Wafer Raises $4 Million Seed Round Led by AI Leaders for Chip Code Optimization AI

# Wafer Secures Seed Funding for AI Code Optimization Wafer has raised $4 million in seed funding from Google’s Jeff Dean, Wojciech Zaremba of OpenAI, and others. The funding supports training AI models to optimize code for efficient running on particular silicon chips. Emilio Andere is the cofounder and CEO of Wafer.

Emilio Andere says Wafer performs reinforcement learning on open source models to teach them to write kernel code that interacts directly with hardware in an operating system. Wafer adds agentic harnesses to existing coding models like Anthropic’s Claude and OpenAI’s GPT to improve their ability to write code that runs directly on chips. Wired reported these details.

Wafer is working with companies including AMD and Amazon to optimize software for efficient running on their hardware.

Industry Context on Custom Silicon Apple and other companies have used custom silicon for years to improve performance and efficiency of software on laptops, tablets, and smartphones.

Google and Amazon produce their own silicon to improve performance of their cloud-computing platforms. Meta said it would deploy 1 gigawatt of compute capacity with a new chip developed with Broadcom. High-end chips from AMD, Amazon Trainium, and Google TPUs offer similar raw floating point performance to Nvidia’s best silicon.

Emilio Andere told the author recently that the best AMD hardware, best Amazon Trainium hardware, and best Google TPUs give the same theoretical flops to Nvidia GPUs. Nvidia’s software ecosystem makes it easier to write and maintain code for its chips.

Challenges in Hardware Optimization Anthropic partnered with Amazon to build its AI models on Trainium and had to rewrite its model’s code from scratch to run efficiently on the hardware.

Emilio Andere says Wafer wants to maximize intelligence per watt. While Nvidia's software ecosystem eases code development, Wafer's CEO believes its AI approach could help address optimization challenges on rival hardware.

Related Startup in AI Chip Design Ricursive Intelligence is a startup founded by Azalia Mirhoseini and Anna Goldie, two ex-Google engineers, developing ways to design computer chips with artificial intelligence.

Azalia Mirhoseini says Ricursive is targeting physical design and design verification in chip design. Azalia Mirhoseini is an assistant professor at Stanford University. Azalia Mirhoseini and Anna Goldie developed a way for AI to optimize the layout of key components of computer chips while at Google.

The AI approach developed by Azalia Mirhoseini and Anna Goldie transformed how Google designs its own processors. The AI approach developed by Azalia Mirhoseini and Anna Goldie is now widely used in the industry to arrange features on different chips.