From HOA Management to $4B in Revenue & $6.3B Deal for Amex GBT | Alexander Taubman, Long Lake
Summary
Long Lake’s announced $6.3 billion take-private of American Express Global Business Travel is a bet that century-old customer trust can become a scaled distribution channel for AI. The 111-year-old company serves close to half the Fortune 500, much of the Global 2000, and government and military customers; Taubman sees a chance to invest in growth and technology across a multi-trillion-dollar travel market growing faster than GDP.
Taubman argues that the binding constraint on AI returns has shifted from model intelligence to deployment inside the real economy. Legacy data, fragmented systems, embedded workflows, and change management make diffusion difficult, even as frontier capability advances rapidly. Because trillions of dollars of future model investment ultimately require real cash returns, Long Lake aims to be “distribution for the labs.”
Long Lake’s thesis is productivity-led growth, not the familiar private-equity playbook of replacing employees and cutting costs. If AI makes a worker 50% more productive, improves customer retention, and creates capacity for 50% more clients, Taubman’s response is to hire more of that worker: “If you have a salesperson with AI that’s now doing 3x quota,” one-third the headcount is plainly the wrong conclusion.
Nexus, Long Lake’s internal AI platform, turns ownership across five service verticals into a shared technical advantage. Roughly 70-80% of its routing, security, and orchestration infrastructure transfers across industries; the difficult final 20-30% requires engineers embedded with frontline employees. An HOA email workflow that can take 45 minutes to an hour or longer can be reduced to 5-10 minutes, and the capability can now address travel, where 55% of agent inquiries arrive by email.
Ownership supplies the alignment needed to change workflows, but Long Lake still refuses to mandate adoption. Taubman describes roughly 30,000 global team members as Nexus’s customers and design partners: “If we have to force someone to use our Nexus platform, then the platform is not good enough.” Adoption that once took about six months can now happen in days, strengthening employee retention as well as customer service.
The combined business is expected to exceed $4 billion of revenue next year while retaining the economics of established, profitable service companies. Long Lake targets businesses already producing 20-25%+ EBITDA margins, and Taubman says existing cash generation means it may never need to raise capital again. Amex GBT adds high-90s logo retention, 100%+ dollar retention, and roughly 15-year relationships with its leading customers.
Long Lake combines an enormous opportunity set with an unusually selective capital-allocation process. It reviews 4,000-5,000 deals annually across 15-20 force-ranked industries but enters only one or two new sectors a year; it has acquired close to 40 companies and has never sold one. Taubman believes public markets could eventually provide the lower-cost acquisition currency for the “maximally ambitious path,” while acknowledging that compounding toward a trillion-dollar company would require decades of unusually consistent execution.
Deep dive
1. Amex GBT is a $6.3 billion bet on trusted distribution
Taubman says Long Lake has announced the $6.3 billion acquisition of American Express Global Business Travel, a 111-year-old business founded in 1915 to help American Express Travelers Cheques customers leave Europe during World War I. The objective is to take it private, invest in growth and technology, and bring AI into the experience of millions of travelers.
The franchise serves close to half the Fortune 500, a large share of the Global 2000, and numerous US and foreign government and military customers. Its strongest relationships average roughly 15 years, with high-90s logo retention and 100%+ dollar retention: evidence that its “century of customer trust” is operational, not ornamental.
Travel is a multi-trillion-dollar global sector growing faster than GDP, but its supplier, customer, and distributor systems carry decades of fragmented data. Taubman sees that complexity as the opportunity: AI could make service more seamless, personalized, and proactive without weakening the human support required when travel goes wrong.
2. The intelligence frontier has outrun the real economy
Taubman’s starting contrast is deliberately extreme: roughly four years ago, impressive chat models still struggled with addition; he says that if someone had predicted then that, four years later, models would solve some of history’s hardest unsolved mathematics problems and may win a Fields Medal, many people would have thought it crazy. Even he, already “super AI-pilled,” did not expect frontier capability to advance this quickly.
Yet GDP has not doubled or tripled, and most businesses still operate conventionally. His explanation is the diffusion problem: legacy systems, legacy data, workflow redesign, and human change management make deployment “actually really hard,” especially in businesses built long before modern software.
The financing loop matters to the thesis. As model investment moves into the trillions of dollars, future cash flows must justify today’s infrastructure spending; Long Lake’s role is to convert intelligence into high-ROI operating improvements, then generate cash that can flow back toward further scaling. Senra’s realization: “You’re distribution for the labs.”
3. AI productivity should expand payrolls, not gut it
Senra explicitly rejects the caricature of technologists and private-equity buyers acquiring service companies, removing employees, and booking cost savings. Taubman calls Long Lake’s view “very positive sum”: remove data entry and process-heavy toil, free people for customers and growth, and let higher productivity expand the business.
The operating math is straightforward. If one employee can support 50% more clients with better service, retention rises and growth accelerates; that makes the employee more valuable, not redundant. Taubman’s sales analogy is sharper: when an AI-enabled salesperson reaches 3x quota, “you obviously want as many of those people as you can have.”
Long Lake began in homeowner-association management, where it has acquired more than 20 companies, and has since added HR services, specialty tax, infrastructure services such as architecture and engineering, and travel. Across close to 40 total acquisitions, Taubman says the older businesses are showing meaningful job growth.
4. A wide search funnel produces concentrated conviction
Long Lake maintains a changing list of roughly 15-20 attractive industries, continuously force-ranked as markets evolve. It sees 4,000-5,000 deals annually but enters only one or two new industries per year—a “big funnel, very tight filter” designed to preserve flexibility until an exceptional business appears.
Taubman borrows Buffett’s joke that being open to both public and private assets doubles the chance of getting a date. Travel was not in Long Lake’s founding plan; after about a year of studying several assets, Amex GBT’s combination of customer durability, mission-critical service, and AI potential made the answer feel obvious.
The mission-critical test appears during disruption. If thousands of employees are traveling in the Middle East when war begins, or a snowstorm strands a company’s team in Chicago at 2 a.m., someone must answer and solve the problem. Travel may be a small corporate cost, but “you don’t want to screw it up.”
5. Nexus turns cross-industry ownership into reusable infrastructure
Nexus is Long Lake’s horizontal AI platform. On one side it connects to multiple models—Long Lake is model-agnostic and may combine several models for one task—and on the other it connects to each company’s data, software, and workflows.
Roughly 70-80% of Nexus’s infrastructure, including model routing, security, and agentic orchestration, is reusable across verticals. The difficult final 20-30% is mapping real workflows and integrating messy operational systems; every new industry therefore both consumes the platform and adds capabilities to it.
Taubman estimates that roughly 70% of Long Lake’s engineering team is sitting with team members in the field across 15-20 states. The model echoes Kelly Johnson’s Skunk Works principle that “you want the engineer sitting in the factory,” paired with Danaher’s maxim: “Common sense rigorously applied.”
In HOA management, a request for a snow-removal vendor may require comparing communities, prices, reviews, and availability before drafting a response—a 45-minute-to-an-hour-or-longer job. Nexus can assemble a reportedly 99%-accurate draft overnight, reducing the manager’s work to 5-10 minutes and increasingly matching that manager’s own voice.
6. A solution built for HOA email transfers directly into travel
The HOA workflow became a cross-vertical asset because most businesses still run heavily through email. Taubman says 55% of inbound traveler requests to agents—new bookings, changes, and cancellations—also arrive by email, giving Amex GBT an immediately applicable capability rather than a greenfield experiment.
Long Lake has declined to turn Nexus into a conventional software product despite interest from investors. Taubman’s objection is alignment: a software vendor can ship a tool without controlling whether workflows change or outcomes improve, whereas an owner-operator is accountable for growth, customer experience, employee experience, compensation, and shareholder returns.
Senra frames this as a human problem rather than a software problem, but Taubman adds an important qualification: ownership does not justify coercion. “If we have to force someone to use our Nexus platform, then the platform is not good enough.” The product must earn adoption through usefulness.
7. Employees are Nexus’s customers and Long Lake’s retention moat
Taubman describes roughly 30,000 team members globally, whom he calls team members, customers, and design partners. Nexus is meant to feel “so magical” that it spreads virally; deployment that once took perhaps six months for a cohort can now occur within days.
Better tools also create an employment moat. Employees describe losing the mental friction associated with the roughly 30% of work spent on repetitive email and data entry; they can instead serve customers, pursue growth, enjoy their jobs more, and potentially make more money.
The resulting flywheel is positive-sum: better tools attract better people; better people produce better customer experiences; higher retention and faster growth generate more profit; and greater profit supports higher pay and further investment. Leaving for an unequipped competitor becomes increasingly unattractive.
8. Long Lake’s organization concentrates on three scarce capabilities
Taubman says the strategy requires three world-class functions simultaneously: acquiring great companies, building technology competitive with leading applied-AI companies or labs, and executing operational change management. Excellence in only one or two leaves the deployment problem unsolved.
Long Lake itself remains small—about 55 or 56 people at the time of recording, with an expectation of approaching 100 by year-end. Roughly 90% work in AI engineering, operations, and deployment, while only about 10% focus on acquisitions, M&A, and finance.
Hiring mirrors deal selection. About 75% of the team has come through referrals, and Long Lake hired roughly 3% of nearly 1,000 applicants in the referenced quarter. Taubman’s central operating lesson is “high bar in everything”: talent, acquisitions, and technology.
The founding network dates to Goldman, where Zach was a summer analyst and Taubman an analyst; Zach later introduced CTO Rasmus, an Oxford computer-science standout. Their early history also included combining roughly 20 locally owned go-kart tracks into a platform.
9. “Threshold resistance” connects Taubman’s family history to AI adoption
Taubman’s grandfather built a real-estate business after World War II and treated retail design almost mathematically. Stores once hid entrances behind display windows or dark passageways; Taubman says even half a step at the threshold could reduce the number of people entering by 20%.
That idea became Taubman’s analogy for Nexus: remove every bit of “threshold resistance” until using AI is effortless. The name Long Lake honors the Michigan road where his family’s business operated for roughly 40-50 of its 75 years, linking the new company to an older culture of continuous improvement.
Michigan also taught him to admire low-profile entrepreneurs in “unsexy” sectors. One Long Lake partner, a Minnesota architecture firm, has built about half the state’s K-12 schools over 70 years. His mission is to bring elite engineers to precisely these enduring businesses while letting their brands and operators remain the protagonists.
10. An investing apprenticeship produced a permanent-capital model
At Goldman’s Special Situations Group, Taubman gained exposure to public and private investments across the capital structure. A mentor, Patrick McCan, let him help manage a few hundred million dollars of invested capital at roughly 22 or 23, giving him unusually early responsibility for real risk.
After business school—and time at Leucadia and with Josh at Thrive—he spent nearly a decade at Oaktree, working with Howard Marks and Bruce Karsh to develop equity, credit, and other investment products. He loved capital allocation but felt he was not “fully expressing” an entrepreneurial impulse.
In parallel, his family investment vehicle completed close to 50 partnerships with founder-owned businesses. Senra cites returns around a 70% IRR; Taubman declines to validate the number beyond saying, twice, “We had a very good run,” preserving the distinction between the host’s claim and the guest’s confirmation.
The lesson was that smaller companies can be extraordinary businesses. Permanent, flexible capital could fund growth and acquisitions without excessive leverage or quarterly earnings pressure, helping founders consolidate competitors they understood better but had never before possessed the resources to acquire.
11. Flexible capital allocation supports a trillion-dollar ambition
Including Amex GBT, Long Lake expects more than $4 billion of revenue next year, with a stable revenue base and roughly 100% dollar-retention-type economics. Its acquisition strategy targets established companies already producing 20-25%+ EBITDA margins, so Taubman believes existing cash generation could fund compounding for decades without another capital raise.
External capital still increases speed and ambition. An early partner who served as Long Lake’s general counsel, led its first round of outside capital, and provided its first $100 million was among those who believed early. Taubman values partners who challenge the team and ask “why not do them all?” He and Senra agree that enduring partnership matters more than optimizing each financing round’s valuation.
Going public is not required, but Taubman calls it the “maximally ambitious path”: deeper capital pools and a lower-cost public currency could finance progressively larger acquisitions. He invokes Henry Singleton, who compounded earnings per share by over 40% a year for 20-25 years while preferring to “steer the boat a little bit every day” rather than follow rigid strategic plans.
John Malone’s first-principles habit—asking why repeatedly until reaching the root cause—frames Long Lake’s post-AGI underwriting. Senra argues that well-positioned businesses can become orders of magnitude more valuable while lower-quality intermediaries disappear. Long Lake has never sold a company and does not plan to, though he retains the hedge “never say never.”
The outer ambition is a trillion-dollar company: tripling value every five years for 20 years can produce staggering outcomes across a market encompassing much of the economy. Taubman refuses to present that as easy; it requires decades of avoiding major mistakes, retaining elite talent, improving technology, and sustaining customer excellence across every service line.