Two Stanford Dropouts Raise $5.2 Million to Let AI Agents Design Computer Chips, With Jeff Dean Investing

A young San Francisco startup is betting that AI can take over one of the slowest, most expensive jobs in tech: designing the chips that AI itself runs on.

The deal

Phinity Labs announced on Oct. 8 that it has raised $5.2 million in seed funding led by Uncork Capital, with Moxxie and angel investors including Google's Jeff Dean taking part. The round follows a pre-seed led by Pear through its PearX program, the company said in a blog post.

What it is building

Designing a new chip today typically takes 9 to 24 months and hundreds of millions of dollars, with separate teams handling design, verification and physical layout, according to Phinity. By the time a chip ships, the workloads it was built for may have changed. Phinity is building software agents that propose chip designs, test them, learn from failures and revise them, with the long-term goal of going from "prompt to silicon." Dealroom reported that the company is targeting production-ready design files in weeks rather than months or years by 2028.

The company claims its system produces verified designs an order of magnitude faster, and with better power, performance and area, than experienced engineers working from the same specs with frontier coding agents. In one AI inference chip task, Uncork partner Amy Saper wrote, a Phinity agent delivered roughly twice the area improvement of the best human result. Those figures come from the company and its investor and have not been independently verified.

The founders and the money

Co-founders Sonya Jin and Aadi Nashikkar met while studying computer science at Stanford and left to start the company, Saper wrote. Jin previously built synthetic data pipelines for NVIDIA's chip design models and did research at the AWS AI Lab; Nashikkar is a previously exited founder. The team includes a former leader of Intel's custom silicon group and former NVIDIA architects.

Phinity says it already works with some of the largest AI model labs and has reached an eight-figure annualized revenue run rate. Saper called it the fastest revenue ramp of any investment she has made in seven years. The company has not named its customers or disclosed a valuation, Dealroom noted.

Why it matters

AI companies are scrambling for computing power, and custom chips are one way to get more out of it. Phinity says it is hiring for chip design and AI research roles.