# AI rollups in 2026: 81 platforms, 341 acquisitions, and what the companies they buy look like online | Mimetic

Source: https://trymimetic.com/research/ai-rollups-2026

[Research](/research)

# AI rollups in 2026: 81 platforms, 341 acquisitions, and what they buy

An AI rollup buys established services businesses, usually accounting firms, property managers, IT providers, insurance brokers or agencies, and rebuilds how they work with AI. General Catalyst, which led the first investment in Long Lake, calls it **"the AI-enabled roll-up"** and calls Long Lake **"the first expression of the 'AI roll-up' as a concept"** ([General Catalyst](https://www.generalcatalyst.com/stories/manufacturing-outliers-long-lake), 2026).

We tracked **81 of them**. 42 have an explicit AI thesis and at least one closed acquisition. Together, the 40 that publish a count claim **341 acquisitions**. And when we looked at the websites of 146 companies they bought, the picture was less AI-native than the pitch.

## The market in numbers

- **81 AI rollups** tracked: 42 core (explicit AI thesis, acquisitions announced), 9 early (thesis and funding, no closed deal yet), and 30 borderline.
- **New.** 25 of the 28 with a known founding year started in 2023 to 2026. 2025 was the peak year for launches.
- **About $9.0 billion of disclosed capital** across 25 platforms, median $60 million. Three hold 85% of it: Metropolis ($3.4 billion, mostly debt), Thrive Holdings ("more than $3 billion") and Long Lake (about $1.3 billion of equity).
- **341 acquisitions** claimed by 40 platforms. The median platform has done 3.5. Four (Current, formerly Crete, Arbio, Long Lake and Buena) account for about half.
- **Where they buy:** accounting leads, then property management and HOA, healthcare administration, insurance brokerage, IT and managed services, legal, engineering and marketing agencies.

Interest is rising with the deals. US searches for "AI rollup" went from about 110 a month in November 2025 to about 320 in October 2026, and posts using the phrase on X went from a handful a month in mid-2024 to more than 100 in most months since mid-2025.

## The thesis: growth, not just cost

The pitch from the people building these companies is that AI changes a services firm's growth rate, not only its margins.

- General Catalyst's target is a **"Rule of 60"**: businesses growing at single digits reaching **"10-20% growth"** with margins of **"30-40%"** ([The Future of Services](https://www.generalcatalyst.com/stories/the-future-of-services), 2025). Its line on why services: **"Customers care about outcomes, not products."**
- Alex Taubman, CEO of Long Lake, calls it **"the first AI-native holding company"** and says it has **"grown both headcount and topline, together"** without a single layoff ([The Next Century of Great Companies](https://llmh.com/blog/next-century-of-great-companies), 2026). General Catalyst reports Long Lake's **"AI-powered sales engine has grown the new customer pipeline tenfold."**
- Thrive's Joshua Kushner frames it as direction: **"Historically, technology transformed industries from the outside in."** ([OpenAI](https://openai.com/index/thrive-holdings), 2025). Thrive Holdings says its goal is to **"transform these sectors from the inside out"** ([Thrive Holdings](https://www.thriveholdings.com/thrive-holdings-x-openai)).
- Elad Gil, an investor in Long Lake: **"If you own the asset, you can [transform it] much more rapidly than if you're just selling software as a vendor"** ([TechCrunch](https://techcrunch.com/2025/06/01/early-ai-investor-elad-gil-finds-his-next-big-bet-ai-powered-rollups/), 2025).
- Sequoia's Julien Bek: **"The next $1T company will be a software company masquerading as a services firm"** ([Services: The New Software](https://sequoiacap.com/article/services-the-new-software), 2026). The same idea, outcomes sold instead of tools, was popularized as "service-as-software" by Foundation Capital in April 2024.

## The skeptics

- Nathan Benaich (Air Street) and Nikola Mrksic (PolyAI): **"The AI roll-up thesis is venture capital's attempt to arbitrage the multiple gap between services and software."** And: **"AI roll-ups may still deliver returns, but not the kind VCs are underwriting"** ([Fortune](https://fortune.com/2025/06/27/ai-rollup-investment-strategy/), 2025).
- Pietro Decio of Tower Research Ventures, on why it may not compound: **"integrating AI into acquired business N+1 is not much easier than doing so for business N"** ([Tower Research Ventures](https://tower-research.com/the-ai-services-opportunity-starts-at-zero), 2025).
- Emergence Capital: **"A roll-up's culture is the weighted average of what you bought"** ([Emergence](https://emcap.com/thoughts/roll-up-or-build-why-the-fastest-path-to-a-venture-scale-ains-business-is-the-one-founders-keep-avoiding), 2026).
- And the classic buy-and-build caution: in BCG's study of buy-and-build deals, **"The contribution of multiple expansion to the IRRs of the buy-and-build deals we studied was 15.3 percentage points"** ([BCG](https://bcg.com/publications/2016/private-equity-power-of-buy-build), 2016). Much of the historic return came from buying small and being valued as big, not from operations.

Even the supporters name the risk. General Catalyst's Marc Bhargava: **"we're only right if the AI enablement works"** ([Cognitive Revolution](https://www.cognitiverevolution.ai/the-revolution-in-services-general-catalysts-ai-rollup-strategy-with-marc-bhargava/), 2025).

## What the acquired companies look like online

If the case is growth, a lot of it has to arrive through each acquired firm's front door: its website, its contact forms, its phone number and its tracking. Most of these firms keep their own brand and site after the deal. So we looked at the homepages of **146 companies acquired by AI rollups** that we could confirm.

- **Four in ten show no Google Analytics.** 60% run Google Analytics 4. On 51 of the 146 (35%) we found no Google Analytics, Google Tag Manager or HubSpot tracking at all.
- **Paid acquisition is rare.** 23% carry a Google Ads tag, 12% a Meta pixel, 10% a LinkedIn tag. Many of these firms grow by referral, which means the first growth lever a rollup pulls often starts from no measurement baseline.
- **The front door is basic.** 62% have a contact form or booking option on the homepage and 60% a tap-to-call link. Only 14% offer online booking, and 6% a chat widget. For businesses being rebuilt around AI, the customer-facing side has barely changed yet.
- **Tracking gets layered.** 9% send data to two or more separate GA4 properties, a common sign of a parent adding its own tag on top of the firm's.
- **Consent is uneven.** 29% show a cookie consent banner. On 26 of those 43, Google or Meta tags were already running before the visitor made a choice.
- **Some sites stop being sites.** 12 of the 146 now redirect to another domain, usually the parent's.

## What this means for buyers and operators

- **Underwrite the front door before the LOI.** A big funnel needs a cheap first filter. Whether a target's site captures leads and whether its numbers can be trusted can be checked from the outside, before anyone opens a data room. See [ecommerce due diligence](/guides/ecommerce-due-diligence).
- **Take a baseline on Day 1.** Most acquired firms start with thin measurement. Without a baseline, "AI grew this firm 20%" cannot be separated from referrals, seasonality or a price change. See [the first 100 days after an acquisition](/guides/post-acquisition-integration-ecommerce).
- **Make the topline levers repeatable.** Lead capture, booking, call tracking and follow-up are the same problems in every firm a platform buys. Fixing them once and rolling the fix out is the services version of the shared platform every AI rollup describes.

## How to read the numbers

- **81 rollups** is a floor, not a census. It counts platforms we could confirm from company sites, press releases, filings and investor posts. Public trackers list more names, most without confirmed acquisitions.
- **Capital and acquisition counts are company-reported.** Funding totals mix equity, debt and committed capital. Acquisition counts may include tuck-ins by member firms. Long Lake has named only one of its 40 acquisitions publicly.
- **146 websites** are the acquired companies with a confirmed domain whose homepage loaded in November 2026. Figures describe homepages only. A firm can track or capture leads elsewhere, so every share is a minimum.
- We do not name individual acquired companies.

Operators and investors can request a company-by-company version of this check for their own portfolio at [trymimetic.com](/).

## Common questions

**What is an AI rollup?**
A company that acquires established businesses, usually in services such as accounting, property management or IT, and transforms how they operate with AI. General Catalyst calls the model the "AI-enabled roll-up"; Long Lake calls itself an "AI-native holding company."

**How many AI rollups are there?**
We tracked 81 as of November 2026, of which 42 have an explicit AI thesis and at least one announced acquisition. Most were founded between 2023 and 2026.

**How much capital has gone into AI rollups?**
About $9.0 billion is disclosed across 25 platforms, but it is concentrated: Metropolis, Thrive Holdings and Long Lake account for 85%.

**What industries do AI rollups buy?**
Accounting is the most common, followed by property management, healthcare administration, insurance brokerage, IT and managed services, legal, engineering and marketing agencies.

**How is an AI rollup different from a traditional roll-up?**
A traditional roll-up creates value mostly through buying small firms at low multiples and combining them. An AI rollup claims to also change how each firm operates, with a shared AI platform, aiming for faster growth and higher margins. Critics argue the returns may still come mostly from the multiple.

**Do AI rollups lay people off?**
Several say no. Long Lake says it has not conducted a single layoff and has grown headcount, and General Catalyst describes its rollups as "not a story about replacing workers." These are company statements.

Mimetic. Public web data, November 2026. Counts are minimums.
Published 2026-11-22.

### See where your own site loses revenue.

[Scan my site](/)
