Architect Labs raises $24M to automate custom chip design with AI

Architect Labs, a Palo Alto-based semiconductor design automation startup, emerged from stealth on June 18, 2026, with a $24 million seed round led by Kindred Ventures. The round included participation from TQ Ventures, Race Capital, Together Fund, and a notable slate of angels: Transformer co-inventor Lukasz Kaiser, Perplexity CEO Aravind Srinivas, former OpenAI executive Srinivas Narayanan, Stanford professor Kunle Olukotun, and executives from NVIDIA, Google, and OpenAI. Kindred founder Steve Jang joins the board.

Architect Labs is building an AI system to design and verify custom chips end-to-end, targeting a fundamental bottleneck in silicon development: the chip design process currently takes 18–24 months and costs hundreds of millions in labor, R&D, and tape-out risk. The company is positioning this as a 'designless' semiconductor industry—analogous to how the fabless model (TSMC) decoupled design from manufacturing two decades ago. Architect Labs wants to decouple design itself from chip companies, allowing software companies, AI labs, and enterprises to specify a workload and receive optimized silicon without becoming semiconductor companies. The founding team has collectively taped out 80+ production chips and includes veterans of Intel's Data Center Division, Meta's custom silicon, and AI research at Anthropic, DeepMind, and xAI.

AI's infrastructure demands are driving the economic case: general-purpose GPUs no longer satisfy specialized compute, networking, and memory bandwidth needs for custom workloads in datacenters, robotics, autonomous systems, defense, and edge devices. Companies like Amazon, Google, and Microsoft already employ chip design teams (AWS Trainium, Google TPU, Azure Maia) because off-the-shelf hardware cannot compete. Broadcom and Marvell extract tens of billions in revenue annually from custom chip design services, but they move slowly. Architect Labs aims to compress design timelines from years to months.

For infrastructure architects and chip procurement strategists, this signals acceleration in the custom silicon race: as AI workloads diverge (training vs. inference vs. specific domains), the ability to design fit-for-purpose silicon quickly becomes a competitive advantage. If Architect Labs succeeds, the design bottleneck dissolves, enabling more players to commission custom chips and reducing Nvidia's leverage. The founding team's pedigree and investor lineup (Kindred, frontier lab researchers) suggest serious technical execution risk is priced in—the challenge is product-market fit and customer adoption at scale.