Ex-OpenAI Trio in Funding Talks at $500 Million Valuation
The Takeaway
- Applied Compute in talks to raise funding at $500 million valuation
- Startup founded by ex-OpenAI staff developing reinforcement learning services
- Valuation would represent a fivefold surge from three months ago
As artificial intelligence developers increasingly rely on reinforcement learning to improve their models, investors are funding startups that focus on helping businesses utilize the technique to automate more tasks.
In the latest example, Applied Compute, founded in May by three former OpenAI staffers, is in talks to raise new funding at a $500 million valuation just three months after a round that valued it at $100 million, according to people involved in talks.
Lux Capital, an investor in other AI firms such as coding startup Cognition, open-source model hub Hugging Face and video AI firm Runway, is in talks to lead the round, one of these people said. The San Francisco-based company previously raised $20 million in funding from investors including Benchmark, Conviction and Sequoia Capital. The terms of the pending deal could change.
Former OpenAI researchers Rhythm Garg, Yash Patil and Linden Li founded Applied Compute to help software developers and businesses use RL to customize AI for specific fields, such as law or finance, one of the people involved in the round said.
The founders studied at Stanford University before joining OpenAI in 2023 and 2024 to develop its reasoning models, coding AI and deep research tool for ChatGPT. Applied Compute hasn’t discussed its work publicly.
Other companies such as Thinking Machines Lab, co-founded by former OpenAI Chief Technology Officer Mira Murati, have also said privately that they aim to offer reinforcement learning services to businesses. Such services could make it much cheaper for businesses to build AI-powered apps that can complete tasks more reliably in their specific industry, like correcting a legal contract or analyzing financial documents.
Murati’s firm recently raised $2 billion at a $10 billion valuation, not including the capital, after hiring numerous ex-OpenAI staffers.
RL techniques help OpenAI, Anthropic, xAI and other AI developers to improve their models by rewarding the AI for accomplishing certain goals and penalizing it for other behaviors. Such methods have become crucial for the AI labs as researchers last year began finding it harder to improve their models using data scraped from the internet.
RL could make it easier to build models that can automate away jobs in certain fields. One senior executive at OpenAI privately said this year that they expect the “entire economy” to become an “RL machine” of sorts, implying that AI might someday train on recordings of professionals handling day-to-day work on their devices.
A spokesperson from Lux Capital declined to comment. Applied Compute did not immediately respond to a request for comment.