Ollama Raises $65 Million Series B, Reaches 8.9 Million Monthly Developer Users

$88 Million Total, 14 Employees

Ollama, the open source tool that lets developers run AI models directly on their PCs, has raised a $65 million Series B led by Theory Ventures. CEO and co-founder Jeff Morgan confirmed the round to TechCrunch. It follows a $15 million Series A led by Benchmark partner Peter Fenton, bringing total funding to $88 million.

The company has 8.9 million monthly active developers, according to Morgan. It has 176,000 stars and nearly 17,000 forks on GitHub. It is embedded in 85% of Fortune 500 companies. All of that with a team of exactly 14 people.

What Ollama Actually Does

Ollama launched in 2023 to solve a specific, boring problem: getting open-weight AI models to actually run on a developer's local machine without hours of configuration headaches.

"Open models started coming out in 2023 but they were really hard to use," Morgan told TechCrunch. They were designed for researchers, not programmers. Ollama abstracted away the hardware configuration layer, the same basic trick its founders pulled off before with Docker.

Morgan and co-founder Michael Chiang previously built Docker Desktop after Docker acquired their earlier startup, Kitematic. Docker solved the cloud portability problem for containerized apps. Ollama is attempting to do the same for local AI inference.

Developers can run smaller models on their own hardware for free, or access larger hosted models through Ollama's neocloud at subscription tiers up to $100 per month. Pricing tracks GPU time, not token limits, a meaningful distinction for developers building inference-heavy applications.

Why Benchmark Wrote the First Check

Fenton, who led the Series A and joined Ollama's board, pointed directly to the founders' track record when explaining his conviction.

"What Jeff and Michael built with Docker is being used by 10 million-plus developers every day," Fenton told TechCrunch. "The creative powers to create a product that goes to ubiquity for developers is extremely rare."

Fenton's bet here was rooted in a demonstrated pattern: two founders who had already scaled a developer tool to eight-figure daily active users, starting over in a new category.

Morgan and Fenton declined to disclose current revenue or the company's post-money valuation from this round.

The Open vs. Closed Model Debate

The skeptical case against Ollama's long-term business centers on this argument: as frontier closed models from Anthropic, OpenAI, and Google continue improving faster than open-weight alternatives, enterprises may ultimately prefer the performance ceiling of a managed API over the operational overhead of running their own inference. Local-first AI looks attractive on cost, but if the quality gap widens, price sensitivity may not be enough to hold enterprise customers.

Fenton pushes back directly on the either/or framing. "I still think that this is the part that most of the debate gets wrong. It's not an either/or," he told TechCrunch. His view: enterprises will run open models for high-volume, cost-sensitive workloads and lean on closed models for tasks that demand best-in-class capability. Both markets can grow simultaneously.

Enterprise IT does not historically collapse to a single vendor in any infrastructure category, and inference cost pressure is real.

The January Inflection Point

Morgan credits a specific moment as the commercial validation for Ollama's model: around January, when larger open-weight models became capable enough to handle agentic coding tasks reliably.

"Obviously, we saw the explosion of the assistants like OpenClaw [sic], and this idea that open models can get real work done," Morgan told TechCrunch. That shift, where open models graduated from research toys to production-grade coding tools, changed the enterprise calculus on local inference.

Since then, the conversation in enterprise AI procurement has noticeably shifted toward asking whether a given workload actually needs a closed frontier model, or whether a well-tuned open model running on owned infrastructure is sufficient.

What Comes Next

Morgan and Fenton have not announced how the $65 million will be deployed, and revenue figures remain private. The unresolved question for Ollama is conversion: 8.9 million monthly users is a large number, but the ratio of free users to paying subscribers and the revenue that paid tiers actually generate has not been disclosed. At a $100/month ceiling on its highest subscription tier, the path to venture-scale revenue requires either significantly expanding that pricing ceiling for enterprise contracts, growing the paid user base substantially, or both.