Architect Labs secures $24 million for the design of custom AI chips.

      Designing a chip is a challenging task. It requires years of effort, hundreds of millions of dollars, and a small group of specialists, most of whom are employed by a few large corporations. Architect Labs aims to change this dynamic with the help of AI. The Palo Alto-based startup emerged from stealth mode on Thursday, securing a $24 million seed funding round. Its objective is to develop an AI system that can design and validate custom chips completely.

      Kindred Ventures led this funding round, joined by TQ Ventures, Race Capital, and Together Fund, along with a notable group of angel investors. According to Reuters, this group includes Google DeepMind chief scientist Jeff Dean, as well as executives from OpenAI and Nvidia. Kindred founder Steve Jang will take a seat on the board.

      Aiming for a ‘designless’ chip industry, the proposal draws on the history of semiconductors. Two decades ago, the fabless model separated design from manufacturing, allowing TSMC to provide high-quality fabrication services to anyone with a design. Architect Labs seeks to replicate this success at the design level, envisioning a “designless” semiconductor industry. In this scenario, companies are no longer required to become chip manufacturers or commit a decade to a specific architecture; instead, they can simply present a workload and receive the corresponding silicon.

      This ambition puts Architect Labs in direct competition with two significant players in the industry. Reuters notes that the company is targeting the custom-chip markets of Broadcom and Marvell, both of which create specialized AI accelerators for major cloud providers like Amazon and Google, generating tens of billions of dollars annually. The demand for custom silicon is continuously increasing, as AI laboratories, hyperscalers, and robotics manufacturers seek chips tailored to their unique workloads, with off-the-shelf hardware falling short.

      Co-founder Ebrahim Hussain pointed out, “AI models have advanced dramatically across nearly every field, yet chip development cycles remain equally slow and painful.” His approach is not to simply integrate AI agents into outdated design processes; rather, he aims to reconstruct the design process from the ground up, positioning AI as a “first-class actor.”

      The founders are fully committed to this vision, having bet their careers on it. Hussain, who began college at the age of 15 after skipping high school, has experience with custom chips from his time at Apple and Tesla. His co-founder, Aaditya Subedi, conducted research on AI code verification at Harvard. The two met at Stanford, then decided to leave school to establish the company.

      The team consists of around 18 individuals, with expertise in machine learning and hardware. Collectively, they have produced over 80 production chips and include alumni from Intel, Meta’s custom silicon projects, and machine learning teams from Anthropic, DeepMind, and xAI.

      Looking ahead, Architect Labs reports that it has begun working with semiconductor partners and anticipates that AI-generated designs will be produced on cutting-edge nodes later this year, although these claims have not yet been independently verified. The new funding will be used to expand computational capacity, enhance research, and support co-design efforts with initial partners. Ultimately, the broader vision is to create a more integrated development cycle, where models, software, and hardware improve in tandem. If successful, this could redefine hardware's role in AI development.