Wafer, a San Francisco-based artificial intelligence hardware startup, has raised $40 million in Series A funding to expand its AI inference technology and infrastructure.

The round was led by Marathon and Chemistry, with participation from Wing Venture Capital, AMD Ventures, Outset Capital, Fifty Years, and Y Combinator.

Founded by Zonathan Zhang and Pranav Sharma, Wafer develops specialized high-performance chips and software designed to make AI inference faster and more cost-efficient.

Its technology is focused on running generative AI workloads at high speed while reducing the computing cost of processing each AI-generated token.

Wafer combines custom inference chips with software designed to manage different types of chip clusters. The company also uses continuous benchmarking tools to test enterprise AI workloads directly on its hardware, helping customers optimize performance before deployment.

The company operates an integrated hardware-and-software model and is targeting the growing demand for high-speed AI inference from AI-native companies and enterprise customers.

Wafer will use the new funding to secure high-performance computing and wafer-level manufacturing capacity, expand its engineering team and accelerate customer onboarding and go-to-market efforts.