In a significant move last month, OpenAI, the leading artificial intelligence developer, finalized a massive $100 billion deal with Nvidia, the world’s most valuable publicly traded company, to secure its supply of advanced computer chips.
Following closely on the heels of the Nvidia deal, OpenAI has now forged a similar partnership with AMD. This collaboration positions AMD, a key competitor, to play a crucial role in powering cutting-edge AI technologies such as OpenAI’s ChatGPT, intensifying the race to dominate the AI chip market.
OpenAI announced on Monday that it plans to integrate AMD chips into its operations starting in the latter half of next year. These chips will be deployed in new data centers, distinct from those already slated for construction in Texas, New Mexico, Ohio, and an upcoming Midwestern location.
This ambitious rollout involves deploying enough AMD chips to draw an astonishing 6 gigawatts of power over several years — a colossal amount equivalent to the total electricity consumption of every household in Massachusetts. For comparison, their earlier agreement with Nvidia committed to deploying chips requiring 10 gigawatts.
While AMD isn’t directly investing in OpenAI, the agreement includes a significant clause: OpenAI can acquire up to 160 million shares in AMD at a mere penny per share. This would grant OpenAI a 10% ownership stake in the chipmaker and potentially provide substantial capital to fund its extensive new computing infrastructure projects over the coming years.
News of the partnership sent AMD’s shares soaring, with a premarket jump of over 20 percent on Monday.
This deal is emblematic of a broader, industry-wide push where technology giants are pouring hundreds of billions into new data center construction. OpenAI, alongside behemoths like Amazon, Google, Meta, and Microsoft, is set to collectively spend over $325 billion on these facilities by the close of this year alone.
While established tech giants such as Amazon, Microsoft, and Google can comfortably fund these massive data center projects from their annual multi-billion dollar profits, newer and smaller players like OpenAI face a different challenge, often needing to raise or borrow tens of billions to build their essential computing infrastructure.
OpenAI’s ambitious ‘Stargate Project’ previously outlined plans to construct new data centers across the United States, in collaboration with Oracle and SoftBank, with an estimated cost exceeding $400 billion. However, as a startup, OpenAI and its partners have had to devise innovative strategies to secure the colossal funding required for these ventures.
The recent Nvidia agreement wasn’t just about chip deployment; it also included a staggering $100 billion investment from Nvidia into OpenAI. Following an initial $10 billion, Nvidia committed to infusing another $90 billion over several years. This highlights a growing trend where OpenAI secures significant funding directly from its key suppliers.
(The New York Times initiated legal action against OpenAI and Microsoft in 2023, alleging copyright infringement of news content used to train their AI systems. Both companies have publicly refuted these allegations.)
