EasyFinance Live
Sign in

OpenAI’s Greg Brockman: AI Delivered “Roughly an Additional Year of Progress” on AMD’s Chip in Under 5 Months

Yahoo! Finance ·

Key points

  • OpenAI reports that its models provided roughly an additional year of progress on AMD’s chip within less than five months, and that some results are included in a design that has already been taped out, indicating a move from lab work to manufacturing readiness.
  • OpenAI has agreed to deploy up to 6 gigawatts of AMD GPUs and holds a warrant for up to 160 million AMD shares at a penny each, with the final tranche contingent on reaching the full 6-gigawatt deployment and a $600 share price.
  • MD/CEO commentary from AMD and market chatter suggest that faster design cycles driven by OpenAI’s models could erode Nvidia’s lead in AI chips and influence AMD’s pricing power, especially if Helios rack systems ship on time to major labs.
  • AMD has nearly tripled its R&D budget since 2021, totaling about $8.09 billion in 2025 (roughly 23.4% of revenue), a spend justified by the need to compete with Nvidia and accelerate chip development.
  • Analysts expect AMD revenue to grow from $34.6 billion in 2025 to $95.9 billion in 2027, a jump heavily dependent on Helios deployments and timely shipments of advanced-capacity chips by TSMC.
  • TSMC’s capital approvals around $40.9 billion, predominantly for advanced fabrication capacity, underpin faster ramp of new designs into production across the OpenAI/AMD/Nvidia ecosystem.
  • OpenAI’s broader strategy includes designing its own accelerators with Broadcom under a separate 10-gigawatt deal, enabling models trained on AMD’s chips to run on OpenAI’s hardware as well, signaling multi-vector leverage in AI hardware development.

Summary

OpenAI’s Greg Brockman claimed that collaboration with AMD engineers accelerated progress on AMD’s chip by roughly a year in under five months, aided by OpenAI models and a taped-out design. The partnership involves deploying up to 6 gigawatts of AMD GPUs and a warrant for up to 160 million AMD shares, aligning OpenAI’s data-center hardware needs with AMD’s roadmap and potentially pressuring Nvidia’s lead in AI chips. Analysts anticipate a substantial revenue rise for AMD by 2027, driven by new designs and large-scale deployments, while the broader supply chain—from TSMC capacity to Broadcom accelerators—supports faster design cycles. If the speedup holds, it could reshape competitive dynamics in AI hardware and pricing, though execution risk remains as key milestones like MI500 and 2027 revenue targets loom. The narrative also highlights OpenAI’s parallel development of its own accelerators and the strategic leverage gained from hardware partnerships and warrants.

This summary was written with AI. The full story has every detail.

Related