ChipsThe story, in brief
OpenAI and Broadcom announce strategic collaboration to deploy 10 gigawatts of OpenAI-designed AI accelerators
10 gigawatts. OpenAI just locked in custom silicon infrastructure that rewrites the compute playbook.
The KeyNews take
Why it matters
OpenAI is moving beyond renting GPU capacity to designing and deploying proprietary accelerators at scale. This 10GW commitment signals a fundamental shift in how leading AI labs will compete on infrastructure—controlling the silicon, not just the models. By 2029, this becomes a competitive moat that smaller players can't easily replicate.
The key facts
6 to knowOpenAI-Broadcom multi-year partnership
10 gigawatts of custom AI accelerators
Deployment target: 2029
Co-developed Ethernet solutions for energy efficiency
Vertical integration play: OpenAI designing its own silicon
Strategic shift from GPU procurement to proprietary compute infrastructure
Go to the source
OpenAI Blogopenai.com
Publisher excerpt: OpenAI and Broadcom announce a multi-year partnership to deploy 10 gigawatts of OpenAI-designed AI accelerators, co-developing next-generation systems and Ethernet solutions to power scalable, energy-efficient AI infrastructure by 2029.