Architect Labs Secures $24M to Build AI Custom Chips

A $24 million seed round has just vaulted a Palo Alto startup out of stealth mode with a bold ambition: replace the slow, expensive, and expert-driven process of chip design with an artificial intelligence system that handles everything from concept to verification. Architect Labs unveiled its plan on Thursday, backed by a roster of investors that includes Kindred Ventures, TQ Ventures, Race Capital, and Together Fund.

The funding round also attracted an impressive list of individual backers. According to Reuters, Google DeepMind chief scientist Jeff Dean is among the angels, alongside executives from OpenAI and Nvidia. Kindred founder Steve Jang has taken a seat on the company’s board.

Architect Labs is essentially trying to replicate the semiconductor industry’s most transformative shift from two decades ago: the fabless model. Back then, companies like TSMC made world-class manufacturing available to anyone with a design, separating fabrication from chip development. Now, Architect Labs wants to separate design itself from the companies that need it.

The startup calls this a “designless” semiconductor industry. In this vision, a company no longer has to become a chip company to get custom silicon. It does not have to lock itself into a single architecture for a decade. Instead, it presents a workload, and the AI generates the optimized chip to run it.

That ambition puts Architect Labs on a collision course with two of the industry’s most lucrative players. Reuters reports that the startup is targeting the custom-chip business currently dominated by Broadcom and Marvell. Those two firms design bespoke AI accelerators for cloud giants such as Amazon and Google, generating tens of billions of dollars in annual revenue together.

The demand for custom silicon is accelerating rapidly. AI labs, hyperscalers, and robotics companies all want chips tailored to their specific workloads. Off-the-shelf hardware can no longer keep up with the pace of innovation, fueling a broader search for new chip architectures.

“AI models have advanced dramatically across nearly every field, yet chip development cycles remain equally slow and painful,” said co-founder Ebrahim Hussain. His solution is not to attach AI agents to a decades-old design “flow.” Instead, he wants to rebuild the entire process from scratch, treating AI as a “first-class actor” in chip development.

Hussain brings a formidable background to the challenge. He skipped high school and started college at 15, then worked on custom chips at Apple and Tesla. His co-founder, Aaditya Subedi, researched AI code verification at Harvard. The two met at Stanford and then dropped out to launch the company.

The team currently numbers about 18 people, split between machine learning and hardware expertise. The company claims its members have collectively taped out more than 80 production chips. The roster includes alumni from Intel, Meta’s custom-silicon division, and machine-learning teams at Anthropic, DeepMind, and xAI.

Architect Labs says it has already deployed its technology with semiconductor partners. It expects AI-generated designs to tape out on leading-edge manufacturing nodes later this year. Neither claim has been independently verified yet.

The new funding will go toward scaling the company’s compute infrastructure, deepening its research, and funding co-design work with early partners. The longer-term vision is a tighter feedback loop where models, software, and hardware improve together. If the bet pays off, hardware will stop being the bottleneck that AI has to work around.