Inside the World's First $6.3B AI Take-Private: Long Lake's Plan to Transform 111-Year-Old Companies
On the latest episode of No Priors, Alex Taubman — co-founder and CEO of Long Lake Management — laid out a vision that most of Silicon Valley hasn’t registered yet: buy legacy service companies outright, inject a shared AI platform, and watch them grow like software companies.
It’s not a pitch deck. It’s happening. Long Lake just announced a $6.3 billion acquisition of American Express Global Business Travel — a company founded in 1915 — in what appears to be the world’s first AI take-private. Before this, Long Lake had already acquired roughly 30 companies across industries like home services, HR outsourcing, and specialty tax.
The Model: AI Rollup, Not AI Software
The traditional Silicon Valley playbook for any industry is: build software, sell it as a vendor. Taubman’s counterintuitive thesis: owning the company is better than selling to it.
His argument proceeds in three parts:
1. The Alignment Problem (for SaaS)
When you’re a software vendor, you don’t control the business outcomes. You can sell AI-powered scheduling software to a home services company, but you can’t force the company to reorganize its workflows around it. You can’t retrain the dispatchers. You can’t convince the CEO to stop doing things the way they’ve always been done.
When you own the company, all of that goes away. You have the authority to redesign processes from the ground up. The feedback loop between the engineering team and the operating team goes from “quarterly check-in with a client” to “daily co-located iteration.”
2. The Talent Flywheel
Taubman claims Long Lake’s portfolio companies have extraordinarily high employee retention. The mechanism is simple: once you’ve used Nexus (Long Lake’s AI platform) to eliminate ~25-30% of the mundane work from someone’s day, leaving Long Lake for a competitor means going back to doing all that work manually.
He compares it to “giving up email” — once you’ve experienced the productivity, you won’t go back. Because employees are more productive, Long Lake can pay them more than competitors. Because they’re better paid and better equipped, they stay. Because they stay, institutional knowledge compounds. This is a textbook positive-sum flywheel.
3. The Growth Paradox
Here’s the non-obvious insight: most legacy service companies want to grow, but can’t profitably. If a company grows 20%, they need ~20% more people. Hiring, training, and managing those people eats most of the incremental margin — Taubman calls this a “very high marginal tax rate on growth.”
When AI makes existing teams 30-40% more productive, the economics flip: now you can grow without proportional headcount increase. The math starts looking like a software company — high incremental margins on every new dollar of revenue. This, more than cost-cutting, is the engine of the model.
The Nexus Platform: 80% Shared, 20% Custom
Long Lake’s secret weapon is Nexus, a horizontal AI platform that handles the common infrastructure across all their portfolio companies. Taubman claims ~80% of the code is shared, with the remaining 20% being industry-specific customization.
This mirrors the Danaher Business System — the legendary manufacturing operating model that allowed Danaher to compound at 20%+ for decades. The difference: Danaher optimized for manufacturing efficiency. Long Lake optimizes for AI-driven service delivery.
Key capabilities:
- Model-agnostic interface: Nexus sits between the models and the business, allowing Long Lake to swap models as capabilities improve
- Workflow mapping: Engineers sit with frontline workers, understand pain points, then build tools within Nexus to solve them
- Data integration: Clean up and connect data sources to make them accessible to AI models
- Deployment velocity: New acquisitions can go from zero to AI-enabled in “days” rather than the year-plus it took with the first acquisition
The Amex GBT Deal: Why This Company?
American Express Global Business Travel isn’t a random target. Taubman describes Long Lake’s approach as “prepared mind” — they maintain a whiteboard of 15-20 industries they believe are high-value AI targets, and travel was always on the list.
The thesis: business travel is mission-critical, high cost-of-failure, and most trips are revenue-generating. If you miss a flight to a client meeting, it’s not just an inconvenience — it’s lost revenue. The incumbent (Amex GBT) has built a century of trust around reliability. Long Lake’s play is to layer AI superpowers on top of that existing trust infrastructure.
The vision: “a travel counselor with AI superpowers” — real-time disruption resolution, predictive rebooking, personalized recommendations based on traveler history. The outcome Taubman targets is not cost reduction, but faster growth driven by better customer experience.
The Danaher of AI?
When asked about the 10-20 year vision, Taubman explicitly invokes Danaher and Berkshire Hathaway: long-term holding companies that compound value through operational excellence. He’s positioning Long Lake not as a private equity flip operation, but as a permanent home for great service businesses.
This matters because it changes the competitive dynamic:
- Sellers: Business owners who’ve spent 40 years building a company want a steward, not a cost-cutter. Long Lake’s permanent capital + AI platform makes them the preferred acquirer.
- Employees: The promise of AI superpowers + higher pay + better growth is a talent magnet in industries where competitors offer none of these.
- Competitors: Traditional PE firms can’t replicate the AI platform. Tech companies can’t replicate the M&A and change management capability.
Risks and Open Questions
1. The Change Management Challenge at Scale
Taubman talks about engineers “living in the office for two years” to drive transformation. This is a people-intensive model. As the portfolio scales to dozens of companies, where do these engineers come from? The applied-AI talent market for people who can both code and run change management is extraordinarily thin.
2. The Model Dependency
Nexus is model-agnostic, which is smart. But it still depends on frontier models continuing to improve at their current trajectory. If progress stalls, the “30-40% productivity gain” thesis looks different.
3. The Vertical Transfer Problem
80% shared infrastructure sounds great, but Taubman acknowledges the remaining 20% is significant work. Going from home services to HR to tax to travel — each requires deep domain expertise. The coordination overhead of managing these parallel transformations is non-trivial.
4. Capital Structure Risk
The Amex GBT deal is $6.3 billion. If the AI transformation takes longer than expected, or if interest rates move against the financing, the leverage could constrain Long Lake’s ability to invest in growth.
The Bigger Picture
Long Lake represents something genuinely new: a synthesis of private equity’s acquisition discipline, Silicon Valley’s engineering talent, and frontier AI’s productivity gains. It’s not a tech company buying tech companies. It’s not a PE fund running Excel models. It’s a hybrid organism purpose-built for the moment when AI’s labor economics become undeniable.
If Taubman is right — if AI can systematically turn 0-5% growth legacy service companies into 20%+ growth compounders — Long Lake won’t just be a successful investment firm. It will be a proof of concept that AI’s greatest economic impact isn’t in software, but in every industry that software has failed to transform.