Amex Global Business Travel: The World's First AI Take Private with Long Lake CEO Alexander Taubman

2026-05-11 - 22 min - source - Read full transcript

Elad Gil (host)Alexander Taubman

Key insights

Long Lake buys operating businesses instead of selling them software because ownership gives it direct control over change management and workflow redesign.

Taubman argues the standard Silicon Valley playbook (build software, sell it as a vendor to an industry) fails here because a vendor doesn't control the buyer's processes or org design. Owning the company and the customer relationship lets Long Lake drive the change management that AI adoption actually requires.

ai-driven-roll-ups

Nexus is a shared horizontal AI platform, not a bespoke build per acquisition.

Roughly 80% of the AI infrastructure is common across all verticals; the remaining work is mapping each new company's workflows, cleaning and integrating its data sources, and connecting them to a model-agnostic layer. This is what lets Long Lake redeploy the same platform into radically different industries (HOA management, architecture, HR services, specialty tax, and now travel).

ai-driven-roll-ups

Time-to-impact after acquisition collapsed from over a year to days as the platform matured.

Long Lake's first deals took over a year to find and realize AI's potential in the business. With Nexus built out, new acquisitions now see measurable time savings and productivity impact within days of closing.

ai-driven-roll-ups

The firm deliberately optimizes for growth and customer experience, not cost-cutting, unlike traditional private equity.

Portfolio companies typically grew 0-5% a year pre-acquisition and now grow 20%+ organically. Taubman frames this as a structural fix: labor-intensive service firms keep only about 20 cents of each incremental revenue dollar after hiring costs, so growth was effectively taxed at a punishing marginal rate. Making existing teams 30-40% more efficient with AI removes that constraint and lets the business behave like a software company with high incremental margins.

ai-productivity-flywheel

Employee productivity gains create a retention flywheel that makes it costly for talent to leave.

Workers who leave a Long Lake-owned company lose access to Nexus and have to resume the manual, low-leverage work Nexus automated, which Taubman compares to giving up email. Because employees are more productive, Long Lake can also pay them more, reinforcing retention and making the company a talent magnet within traditionally low-tech industries.

ai-productivity-flywheel

Executing this model requires three disciplines that rarely coexist in one team: PE-style deal execution, applied AI engineering, and hands-on change management.

Taubman says he has talked to dozens of people attempting AI roll-ups and few combine all three. Long Lake's early hires came almost entirely through founders' personal networks (100% of the first 20 people), pulling engineers from Palantir, Ramp, Robinhood, and Glean and M&A talent from GTCR, Blackstone, TPG, and HIG - explicitly targeting PE professionals who believe in the AI thesis but work at firms that are not AI-native.

talent-and-retention

Long Lake's engineers are embedded on-site with portfolio company teams rather than working remotely from headquarters.

Taubman describes engineering staff distributed across roughly 20 states, co-located with team members in each acquired business, deliberately mirroring the 'skunk works' principle of putting engineers and the factory floor together to shorten feedback loops and identify real pain points to automate.

ai-driven-roll-ups

The $6.3 billion Amex GBT take-private is framed as the first AI-native buyout of a public company, applying the same Nexus thesis to a 111-year-old business.

Amex GBT traces to 1915 (originally booking rail and boat travel for American Express customers stranded in Europe during WWI) and recently acquired Carlson Wagonlit, founded in 1876. Taubman positions the deal as doubling down on Amex GBT's existing AI transformation strategy, giving travel counselors 'AI superpowers' to improve response times and issue resolution.

ai-driven-roll-ups

Long Lake explicitly models itself on Danaher and TransDigm: compound advantage over decades rather than the buy-improve-flip cycle typical of private equity.

Taubman says building a category-leading company only to sell it in a few years 'doesn't make sense' because the productivity and trust advantages compound over a multi-year (not multi-month) transformation cycle. He wants to be a permanent capital partner and expects Long Lake's cost of capital to fall further as the model proves out, similar to how Danaher and TransDigm built pricing power by demonstrating a repeatable operating system.

long-term-ownership-model

Long Lake wins competitive deals and off-market sales partly because it offers founders equity rollover and alignment, not just capital.

Sellers get a cross-functional deal, engineering, and change-management team embedded from day one, and founders/management are encouraged to roll equity into the new business so they share in the AI-driven upside. Taubman says this has produced significant rollover participation across Long Lake's first four verticals and increasingly makes Long Lake the only buyer some owners want to sell to.

deal-sourcing-and-alignment

Enterprise AI adoption remains extremely early, which Taubman treats as the core opportunity.

He estimates real enterprise AI use cases are only about 1% penetrated. Most of the US economy is small businesses without resources to build AI tooling, and even large companies struggle to drive impact from it, leaving a large gap between what foundation-model labs are building and what actually gets deployed in day-to-day operations.

ai-driven-roll-ups

Target industries are chosen with a 'prepared mind' framework favoring mission-critical, high-cost-of-failure services.

Long Lake maintains a running list of roughly 15-20 target industries; corporate travel qualified because trips are typically revenue-generating and failures (missed flights, botched bookings) are costly, which rewards the trust a century-old brand like Amex GBT has built and gives Nexus's reliability improvements outsized value.

deal-sourcing-and-alignment

Companies

Long Lake - AI-native holding company Taubman co-founded and runs as CEO; acquires service businesses and layers a shared AI platform across them.
American Express Global Business Travel (Amex GBT) - 111-year-old corporate travel platform Long Lake agreed to take private for $6.3 billion, discussed as the episode's centerpiece deal.
Carlson Wagonlit - Travel management company founded in 1876 that Amex GBT acquired in late 2025, cited to underline the century-plus age of the assets Long Lake buys.
GTCR - Top private-equity firm Long Lake's M&A co-lead left, reportedly named employee of the month there the same month he departed for Long Lake.
Blackstone - One of the large-cap PE firms Long Lake has recruited M&A talent from.
TPG - One of the large-cap PE firms Long Lake has recruited M&A talent from.
HIG Capital - One of the large-cap PE firms Long Lake has recruited M&A talent from.
Palantir - One of the source companies for Long Lake's early engineering hires.
Ramp - One of the source companies for Long Lake's early engineering hires.
Robinhood - One of the source companies for Long Lake's early engineering hires.
Glean - One of the source companies for Long Lake's early engineering hires.
Danaher - Cited repeatedly as the operating-model template Long Lake is following: compounding growth through a differentiated operating system and long-term ownership rather than PE's typical hold-and-flip cycle.
TransDigm - Named alongside Danaher as an example of an operator that built pricing power and cost-of-capital advantage by proving out its operating model over many deals.

Techniques and frameworks

Nexus platform - Long Lake's horizontal AI infrastructure, roughly 80% shared across verticals, that sits between foundation models and each portfolio company's data, workflows, and skills; customized per acquisition through workflow mapping and data integration.
AI-driven roll-up / AI take-private - Taubman's term for buying operating businesses outright (rather than selling them software) so the acquirer controls change management, workflows, and org design directly.
Danaher Business System-style compounding - Long-term ownership model: buy, apply a differentiated operating system, reinvest gains into growth, and never sell, so improvements compound across a multi-year transformation cycle.

Summary

Elad Gil interviews Alexander Taubman, co-founder and CEO of Long Lake, about the firm's roughly $6.3 billion agreement to take American Express Global Business Travel private, which Taubman frames as the first AI-driven take-private of a public company. Long Lake's underlying model, tested across roughly 30 prior acquisitions in HOA management, architecture, HR services, and specialty tax, is to buy traditional, labor-intensive service businesses outright and layer a shared horizontal AI platform called Nexus across them, rather than selling AI software as a vendor to industries that are slow to adopt it.

Taubman's central argument is that ownership, not vendor relationships, is what makes AI transformation actually happen. Roughly 80% of Nexus's infrastructure is common across verticals; the remaining work is mapping each new company's workflows and data into a model-agnostic layer. Early acquisitions took over a year to show results; now the platform delivers measurable time savings within days of closing. Crucially, Long Lake does not run the classic private-equity playbook of cutting costs and flipping the business - portfolio companies have gone from 0-5% organic growth to 20%+ because AI-driven efficiency turns growth from a high-marginal-cost activity (keeping only about 20 cents of each incremental revenue dollar after hiring costs) into a high-margin, software-like one.

That productivity gain also drives a retention flywheel: employees made significantly more productive get paid more and are reluctant to leave, since doing so means giving up Nexus's tooling and returning to manual work. Building this required combining three disciplines Taubman says rarely coexist - private-equity deal execution, applied AI engineering, and on-the-ground change management. Long Lake's first 20 hires came entirely through founders' personal networks, pulling engineers from companies like Palantir, Ramp, Robinhood, and Glean, and M&A talent from GTCR, Blackstone, TPG, and HIG specifically among PE professionals who believe in the AI thesis but work at firms that aren't AI-native. Engineers are embedded on-site with each portfolio company rather than centralized, mirroring a "skunk works" co-location approach to shorten feedback loops.

On strategy, Taubman repeatedly contrasts Long Lake's approach with typical PE: he explicitly models the firm on Danaher and TransDigm, aiming to compound operating advantages over decades as a permanent owner rather than exit after a few years, and expects Long Lake's cost of capital to improve as its track record proves out. This positioning, plus equity rollover for founders and management, is presented as why Long Lake increasingly wins competitive processes or becomes the only buyer sellers want. The Amex GBT deal is discussed within this frame: a 111-year-old company (with a 1876-founded subsidiary, Carlson Wagonlit) chosen partly because Long Lake targets mission-critical, high-cost-of-failure services where trust built over a century has outsized value, and where giving travel counselors "AI superpowers" fits the existing Nexus playbook.

Taubman closes by sizing the opportunity: he estimates real enterprise AI use cases are only about 1% penetrated, with most of the US economy made up of small businesses lacking resources to build AI tooling themselves, and even large companies struggling to extract value from AI - the gap Long Lake is positioning itself to fill as a long-term operator rather than a software vendor.

Notable Quotes

"We actually think AI makes people more productive and we have more productive people... When your customers are happier, you grow faster... We create jobs and everybody wins." - Alexander Taubman

"If you now leave Long Lake or you leave one of our partner companies to go to a competitor, you have to start doing all this mundane work again... it's like giving up email." - Alexander Taubman

"AI is very, very underpenetrated. It's probably around 1% penetrated in terms of real enterprise AI use cases." - Alexander Taubman

"I'd want to own that company forever and compound on that advantage for decades to come, and then extrapolate it into ancillary areas and other service lines." - Alexander Taubman