply NDC API services with at least either other companies, losing bids to [Redacted] (Carter Tr. 270-71) While, in the overall context, Farelogix has competed well against other NDC API suppliers (see FF 164-70), and has some unique advantages (see FF 140-50), these other suppliers have also enjoyed success, demonstrating that new entrants can develop a "good reputation and track record" despite Farelogix's established position in the market. The large number of suppliers who have won bids further suggests that GDS contract provisions have not shut out new NDC API providers.

At trial, and again in the post-trial briefing, the parties devoted a great deal of attention to what was marked as Davidson Demonstrative 1, which listed NDC API projects Farelogix bid on in recent years, purporting to depict the results of its efforts. In the Court's view, DOJ has the more persuasive interpretation of this demonstrative. It shows – consistent with the facts supported by the record – that Farelogix has fared just fine in recent years. [Redacted] Furthermore, Farelogix currently provides its services to two of the largest U.S. airlines as well as the largest airlines in Australia and Mexico. (See Davidson Demonstrative I; Davidson Tr. 469; FF 166) Farelogix is currently working on 26 GDS integration installations, far more than any other NDC API provider (as far as the record reveals). The Court is not persuaded by Defendants' puzzling attempts at trial to portray Farelogix as a failure.

Notwithstanding all this, however, the reality that Farelogix is competing successfully against other NDC API suppliers does nothing to alter the fact that there arc not significant barriers to entry into, and success in, the NDC API market.

2. DOJ Failed To Prove That A Post-Merger Sabre Will Harm Competition By Eliminating FLX OC Or Raising FLX OC Or GDS Prices

The Court is persuaded that at various points Sabre has viewed FLX OC as a competitive threat. Among other things, Sabre believed that FLX OC was "a real alternative to the GDS" (PX367 at -277-78), and that Farelogix's NDC technology has [Redacted] (PX247 at -766). Further, the Court is persuaded that some at Sabre viewed the acquisition of Farelogix as way to neutralize this perceived threat, including employees who wrote that the deal would "[m]itigate risk from potential GDS bypass" (PX011 at 3; PX012 at 7), in a document that was at one point sent to Sabre's board (see, e.g. , FF 240-41). The record also demonstrates that, historically, Sabre has resisted change while Farelogix has been a pioneering innovator and disruptor of the airline travel services ecosystem. (See FF 101-08) Moreover, this evidence suggests that Sabre will have the incentive to raise prices, reduce availability of FLX OC, and stifle innovation.

Nevertheless, DOJ has not persuaded the Court that Sabre will likely act consistent with its history or these incentives and actually harm competition if it is permitted to complete the acquisition of Farelogix. This is yet another problem for the government's case because the claim the government has brought necessarily requires the Court to undertake a forward-looking analysis. See generally United States v. Baker Hughes , 908 F.2d 981, 988, 991 (D.C. Cir. 1990) (explaining that merger analysis "focus[es] on the future" and requires Court to "[p]redict[ ] future competitive conditions").

The evidence does not suggest that Sabre is acquiring Farelogix to eliminate FLX OC from the marketplace. Rather, it supports the opposite conclusion: that Sabre intends to continue offering FLX OC by integrating it into the Sabre GDS platform, allowing Sabre to better meet the demands of airlines and travel agencies. In a 2018 presentation to Sabre's board of directors, Sabre's management sought approval for the merger by arguing it would allow Sabre to integrate FLX OC's NDC capabilities. (See DX145 at -005; see also Boyle Tr. 555 (Sabre VP of Corporate Development testifying that deal is intended, in part, to "bring[ ] NDC-enabled distribution to scale in the marketplace"); DX086 at -037) Several industry participants have applauded the deal for this very reason. (See, e.g. , FF 256-60) All of this is consistent with the strong evidence that the future of airline ticket distribution is more likely to be characterized by GDS passthrough – i.e., integrating NDC capabilities into the GDS platform – rather than GDS bypass. (See, e.g. , PX081 at -045-47; supra FF Part N)

Further supporting these conclusions are the public commitments Sabre CEO Menke has made, including to continue offering FLX OC to airlines (at their current or better prices) after the merger closes. (See supra FF Part T) DOJ did not persuade the Court that Sabre is likely to raise prices for its GDS platform after the merger, even though the Court has found that the merger will likely reduce some of the leverage airlines currently have in negotiations with Sabre. (See FF 265, 279-83) While a Farelogix board member recommended that Sabre increase its post-merger GDS booking fees (see Boyle Tr. 494-95; PX007 at -734), no evidence has been presented that anyone at Sabre or Farelogix has taken steps to act on this recommendation. (See also FF 265-77) Also, because Sabre competes with rival GDS providers Amadeus and Travelport for travel agency customers, it faces constraints on the prices it can charge. Indeed, CWT CEO Kurt Ekert testified that "if ... Sabre was to lose content or enable content but on a noncompetitive basis to how another GDS provides that, then we have the ability to shift business away from them to another GDS." (Ekert Tr. 1186; see also Murphy Tr. 1433)

DOJ also failed to persuade the Court that the merger is likely to lead to higher prices for FLX OC. DOJ points to Sabre's January 2019 proposal to raise an airline's post-merger price for FLX OC services. (D.I. 233 at 25; see also Gilchrist Tr. 751-60; Wilding Tr. 828-30; PX392 at -026-27; FF 271) However, that airline rejected Sabre's proposal, with one Sabre employee reporting that the proposal "caused a kerfuffle." (PX393 at -079; see also Gilchrist Tr. 758) If anything, this evidence suggests that Sabre will face serious resistance if it tries to raise FLX OC prices after the merger.

Farelogix's rivals will likely further constrain Sabre's ability to raise prices for FLX OC. As discussed above, Farelogix now vigorously competes with – and often loses business to – other NDC API providers. Farelogix also faces some threat that airlines will self-supply NDC API services (although many airlines do not view this as an attractive option). Thus, it is unlikely that Sabre will be able to raise prices on FLX OC services without driving customers to competitively priced alternatives.

For at least these reasons, DOJ failed to prove that Sabre, after the merger, will likely harm competition by raising prices or eliminating FLX OC.

Because DOJ failed to prove a relevant market, the Court need not and will not address Dr. Nevo's SSNIP analysis in any greater detail than it already has elsewhere in this Opinion.
--------

3. DOJ Did Not Prove That The Merger Will Harm Innovation

"A merger can substantially lessen competition by diminishing innovation if it would ‘encourag[e] the merged firm to curtail its innovative efforts below the level that would prevail in the absence of the merger.’ " United States v. Anthem , 236 F. Supp. 3d at 229-30 (quoting 2010 DOJ and FTC Horizontal Merger Guidelines §§ 1, 6.4 (2010)) While DOJ argues that Sabre's purchase of Farelogix will "reduce competition to innovate" (D.I. 233 at 28), the Court is not persuaded. DOJ offers nothing more in support of its contention than vague theories from Dr. Nevo, who testified that the merger involves "taking an innovative firm that has been driving, driving ahead, creating the industry shakeup, and putting that under a firm that does not have the same incentive." (Nevo Tr. 950-51) These generalities do not help the Court conclude that the merger would harm innovation.

Moreover, the record evidence, overall, suggests the opposite. While it is undisputed that Farelogix developed the original NDC schema and has been an innovator (see FF 58; see also supra FF Part I), no party offered evidence that Farelogix has more recently created or introduced innovative products or services. Further, as has been repeatedly noted, the public availability of the NDC schema has allowed (and will continue to allow) new firms to enter the market and compete successfully with FLX OC.

DOJ also failed to persuade the Court that Sabre is purchasing Farelogix to delay innovation. (See D.I. 233 at 28) As discussed above, the evidence suggests that Sabre seeks to integrate FLX OC into its platform, not eliminate it, and that in doing so Sabre is following the industry trend towards GDS passthrough. (See, e.g. , supra FF Parts N, S)

Finally, the Court is persuaded that the merger may well promote innovation, as numerous witnesses have forecast. (See, e.g. , Ekert Tr. 1199-1201 (CWT CEO stating that merged Sabre-Farelogix could offer additional technology services to airlines and allow travel agencies to "drive more high yield traffic to the airlines"); Kunz-Cho Tr. 1316 (Fareportal CEO stating that merged entity "will be able to embrace a wider array of content, particularly in the low cost carrier domain which is very important to us")) Moreover, Farelogix Chief Technology Officer Tim Reiz credibly predicted that the merger will allow him to develop new projects (such as a next-generation PSS) using artificial intelligence, because he would have access to Sabre's resources, including data, the lack of which was previously a "roadblock." (Reiz Tr. 1346-47)

On balance, then, DOJ has not proven that the Sabre-Farelogix merger will harm innovation.

CONCLUSION

The Court recognizes that the outcome here may strike some, including the litigants, as somewhat odd. On several points that received a great deal of attention at trial – whether Farelogix is a valuable company enjoying relative success in the market, whether Sabre and Farelogix compete, whether Sabre understands GDS bypass is a threat, whether Sabre stands to lose revenue even from the expansion of GDS passthrough, and Sabre's motivation for its proposed acquisition of Farelogix – the Court is more persuaded by DOJ than by Defendants. This is largely due to the surprising lack of credibility on these points of certain defense witnesses, including Sabre CEO Menke, Sabre deal leader Boyle, and Farelogix CEO Davidson.

Despite these findings and conclusions, however, Defendants have won this case. This is because the burden of proof was on DOJ, not Defendants. Defendants opted to tell the Court a story that is not adequately supported by the facts, but it was their choice whether to do so, and their failing does not determine the outcome of this case. Instead, it is DOJ which, under the law, has the obligation to prove its contention that the Sabre-Farelogix transaction will harm competition in a relevant product and geographic market. DOJ failed. It based its case on the expert analysis of Dr. Nevo, but that analysis – including Dr. Nevo's explanation and defense of it – was simply unpersuasive. Unlike Defendants' evidentiary failings, DOJ's are dispositive.

Under our laws, and in our (regulated) market economy, private entities like Sabre and Farelogix are generally free to enter into agreements and relationships with one another, whether or not the government prefers that they do so. If DOJ is to get the Court to enjoin such a transaction, it must meet its burden of proof. Here, the government has not done so. Accordingly, the Court must enter judgment for Defendants.

APPENDIX

CLOSING THE COURTROOM AND SEALED PORTIONS OF THE RECORD

While the public enjoys a "common law right of access [ ] to judicial proceedings and records," this right is "not absolute." In re Avandia Marketing, Sales Practices and Prods. Liability Litig. , 924 F.3d 662, 672 (3d Cir. 2019). To overcome the presumption of access, parties must show that the information they seek to protect is "the kind of information that courts will protect and that disclosure will work a clearly defined and serious injury to the party seeking closure." Id. at 672 ; see also Genentech, Inc. v. Amgen Inc. , C.A. No. 17-1407-CFC D.I. 659 at 6 (D. Del. Mar. 30, 2020). Such information includes "sources of business information that might harm a litigant's competitive standing." Avandia , 924 F.3d at 679. Moreover, in granting a party's request to protect information, courts "must [1] articulate the compelling, countervailing interests to be protected, [2] make specific findings on the record concerning the effects of disclosure, and [3] provide[ ] an opportunity for interested third parties to be heard." Id. at 672-73 (internal quotations omitted and numerals added). Further, they must "conduct a document-by-document review of the contents of the challenged documents." Id. at 673 (internal formatting and quotations omitted).

This issue often emerges in trials of government challenges to a proposed merger based on antitrust concerns, during which witnesses almost inevitably discuss sensitive business information such as prices, forward-looking strategy, and analyses of competitive conditions. To balance the public's right of access with the parties' interest in avoiding competitive injury, courts routinely close the courtroom when witnesses discuss "competitively sensitive information," while otherwise allowing full public access. See, e.g. , FTC v. Tronox Ltd. , 18-cv-01622 D.I. 80 at 8-9 (D.D.C. July 26, 2018) (agreeing to seal courtroom for witness testimony about "competitively sensitive information"); FTC v. Wilhelmsen et al. , 18-cv-00414 D.I. 62 at 85 (D.D.C. May 29, 2018) ("[W]e will have portions of the trial in which there's confidential commercial information that will be elicited. And so I'm going to have to close the courtroom briefly."); FTC v. Sysco Corp. , 15-cv-00256 D.I. 183 at 465-66 (D.D.C. June 26, 2015) (sealing courtroom to "accommodate confidential business information").

The Court followed this approach during trial. At various points, the litigants and several third parties asked the Court to close the courtroom to protect competitively sensitive information, such as forward-looking business analyses, commercial agreements, and strategic plans. (See, e.g. , D.I. 212 at 3-5; D.I. 219 Ex. 1 at 1-2) There were no objections to these requests. The Court evaluated each request separately, and granted requests to close the courtroom after describing the requesting party's interest in protecting commercially sensitive information or specifically referring to prior rulings where it had articulated this interest. (See, e.g. , Tr. 20 ("[T]here is some forward-looking financial competitively sensitive information that ... is not something that ... the public necessarily has a right to know because of the clearly defined and serious injury that it would cause to those defendants and to the third parties, particularly as they contemplate further transactions, negotiations, [and] agreements, in the future."); Tr. 87 ("So I am persuaded for the reasons discussed earlier, the reasons in the letters I received over the weekend, the declaration that I received over the weekend as well as further discussion today that, particularly what I have been told about the NDA, that this witness is bound by ... that it is appropriate to close the courtroom for limited portions of the testimony that will discuss those commercially sensitive negotiations."); see also Tr. 196-97, 229, 360-61, 662-63, 1088) The Court also allowed limited redactions of documents for the same reasons it occasionally limited access to the courtroom and ensured that the parties made the complete redacted record of the trial available to the public in a timely manner. (See D.I. 252, 254, 256, 259, 262, 264, 266, 268; see also Tr. 1742-43)

The parties also provided the Court with a helpful summary of the limited information they believed would need to be redacted if it were included in this Opinion. (See D.I. 271) With the parties' cooperation, the Court will be able to issue a public version of this Opinion quickly, notwithstanding its length.

The Court commends counsel for working cooperatively with one another to ensure the efficient presentation of the evidence at the expedited trial while assisting the Court in ensuring the public's access to the courtroom, the trial record, and this Opinion.

ORDER

At Wilmington this 7th day of April, 2020 :

For the reasons stated in the Opinion issued this same date, IT IS HEREBY ORDERED that judgment is entered FOR Defendants and AGAINST Plaintiff.

IT IS FURTHER ORDERED that, because the Opinion was issued under seal, the parties shall meet and confer and provide the Court a proposed redacted version, consistent with their submission earlier today (see D.I. 271), no later than April 8, 2020 at 10:00 a.m. Thereafter, the Court will release a public version of its Opinion.

IT IS FURTHER ORDERED that the parties shall meet and confer and, no later than April 14, 2020 , submit a joint status report, advising the Court as to what, if anything, remains to be done in this case before it is closed.