What happened
Ollama, the open source tool for running open-weight AI models on your own machine, has closed a $65 million Series B. Theory Ventures led the round, and CEO Jeff Morgan confirmed it to TechCrunch.
The company's earlier $15 million Series A was led by Benchmark's Peter Fenton. Add it all up and Ollama has raised $88 million in total.
The tool launched in 2023 and is built to get developers up and running with local models in minutes. Its GitHub repo has 176,000 stars and nearly 17,000 forks, and the user count sits at nearly 9 million.
There is also a paid side. Ollama hosts larger, heavier models on its own neocloud, reachable through subscription tiers that run from free up to $100 a month. Usage is tracked by GPU time rather than token limits.
Morgan and co-founder Michael Chiang have done this before. They helped build Docker Desktop after Docker bought their earlier startup, Kitematic.
Why it matters
Local AI is no longer a hobby for people who enjoy fan noise. Nearly 9 million users is a real audience, and investors have now put real money behind the idea that developers want to run models themselves.
The business model is the interesting part. Free local use gets people in the door, and hosted models cover the cases where a laptop simply gives up. Billing by GPU time instead of tokens is a deliberate contrast with how most hosted AI services count.
The founders' Docker history also tells you the pitch. Last time, the job was making cloud apps easy to move around. This time, it is making models easy to run on the machine already on your desk.
Local models just got a $65 million vote of confidence.
One caveat: a funding round is a bet, not a result. Ollama has the users and the stars, but the paid tiers still have to carry the company.
What to do with it
- Try a local model.Install Ollama and run an open-weight model on your own machine to see what your hardware can handle.
- Check your data rules.If prompts contain anything sensitive, local inference keeps them on your device, which is worth raising with your team.
- Compare the pricing.GPU-time billing behaves differently from token billing, so run your real workload through both before committing.
- Watch the free tier.It is generous today, and a freshly funded company is exactly when plans start to change.
Ciao, sapiens. Your laptop has been promoted, and it did not even ask for a raise.


