ChipAgents, a startup building AI agents that design and test semiconductors, has raised $60 million. The company, a partner of Nvidia, is betting that the same agentic AI reshaping software can be turned on the painstaking business of making chips.
The raise adds to an already busy year. ChipAgents closed a $21 million Series A last October and a $50 million round in February, so the latest cheque takes its total funding to roughly $134 million in under a year, a pace that tracks the wider rush into AI-native chip-design tools.
“We’re enabling the industry to move beyond AI assistants toward autonomous agents that can execute meaningful engineering work,” said William Wang, CEO and Founder, ChipAgents.
The company’s product sits in a corner of the industry known as EDA. Electronic design automation is the software engineers use to lay out and verify chips, a market long dominated by a handful of incumbents, and the one ChipAgents wants to rebuild around autonomous agents rather than human-driven tools.
Verification is the bottleneck it is aiming at. Checking that a chip actually does what it is supposed to can eat more than 70% of a design project’s time, and ChipAgents says its agents can read a specification, write the tests, run them, and iterate without an engineer holding their hand at each step.
The claims are bold, as they tend to be. The company says its agents comprehend specifications many times faster than people and generate formal checks almost instantly, and that the platform is already in production at several chipmakers, though those figures come from the company rather than any independent audit.
Its founder frames it as a change of model. William Wang, ChipAgents’ chief executive, describes agents that “read specifications, break down objectives, discuss and implement solutions, validate results and iterate relentlessly,” a pitch for software that does the job rather than assists the person doing it.
The company is the product of Alpha Design AI, based in San Jose with research roots near Santa Barbara. Wang, an academic turned founder, has leaned on that pedigree to recruit both talent and early customers.
The investor base has been heavy on strategics. Earlier rounds drew in Bessemer Venture Partners alongside chip-world names including Micron, MediaTek, and Ericsson, the kind of backers whose interest doubles as a signal that the technology is being taken seriously inside the industry.
The Nvidia label is the eye-catching part. Reuters cast ChipAgents as an Nvidia partner, a useful association in a market where Nvidia’s chips and software sit at the centre of the AI boom, even if the exact terms of the relationship were not spelled out.
It is not alone in the idea. A wave of startups is pitching AI at different layers of the silicon stack, from Oxmiq’s chip architecture to Kandou’s interconnects, and agentic design tools are the newest front in that competition.
The logic is almost recursive. AI needs ever more chips, chips are slow and costly to design, and so the industry is reaching for AI to design the chips that will run more AI, a loop that investors have found hard to resist.
The incumbents are not standing still, which is the risk. The established EDA vendors have their own AI features and deep ties to every major chipmaker, and a startup promising to replace their tools has to prove its agents are reliable enough to trust with silicon that costs millions to fabricate.
That trust is the real product. For all the talk of speed, what ChipAgents has to sell is the confidence that an agent’s verification can be believed, and $60 million more is the market’s wager that it can get there.
The broader bet, shared across a field of AI-agent startups, is that agents will soon do skilled technical work, not merely talk about it.
Get the TNW newsletter #
Get the most important tech news in your inbox each week.