Wilson Vu, senior director of hardware architecture for AI systems at Microsoft, has left the company after 12 years to join OpenAI as a member of the technical staff working on compute architecture.

Vu spent those 12 years at Microsoft across three separate stints. In his most recent role, he ran a team developing AI and GPU platforms.

His move lands at OpenAI as the generative AI firm builds out its own AI hardware to support a growing compute fleet. That fleet currently includes:

• Nvidia GPUs
• AMD GPUs
• Google TPUs
• OpenAI's own Jalapeño chips

What does the hire signal?

OpenAI's recruitment of a senior hardware architect points to the company's continued push to design and deploy custom silicon rather than rely solely on external suppliers. The compute architecture group Vu joins sits at the center of that effort.

Jalapeño, OpenAI's in-house chip, is expected to be deployed in limited quantities later this year. A wider deployment is planned for next year.

Where did Vu work before?

Vu's career spans several major infrastructure and cloud players. Before his most recent Microsoft role, his positions included:

• Hardware engineer on AI systems and accelerated platforms at Meta
• Systems design lead at Salesforce
• Hardware development engineer at Supermicro
• Principal engineering manager at GoDaddy

At Microsoft, his remit covered AI and GPU platform development — the same domain OpenAI is now staffing as it scales its compute fleet across multiple chip suppliers and its own silicon.

The hire continues the pattern of talent moving from hyperscaler hardware teams into OpenAI as the company expands its infrastructure ambitions.