BUYER UNIVERSEUpdated October 21, 2026
Short answer
Three groups buy AI companies. Strategic acquirers, including hyperscalers, chip makers and software, security and payments platforms, pay the most for scarce teams, data and models. Private equity buys applied-AI software with real revenue, and growth funds back model and infrastructure leaders. Prices run from 3x to 6x revenue for enterprise AI applications to 12x to 20x for LLM and generative-AI assets. Venture-backed AI companies made 195 acquisitions of AI startups through September 29, 2026, according to Crunchbase.
THE THREE BUYERS
AI is the most mispriced category in the market, so the buyer you run toward matters more here than almost anywhere else. The same business can be worth a strategic premium to one acquirer and a mature-software multiple to another. Three distinct pools compete, each pricing your company on a different basis: capability and scarcity, return on revenue, or category leadership. Knowing who buys what, and why, is the difference between a clean exit and leaving real money on the table.
WHAT EACH BUYER PAYS
Strategic acquirers. Hyperscalers and cloud platforms, large enterprise-software vendors, and security, data, and chip incumbents buying a capability, a dataset, or a team to defend or extend their position. They pay the most for genuinely scarce technology or talent, and at the model layer they often acquire the team and intellectual property rather than a revenue line.
Private equity platforms buy applied-AI vertical software that already has real revenue and retention, underwriting to a return and paying for proven unit economics and a defensible workflow position, not a research narrative. Growth and crossover sponsors fund model, data, and infrastructure leaders through their scale-up phase and set up a later strategic sale, paying for category leadership and the durability of the advantage.
HOW TO POSITION
Running the wrong story to the wrong buyer is how a 12x business receives a 4x offer.
HOW THESE DEALS ARE STRUCTURED
AI deal structure varies more by layer than almost any other sector. At the model and research layer, deals are often talent-and-IP acquisitions, where much of the consideration is tied to retaining the team through vesting and milestones rather than paid as cash at close. At the application layer, structure looks like the rest of software: meaningful cash at close, frequently with rollover equity when a sponsor is the buyer, an escrow against reps and warranties, and sometimes an earnout tied to retention or growth.
WHAT BUYERS EXAMINE
MARKET CONTEXT
AI M&A is being driven by a capability race at the top of the market and a proving-out of economics at the bottom. Hyperscalers and large platforms are paying extraordinary prices, often through talent-and-IP structures, to secure scarce model and research teams. One layer down, the question has shifted from narrative to numbers: which applied-AI businesses actually retain customers and earn a margin after inference cost. Capital has moved toward the assets where the unit economics prove out.
The buyers most active in AI M&A, by layer of the stack:
| Infrastructure & data platforms | Nvidia, Microsoft, Google, Amazon, Databricks, Snowflake | Compute, data, model hosting, and ML infrastructure |
| Application & enterprise strategics | Salesforce, ServiceNow, Adobe, SAP, Intuit | Applied AI embedded into existing products and workflows |
| Private equity & growth | Thoma Bravo, Vista Equity, General Catalyst, growth and crossover funds | Applied and vertical AI with real recurring revenue and defensible economics |
| Hyperscaler & foundation buyers | Microsoft, Google, Amazon, Meta, OpenAI, Anthropic | Teams, data, and model capability (often acqui-hire structures) |
Identifying and positioning to the right acquirer in your sub-sector is core sell-side work. Request a confidential valuation to see which of these buyers is active on a company like yours.
FAQ
Three groups. Strategic acquirers such as hyperscalers, enterprise-software and security platforms, and data and chip incumbents buying capability, data, or a team; private equity platforms buying applied-AI vertical software with real revenue; and growth and crossover sponsors funding model and infrastructure leaders toward a later strategic exit.
Foundation-model and LLM companies are bought mostly by hyperscalers and large platforms, often as talent-and-IP deals, and price at 12 to 20x revenue or far higher at the earliest stages. Enterprise AI application companies are bought by software strategics and PE platforms at 3 to 6x, closer to mature SaaS once growth is removed.
It splits sharply by layer. LLM and generative-AI assets with defensible data or inference advantages clear 12 to 20x revenue. Data and MLOps platforms run 6 to 9x, infrastructure-adjacent software 6 to 10x, and enterprise AI applications 3 to 6x.
At the model and research layer, buyers are often acquiring scarce engineering teams and intellectual property rather than a revenue stream. These deals price on the team, the technology, and strategic scarcity, which is why headline multiples can look detached from current revenue.
Model-layer deals often tie much of the consideration to retaining the team through vesting and milestones. Application-layer deals look like software: cash at close, often rollover equity with a sponsor, an escrow, and sometimes an earnout. The cash you actually realize depends heavily on how much is contingent.
Through a confidential, targeted process. An advisor identifies the specific strategics and sponsors whose roadmap or thesis matches your layer of the stack, approaches them under NDA, and runs them in parallel so you get genuine competition without signaling to competitors or customers.
Among venture-backed AI companies, OpenAI was the most active, with 10 acquisitions in 2026 through September 29. Anthropic and Legora announced five each, Harvey four, and Sierra and Cursor three each. Crunchbase counted 195 acquisitions of AI startups by venture-backed AI companies over that period, 14% more than in all of 2025, while the number of buyers grew only 2%. Prices were disclosed for just 12 of the 195 (Crunchbase News, October 6, 2026).
Nvidia agreed on September 3, 2026 to buy Hugging Face for about $12.9 billion, including up to $1 billion of retention equity for employees joining Nvidia, with closing expected in the first half of 2027 (Adweek). Stripe agreed in August to buy OpenRouter, an AI model routing platform, in a deal reported at about $7.5 billion. Visa agreed on August 3 to buy BioCatch, which uses AI to detect fraud, for $2.4 billion in cash (Bloomberg Law).
An acqui-hire is a purchase made mainly for the team and its intellectual property rather than for revenue. Crunchbase described OpenAI's February 2026 deal for OpenClaw, which brought in the open-source AI agent and its creator, Peter Steinberger, as an acqui-hire. In these deals much of the value is tied to keeping people through vesting and milestones. Full acquisitions carry retention terms too; up to $1 billion of Nvidia's Hugging Face agreement is retention equity for employees.
Windsor Drake advises founder-led companies on sell-side M&A. A first conversation is confidential and starts with your numbers.
Research, not investment advice, a fairness opinion or a valuation of any specific company. Third-party data is cited to its source and remains its owner’s. Windsor Drake is not a registered broker-dealer and does not offer, sell or place securities. When a transaction requires a registered broker-dealer, that work is handled by a registered broker-dealer partner.
