Pershing Square Challenge 2026 finalists pitch Amadeus $AMS | the toll booth on global travel

Summary

• The team’s core pitch: Amadeus is “the toll booth on global travel” — a fee every time a flight is booked or a passenger boards — with ~50% share in its core segments, 7% revenue growth, 11% EPS compounding, and 400+ airline connections built over 30 years. At 15x earnings versus a historical 22–24x and a DCF-fair ~22x, with 8% topline / 15% EPS growth projected and ~3% dividend yield, they get “close to a 20% return over the next three to four years” with zero multiple rerating assumed.
• The AI-disruption fear that crushed the stock is, in their primary work, misdirected: air IT solutions is a mission-critical, deterministic system of record costing around 1% of revenue. “If the system is not working, the plane will just not take off.” Their best specimen: a Microsoft AI director — the person you’d expect most bullish on AI eating software — said on PSS specifically, “I would bet my money on that this is not going to go away in five, ten years.”
• The distribution business is protected by economics AI doesn’t change: inference runs ~30x the cost of an API call for a deterministic task, scraping airline content violates terms and prices are too dynamic anyway, so partnerships are the only road — and Amadeus has 400+ of them. Google already tried, buying ITA in the mid-2010s, watched the tech go obsolete, and “came back and partnered with Amadeus last year”; a $6B GDS revenue pool isn’t worth chasing when OTA demand aggregation is a $65B pool. LLM uptime of ~99.3% (~87 hours/year down) versus the ~99.99% standard discussed (~50 minutes) seals it.
• Andrew Walker’s pushbacks are worth the listen: “it doesn’t look that obviously cheap to me” at ~10x EBITDA for a GDP-plus grower, and the tech debt behind airline counters “running 1990s tech” invites eventual replacement. The team’s answers: Andrew considers the company’s EBITDA and adjustments clean, most SaaS is far pricier on comparable GAAP numbers, and a former Amadeus SVP of engineering said no one swaps a working system costing “about 1% of your revenue” for one that “might work well” — migration means running two parallel instances.
• The Sabre subplot is signal for Constellation Software watchers: Sabre (15% share, 3x smaller, ~9x levered after repeated PE ownership) accused Amadeus of monopoly on its Q1 call and asked for a DOJ investigation. The team’s history lesson — Sabre was once #1, but leverage starved reinvestment while Amadeus, in COVID, “gave them better payment terms in exchange for larger content deals” and took share Sabre couldn’t contest.
• A surprise ~$1B biometrics acquisition landed the morning before their Pershing Square finals — Andrew flags M&A as a classic ex-growth tell, but the team got comfortable: opportunistic PE exit, sticky airport/government customers, 10x EBITDA paid that pencils to ~6x by 2028–29 with synergies, and accretive to their IRR.
• The galaxy-brain upside both sides entertain: if Amadeus truly can’t be ripped out, it’s an AI beneficiary, not victim. It employs more engineers than Microsoft Office and Outlook combined; experts told them to expect a 1.5–10x productivity boost, and even the 1.5x low end grows earnings ~10% — though Andrew notes Sabre’s 30% higher revenue-per-employee may just measure underinvestment, itself a bull case for Amadeus share gains.

Deep dive

1. Why this pitch: an “AI-proof” SaaS down 30%, with a Constellation kicker

• Andrew’s setup for why Amadeus keeps hitting his radar: multiple people have pinged him calling it “about the most AI-proof company in SaaS land that I can think of” after a ~30% slide, and Constellation Software bought ~10% of publicly traded competitor Sabre — “who is quite levered” — on the open market.
• The team — Kabir and Fran (both ex-consulting), plus a third teammate ([Ethan?] — ex-buy-side and corporate strategy), all CBS MBAs — ran a disciplined funnel: each brought 2–3 company primers every three days for weeks, then ranked and voted. Amadeus won on Pershing-portfolio fit plus “the most interesting thesis, mostly because of the AI selloff during that period.” Judges told Andrew this was the best set of contestants they had seen by far; one was “pounding the table” on this pitch.

2. What Amadeus actually is: “the toll booth on global travel”

• Kabir’s one-liner: “Amadeus is like the toll booth on global travel” — it earns a fee every time a flight is booked through a travel agency or an airline boards a passenger. #1 player with roughly 50% share in core segments, 7% revenue growth, 11% EPS compounding, across three segments: air distribution (two-sided platform connecting 400+ airlines to travel sellers), air IT solutions (airline operations backbone), and hospitality (hotel reservation/guest systems).
• Fran’s mental model for IT solutions: “the SAP of airlines” — check-in, luggage, airport systems, takeoff, everything. Distribution shows airline content — flights, prices, seats — to OTAs (Expedia, Booking.com), brick-and-mortar agencies, and TMCs like Amex Travel and BCD, and handles the booking internally.
• Kabir’s caveat on the SaaS comparison: this is not a per-seat business — Amadeus sells directly to airlines and aggregates airline content for travel sellers — “so it’s not impacted by whether there’s going to be lesser users in the future.”

3. The valuation argument: cheap, but Andrew isn’t fully sold

• Andrew’s honest framing: stock from low-70s to mid-50s, ~10x EBITDA / 15x P/E, mid-to-high-single-digit grower, buybacks just started — “it’s just not striking me as like this huge dislocation… I came in thinking it was a SaaS victim and it’s just not obvious that it is to me.”
• [Ethan?]’s rebuttal: “it is cheap to me.” Historical range 22–24x P/E, now 15x; a DCF spits out ~22–23x as fair; consensus EPS growth ~12%, in line with guidance, on an essentially unlevered company. Against software comps you must use GAAP numbers — and on GAAP, most SaaS “is a lot more expensive than 15 times.”
• Andrew agrees that the company’s EBITDA and adjustments are clean, contrasting with his usual gripe about companies claiming $500M free cash flow atop $700M of stock comp: “the EBITDA and the adjustments here is very clean.”

4. Slide 15: mission-critical + deterministic = least AI-exposed software in their matrix

• The deck’s 2x2 — Andrew calls it “one of the best slides I’ve seen” — plots AI exposure: Duolingo worst-positioned, with Amadeus air IT solutions and air distribution placed as safer than Visa, Oracle, or SAP. Fran’s logic: 40–50-year-old systems of record, fully relied upon, at around 1% of revenue. “If the system is not working, the plane will just not take off. As simple as that.”
• The load-bearing primary call: a Microsoft AI director who builds agents for enterprises and governments — “you would think that he would be the most bullish on AI is going to eat software” — said the opposite: mission-critical deterministic systems of record aren’t being replaced, and on PSS specifically, “I would bet my money on that this is not going to go away in five, ten years.”
• Andrew’s pushback: “the AI systems we’re working with today are the worst we’re ever going to work with,” and behind the JFK counter “it’s like running 1990s tech” — won’t tech debt eventually force replacement, or won’t startup airlines begin debt-free? Kabir, via a former Amadeus SVP of engineering: you wouldn’t “replace a system that’s working too well that costs about 1% of your revenue with something that might work well” — migration means running two live instances in parallel, and Amadeus keeps reinvesting (Nevio, with Finnair an early adopter).
• The team’s customer research added that the technical-debt concern was mostly about large U.S. airlines running 20–30-year-old internal software — “one of the biggest opportunities for Amadeus to come in and take.”

5. The Sabre subplot: a levered #2 crying monopoly

• Andrew, prepping from Sabre’s Q1 call: Sabre accused Amadeus of taking share “because they are a monopoly” and asked for a DOJ investigation — “I’ve never heard an analyst ask someone on a call, hey, your competitor accused you of being a monopoly.” Amadeus’s answer: customers are voting with their dollars because the product is better — Andrew’s analogy, Netflix versus legacy cable.
• The team’s history: Sabre was once #1, but repeated private equity ownership left it ~9x levered and starved of reinvestment while Amadeus compounded product quality and network effects. In COVID, Amadeus “gave them better payment terms in exchange for larger content deals” — brilliant share-taking that Sabre, at 5–6x leverage going in, couldn’t match. Core-market shares today: Amadeus ~50%, Sabre ~15%.

6. Why nobody rebuilds distribution — not even Google

• Andrew’s Uber-driver challenge: drivers multi-home across five phones, so why wouldn’t airlines connect to any new AI-native distributor for free? Fran’s answer: there are only two ways to extract airline content — scraping (against airline terms, and pricing is too dynamic anyway) or API partnerships, and Amadeus’s 400+ airline connections took 30 years to build. Leaving means a carrier’s flights effectively vanish from travelers’ view; agencies would forgo Amadeus incentives. Fran’s structural point: “There are only three companies in the world. That’s for a reason.”
• The Google case study: it bought ITA in the mid-2010s to enter the space, the technology became obsolete, and it “came back and partnered with Amadeus last year.” A large AI firm can’t justify building for a $6B distribution revenue pool when it sits atop a $65B OTA demand-aggregation pool. And LLM uptime of ~99.3% (~87 hours down per year) versus the ~99.99% standard discussed (~50 minutes) “will not work for an airline” — plus inference costs ~30x an API call for what is a deterministic task.

7. The biometrics surprise, the return math, and the AI-beneficiary flip

• Amadeus announced a ~$1B biometrics acquisition the morning before the finals — “we got completely caught by surprise.” Andrew flags M&A as a classic tell that growth is plateauing; the team’s post-mortem: high-quality PE-owned asset exiting opportunistically, sticky airport and government customers, fits the “orchestrator of all of travel” strategy, and 10x EBITDA paid pencils to ~6x by 2028–29 with synergies — accretive to their IRR.
• The return math: 8% topline / 15% EPS growth for the next 3–4 years, ~3% dividend yield, no rerating assumed despite a 22x fair multiple — “close to a 20% return over the next three to four years” just holding.
• Andrew’s galaxy-brain close, which the team embraces: if this truly can’t be ripped out, Amadeus is an AI beneficiary — margin expansion from engineer productivity plus possibly GDP-plus-plus travel growth. Kabir: Amadeus has more engineers than Microsoft Office and Outlook combined; experts at Gemini and Anthropic expect a 1.5–10x productivity boost, and even 1.5x grows earnings ~10%. Andrew’s counter on the Sabre benchmark: 30% higher revenue-per-FTE at an overlevered Sabre may just measure chronic underinvestment — “maybe it’s a bull case for Amadeus on the other side,” ceding share over the long run.