Training and fine-tuning
Platforms and services that adapt models to a company's own data and tasks.
Who is making money
The companies earning or raising real money in this part of AI. Suppliers are marked: they sell to the builders rather than to end users.
Sells expert human feedback and training scenarios to labs that tune their models.
About $2B gross annualised run rate in June 2026, reported by The Information. Gross means before contractor pay, which Sacra says is 60 to 70 percent of it
Open profileGPU cloud that bought Weights and Biases and OpenPipe to add model tracking and fine-tuning tools.
Quarterly revenue $2.6B, up 112%. $104B backlog. Still loss-making
Open profileProvides human feedback and evaluation for model tuning, now with a large Meta stake.
Meta paid $14.3B for 49%
Open profileSupplies human feedback and training scenarios to frontier labs for post-training.
About $1.2B in 2024 (Sacra estimate)
Open profilemanaged fine-tuning across many open models
Annualised run rate above $100M, self-reported and unaudited
Open profilefine-tuning plus hosting
About $1B annualised (Sacra estimate, Feb 2026)
Open profilepay-as-you-go GPUs for training jobs
About $300M annualised (CEO, 2026, reported)
Open profileDeveloper platform that makes it easier to fine-tune and deploy open language models.
Reported sale price of $100M to $500M to Rubrik in June 2025 (reported, not official)
Open profileWhere the money flows
Companies and labs pay per job or per GPU hour. The platforms pay GPU clouds.
- AnthropicpaysMercorMercor's first-half 2026 gross revenue about $614M, with OpenAI and Anthropic named as main buyers
Buys expert training and feedback work through Mercor (reported)
aiweekly.co
The openings
Gaps still open in this area, from ones a small business could take to ones only giants can play.
- Small business Done-for-you fine-tuning and testing projects for mid-size firms
- Startup money Training scenarios for specific business workflows, sold to labs
- Giants only A general post-training platform
Frontier models keep getting good enough that good instructions and company data often beat fine-tuning.
Sources: dealroom.co, sacra.com
The map updates three times a week.
241 companies across 25 areas, 62 money flows and 89 openings so far. New research lands every Tuesday, Thursday and Saturday.
Back to the full mapFree daily email
Get the map updates in your inbox
A short read every morning on what is moving in AI, including new areas of the map as they go live. Free.
You are in
Your first AI Pulse lands tomorrow morning. Keep an eye on your inbox.