# Long Lake AI Rollup: How Alex Taubman's AI-Native Holding Company Works

Source: https://trymimetic.com/guides/long-lake-ai-rollup

[Guides](/guides) / Revenue QA

# Long Lake: how the first AI-native holding company works

Long Lake buys established services businesses and runs them on its own AI platform. It was founded in 2023, has made 40 acquisitions, and in 2026 took Amex GBT private in a deal valued at about $6.3 billion. General Catalyst, which led its first investment, calls it **"the first expression of the 'AI roll-up' as a concept"** ([General Catalyst](https://www.generalcatalyst.com/stories/manufacturing-outliers-long-lake), 2026). Its CEO, Alex Taubman, calls it **"the first AI-native holding company"** ([The Next Century of Great Companies](https://llmh.com/blog/next-century-of-great-companies), 2026).

This is a case study built from what Long Lake, its investors and its SEC filings have said publicly, and what is still unproven.

## Long Lake at a glance

- **Founded:** 2023, by Alex Taubman (CEO, formerly a managing director at Oaktree) and Rasmus Visman (CTO). Offices in New York and San Francisco. The company's site is llmh.com.
- **Backers:** General Catalyst, whose Creation strategy led the first investment in February 2024, plus Alpha Wave Global, D1 Capital, Thrive Capital and Elad Gil.
- **Capital:** about $1.3 billion of equity across four rounds, according to Dealroom. The company has not confirmed the total.
- **Acquisitions:** 40, with Amex GBT the 40th. It is the only one Long Lake has named publicly.
- **Scale:** **"nearly 30,000 people"** across its companies, and five service lines: HOA management, HR and payroll services, specialty tax, architecture and engineering, and travel.

## What it buys

Long Lake started in HOA management. Taubman says it has bought "20 something" companies in that industry alone (Founders with David Senra, 2026). The pattern across its service lines is the same: fragmented industries, recurring customer relationships, and a lot of work done in email and documents.

Taubman says the company sees four to five thousand potential deals a year and enters only one or two new industries a year: **"big funnel, very tight filter"** (Founders with David Senra, 2026). He describes buying established, profitable businesses rather than turnarounds, and the company's own description of what lasts is **"deep customer relationships, trusted brands, and strong supplier partnerships."**

## How it works: Nexus and the people inside the companies

Long Lake's platform is called Nexus. Taubman: **"Nexus is built on frontier models, so every advance in AI makes it more capable, and every Long Lake company benefits as soon as it does."** He says about 70 to 80% of the infrastructure is shared across industries, so each new industry adds capabilities the others can use.

The work is done inside the acquired companies. From Long Lake's site: **"Our engineers and operators work inside our companies, codesigning systems"** ([Long Lake](https://llmh.com)). Taubman describes Long Lake's own team as small and mostly AI engineers and deployment staff, with a minority working on acquisitions.

Speed has changed. Taubman says it took **"over a year"** with the first acquisitions to see the real potential, **"But now, within days of partnering with a company, we can deploy this very quickly"** (No Priors, 2026).

## The thesis: AI as a growth lever

The central claim is that AI makes services firms grow faster.

- **"we're not focused on cost saving. We're actually focused on driving growth and customer experience"** (Taubman, No Priors, 2026).
- Taubman says **"we have not conducted a single layoff across our portfolio"** and that **"we've grown both headcount and topline, together"** (his essay).
- His economics argument: a labor-heavy services firm keeps only a small share of each new dollar of revenue, so growth is expensive. Make the team substantially more efficient and the same firm's growth starts to look like a software company's (No Priors, 2026).

The operating claims published so far, all company-reported:

- HOA businesses that were **"typically growing 0 to 5% a year in terms of volume"** are **"now growing 20 plus percent a year"** (Taubman, No Priors, 2026).
- **"Top-performing HOA managers are reclaiming over 65 hours per month,"** a **"30%+ capacity increase for payroll processors,"** and an **"AI-powered sales engine has grown the new customer pipeline tenfold"** (General Catalyst, 2026).
- Taubman told the Founders podcast the company expects to do over $4 billion of revenue next year, and that because it buys businesses that already make 20 to 25%-plus EBITDA margins, it does not expect to need to raise capital again. Both are forward-looking statements.

## Permanent capital

**"Long Lake offers permanent capital: our plan is to buy, hold, and improve forever."** That separates it from a private equity fund with a five-to-seven-year exit. Taubman cites Danaher and TransDigm as models, and argues that a platform able to run businesses better can afford to pay sellers more, so it does not lose deals on price.

## The Amex GBT deal

Long Lake agreed to acquire Amex GBT on May 4, 2026, at $9.50 a share, about $6.3 billion all-cash, and closed on September 29, 2026. The SEC merger proxy shows the financing: up to $3.139 billion of Long Lake equity, up to $1.25 billion from Koch Equity Development, and debt facilities of $1.5 billion in term loans, a $250 million revolver and a $1.0 billion bridge ([SEC DEFM14A](https://www.sec.gov/Archives/edgar/data/1820872/000114036126027649/ny20072693x2_defm14a.htm), 2026).

The same proxy shows that another bidder told the company it would only continue if it was **"partnered with a financial partner with an artificial intelligence angle."** An AI operating story was a factor in the sale of a public company.

Taubman at signing: **"The future of business travel will be defined by AI and human agents working seamlessly together"** ([Amex GBT](https://www.amexglobalbusinesstravel.com/press-releases/long-lake-agrees-to-acquire-amex-gbt-with-support-from-general-catalyst-and-alpha-wave/), 2026). Amex GBT kept its name. Taubman describes the brand as one Long Lake wants to **"partner with and let them shine."**

## What is still unproven

- **One named acquisition out of 40.** The other 39 are not public, so the operating claims cannot be checked company by company.
- **The numbers are company-reported.** Growth rates, hours saved and pipeline figures come from Long Lake and its lead investor. None are audited.
- **The return may look like private equity, not venture.** Critics such as Newcomer have argued that PE-like strategies may return 2x or 3x rather than 10x. Nathan Benaich and Nikola Mrksic argue that AI rollups **"may still deliver returns, but not the kind VCs are underwriting"** ([Fortune](https://fortune.com/2025/06/27/ai-rollup-investment-strategy/), 2025).
- **The competencies are rare.** Elad Gil, an investor, says the model needs three things at once: private equity dealmaking, AI engineering and change management (No Priors, 2026).
- **Integration at larger scale is new.** Twenty-something HOA firms are a different integration problem from a global travel company with tens of thousands of staff.

## What operators and investors can take from it

1. **Pick industries before companies.** A tight list of industries with shared workflows is what makes a shared platform pay off.
2. **Make the platform the asset.** Each acquisition should make the next one faster to integrate, or the model is a traditional roll-up with software costs.
3. **Measure growth.** If the claim is growth, count inquiries, bookings and new customers in every company from Day 1, the same way each time. In our check of 146 companies bought by AI rollups, 35% showed no Google Analytics, Tag Manager or HubSpot tracking on their homepage, so many start without a baseline ([AI rollups in 2026](/research/ai-rollups-2026)).
4. **Keep each company's name.** Long Lake, ONAR and most of the platforms we track keep each acquired company's name and local identity.

For the broader playbook, see [buy and build with AI](/guides/ai-buy-and-build-strategy) and [services as software](/guides/services-as-software).

Quotes marked No Priors (Elad Gil's podcast, "The World's First AI Take Private") and Founders with David Senra ("Bringing AI to the Real Economy") are transcribed from the episodes.

## How Mimetic helps

Mimetic checks a company's site the way its customers use it, and whether its tracking adds up. For an acquirer running dozens of companies, it gives the same read on each one: before the deal, on Day 1, and after each change.

## Common questions

**What is Long Lake?**
An AI-native holding company founded in 2023 that acquires established services businesses, such as HOA management, payroll, specialty tax, architecture and engineering, and travel, and runs them on its AI platform, Nexus.

**Who is Alex Taubman?**
The co-founder and CEO of Long Lake. He was previously a managing director at Oaktree Capital.

**How many companies has Long Lake acquired?**
40 as of late 2026, according to the company. Amex GBT, acquired for about $6.3 billion, was the 40th and is the only one it has named.

**How much has Long Lake raised?**
About $1.3 billion of equity across four rounds, according to Dealroom. Its backers include General Catalyst, Alpha Wave Global, D1 Capital, Thrive Capital and Elad Gil. The Amex GBT deal was financed separately with Long Lake equity, Koch preferred equity and debt.

**Does Long Lake lay people off?**
The company says it has not conducted a single layoff across its portfolio and has grown headcount. This is a company statement.

**Is Long Lake an AI rollup?**
Yes. General Catalyst calls it the first expression of the AI roll-up as a concept. Long Lake describes itself as an AI-native holding company with permanent capital.

Written by Coleman Hindes, founder, Mimetic.
Published 2026-11-22.

### Running a buy-and-build? Check each company's front door the same way.

[Check a site](/)
