Applied Compute Raises $350M at $3.25B Valuation After $100M Run Rate

Applied Compute, a San Francisco startup that builds custom AI models for large companies, has raised $350 million at a $3.25 billion valuation. The round was led by investor Elad Gil and announced in early November 2026. The company says it reached a $100 million annualized revenue run rate less than 18 months after it was founded. That makes it one of the fastest-growing enterprise AI startups, and the round is a bet that companies want AI models trained on their own data rather than off-the-shelf chatbots.

What happened?

Applied Compute closed a $350 million funding round at a $3.25 billion valuation. Elad Gil led it, joined by existing backers Kleiner Perkins and Lux Capital and new investors including NVIDIA, Blackstone, Shopify Ventures and HubSpot Ventures.

The deal was first reported in September, when Forbes said Applied Compute was in talks to raise $350 million at that valuation. The company has now confirmed it alongside its revenue milestone.

How fast is Applied Compute growing?

The company’s growth figures are striking:

• November 2025: about $13 million in annualized revenue.
• August 2026: about $50 million, as previously reported.
• November 2026: a $100 million annualized run rate, according to the company.

Its valuation has climbed just as fast. In April 2026, Applied Compute raised $80 million at a $1.3 billion post-money valuation in a round led by Kleiner Perkins, bringing its total funding to $160 million at the time. The new round values it at about two and a half times that figure seven months later.

Applied Compute says it has more than doubled the number of enterprises it serves since its last raise. Its team has tripled since January, and 85% of staff are researchers and engineers.

What does Applied Compute do?

Applied Compute helps companies build what it calls “specific intelligence”: AI models trained on a company’s own data and standards. Its engineers work directly with enterprise customers to train models, often starting from open-source models, and then run them in production.

About three months ago it launched AC2, the Applied Compute Agent Cloud, which brings model training, inference (running the model) and continual improvement into one platform. The company says:

• More than 1,700 training runs have taken place on AC2 since July.
• Its inference service handles more than one trillion tokens a day.
• 40% of customers use both its training and inference products.

Customers include Microsoft, Shopify, Mercor, Harvey, DoorDash, Cognition, Clay and Decagon. Applied Compute says a project for legal AI company Harvey produced better answers and citations at less than half the running cost of a leading frontier model. It also says a custom model router for Clay cut production AI costs by up to 47%.

Who founded Applied Compute?

The company was founded by three former OpenAI researchers: chief executive Yash Patil, Rhythm Garg and Linden Li. They left OpenAI to start a company focused on enterprise AI built on open models. Elad Gil, who invested in earlier rounds, told Forbes in September that he had always believed open source would play a major role in AI.

Why it matters

Applied Compute’s rise reflects a shift in how big companies want to use AI. Instead of paying per query for a general model, many want models tuned to their own workflows that are cheaper to run and that they control. Cost is a big part of the pitch: the company’s case studies focus on matching frontier-model quality at a fraction of the inference bill.

The round also shows that investors are still writing very large cheques for AI companies with fast revenue growth, even as public tech stocks wobble. That trend runs through recent deals such as fal’s $1.3 billion Series F. Established software companies are also buying their way into custom AI, as seen in Five9’s $325 million deal for Level AI.

What happens next?

Applied Compute says it will use the money to expand its computing capacity, keep growing its team and speed up deployments with enterprise customers. The test will be whether it can keep doubling revenue while competing with AI platforms and cloud providers that also offer tools to fine-tune and host open-source models.