The first gigawatt-scale deployment of the OpenAI deal is slated for the second half of next year, marking a pivotal milestone in AMD’s AI infrastructure push.

The article introduces a PC gaming writer as a profile subject and includes a note about audience comment policy unrelated to the core content.

AMD’s strategy aligns with hyperscalers building more in-house silicon and represents a long-term rebalance after years of rebuilding in high-performance computing.

AMD’s chief executive argues that the AI boom is not a bubble and that demand for AI chips will expand as companies require immense computing power from AMD and peers alike.

AMD’s AI ambitions run parallel to Nvidia’s gains, underscoring market enthusiasm for AI potential.

Su stresses AMD’s challenge is advancing its own roadmap quickly to stay ahead of future deployments, not merely matching rivals like Nvidia, Google, or Amazon.

AMD sees potential benefits from re-entering certain export markets under favorable terms, despite acknowledging up to $800 million in charges tied to inventory and purchase commitments.

A Wired interview hints that Su’s broader views are not fully disclosed because the full interview hasn’t been published yet.

AMD is pursuing large data-center commitments, including a multi-gigawatt OpenAI deployment and the resumption of MI308 shipments to China under a new export-control regime.

China will not drive near-term AMD data-center revenue, but the region remains capable of absorbing large accelerator shipments quickly.

The broader AI debate features governments showing interest and potential support for adoption, while industry figures warn against overinvestment.

Each generation of AI models raises performance expectations, sustaining investments in training and inference clusters and testing AMD’s ability to translate confidence into real hardware and commitments.