Nvidia is not happy with AMD and OpenAI (and it’s hilarious)

Quick Overview

Jensen Huang views AMD's massive partnership deal with OpenAI, which involves giving away a 10% stake in AMD, as a desperate and "imaginative" move driven by AMD falling far behind Nvidia's growth curve in the AI data center market, where Nvidia currently dominates with 92% market share.

Key Points: Jensen Huang described the AMD deal with OpenAI as "imaginative," "unique and surprising," especially since AMD is giving away 10% of the company before building out the product. AMD expects over $100 billion in revenue from OpenAI due to the deal, which is approximately three times AMD's previous annual revenue of about $30 billion. Nvidia holds 92% of the AI data center GPU market, and AMD is desperate to catch up, as evidenced by their willingness to cede 10% equity for partnership status. The Nvidia deal structure involves Nvidia investing $10 billion per gigawatt deployed, leading to $40 to $60 billion in returns for every gigawatt, effectively acting as an "infinite money glitch." OpenAI's motivation for the AMD deal is likely to introduce competition to force Nvidia's premium prices down, as OpenAI currently spends heavily on infrastructure for training models. Nvidia's strength is its entire ecosystem, including proprietary software like CUDA, making its converged system "much, much greater than AMD's is right now," despite AMD sometimes beating Nvidia on raw chip performance (flops). Jensen Huang projects the AI data center market value to reach upwards of $1 trillion by 2028, accelerating to three to four trillion by 2030.

Context: The transcript analyzes the commentary from Nvidia CEO Jensen Huang regarding a significant strategic partnership between AMD and OpenAI. This deal involves OpenAI potentially taking a 10% stake in AMD in exchange for substantial future purchases of AMD's AI goods, positioning AMD as a direct competitor to Nvidia, which currently dominates the highly lucrative AI chip market.

Detailed Analysis

Jensen Huang expressed thinly veiled disdain for AMD's deal with OpenAI, labeling it "clever" and "imaginative," interpreting it as a clear sign of desperation because AMD is significantly trailing Nvidia in revenue growth since the GPT moment, with Nvidia's revenue dwarfing AMD's $30 billion annual figure. The AMD deal promises OpenAI a 10% stake in AMD for effectively $0 per share while securing hundreds of thousands of AMD chips, whereas Nvidia's deal involves Nvidia investing directly into OpenAI, yielding massive returns based on AI demand. Huang emphasized that while AMD's raw chip performance might sometimes equal or beat Nvidia's (like the Instinct chip versus Blackwell), Nvidia maintains an insurmountable advantage through its complete, proprietary infrastructure—the convergence of GPUs, networking, memory, and the critical CUDA software ecosystem, which AMD lacks. OpenAI's underlying goal in fostering this competition is to leverage AMD to drive down Nvidia's premium pricing, as OpenAI faces immense infrastructure costs, potentially up to $60 billion per gigawatt deployed.