Elon Musk's generative AI company xAI is hiring for custom silicon developers.

In a previously unreported job listing, the company says that it is looking for employees able to "co-design next-generation AI systems, spanning from silicon to software compilers to models."

The role involves "designing and refining new hardware architectures to push the boundaries of computational efficiency," as well as using AI in the hardware design process.

Candidates are expected to have familiarity with accelerator-friendly hardware design languages like Chisel, alongside VHDL and Verilog.

They should ideally have experience "simulating training workloads on novel AI hardware architectures."

The hire would join an xAI team that aims to "pioneer next-generation AI systems that achieve breakthrough efficiency and scalability across new hardware, compilers, and models."

The team appears to be led by Xiao Sun, an xAI member of the technical staff for the past two years. He previously worked at Meta, and before that developed machine learning hardware and algorithms for IBM.

At IBM, he also developed 'beyond silicon' CMOS devices. Sun and xAI did not respond to requests for comment.

It is unclear if xAI is partnering with another company for the silicon development. Google worked with Broadcom to build its TPU chip family, and OpenAI is also working with Broadcom for its AI chip (after hiring TPU lead Richard Ho to run the project).

Musk's sister company Tesla also developed its own chips focused on processing video data to help train its semi-autonomous vehicles.

However, most of the team behind the 'Dojo D1' chip have left Tesla, and the state of its ongoing development is unclear. In 2023, Musk said that the chip was being used because of shortages of Nvidia GPUs. "Frankly, I don’t know if they could deliver us enough GPUs - we might not need Dojo - but they can’t," he said in an earnings call.

Like Tesla, xAI has deployed a significant quantity of GPUs, with more than 200,000 at the Colossus supercomputer in Memphis. Musk hopes to expand Colossus to one million GPUs, and xAI has purchased a site for a second data center in Memphis.

Earlier this month, xAI tweaked Grok to be less "politically correct." The system began spouting antisemitic tropes and referred to itself as MechaHitler. Several days later, xAI's head of infrastructure Uday Ruddarraju departed for OpenAI.