{
  "id": 20496434,
  "title": "OpenAI pushes forward with AMD partnership amid ongoing efforts to diversify from Nvidia",
  "url": "https://urgent.news/2026/11/17/openai-pushes-forward-with-amd-partnership-amid-ongoing-efforts-to",
  "topic": "finance",
  "section": "Finance & Markets",
  "published": "2026-11-17T20:16:07.000Z",
  "source": {
    "name": "Fortune",
    "slug": "fortune",
    "url": "https://fortune.com/2026/11/17/openai-reveals-first-amd-neocloud-provider-amid-nvidia-collapsejensen-huang-launching-his-comeback-campaign/"
  },
  "original_language": "en",
  "account": "In today's AI news, OpenAI is forging ahead with a partnership with AMD, despite being heavily reliant on Nvidia GPUs for its operations. OpenAI's CEO Sam Altman has expressed that a world without Nvidia would be \"hard to imagine,\" given the company's dependence on their hardware. Meanwhile, Anthropic has built a diverse chip base with a mix of Google TPUs, AMD GPUs, Amazon Trainium GPUs, and Nvidia GPUs. OpenAI is pursuing diversification through various initiatives, including designing its own chip architecture (Jalapeño), signing deals with non-Nvidia suppliers, and acquiring Google's TPU chips. OpenAI also recently announced a major compute deal with AMD to provide 6 gigawatts of capacity, which will be split into a first gigawatt coming online later this year and additional capacity expected in 2027. OpenAI's proprietary Jalapeño chip, developed in partnership with Broadcom, is set to scale up production in 2027. Despite these efforts, OpenAI may still benefit from Nvidia CEO Jensen Huang's potential return to the company.",
  "summary": "The major AI labs have long been trying to reduce Nvidia dependency, but Huang is determined to stay in the mix.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}