Meta Platforms (NASDAQ: META) announced today a multiyear partnership with Advanced Micro Devices, Inc. (NASDAQ: AMD) to deploy up to 6 gigawatts worth of AMD’s graphics processing units in its AI data centers. This move comes right after Meta’s commitment last week to use millions of processors from NVIDIA Corporation (NASDAQ: NVDA) for its artificial intelligence expansion. Think of gigawatts as a measure of power capacity. In AI terms, 6 gigawatts could support the equivalent of 2.4 million to 3 million GPUs running tasks like training large language models or processing real time data.
Data centers form the foundation for companies like Meta to handle AI workloads. These facilities pack thousands of servers with specialized chips that perform calculations at speeds regular computers cannot match. Graphics processing units, or GPUs, shine here because they handle many operations at once, perfect for the math heavy demands of AI. AMD plans to supply its Instinct series, including models like the MI450, along with EPYC processors and rack systems called Helios. The deal spans through 2030 and covers next generation architectures such as CDNA 4 and CDNA 5.
What sets this apart goes beyond buying hardware. AMD gave Meta a performance-based warrant to purchase up to 160 million shares, which equals about 10% of AMD’s total shares outstanding. These shares come at a low exercise price of $0.01 each and vest in steps tied to deployment milestones. The first tranche activates after Meta installs the initial 1 gigawatt, with further portions unlocking as it reaches higher levels up to 6 gigawatts. Later stages also depend on AMD’s stock price hitting targets, including hurdles as high as $600 per share. This structure aligns their goals. Meta gains early access to cutting edge tech, while AMD locks in committed purchases over years.
Consider the context of Meta’s choices. Just last week, on February 17th, Meta announced plans to integrate millions of Nvidia chips, including Grace CPUs, into its infrastructure. Nvidia has led the AI chip market for years, but supply constraints and rising costs push big users to explore options. By adding AMD’s GPUs, CPUs like the sixth generation EPYC Venice, and full rack solutions, Meta spreads its bets. Analysts estimate the AMD commitment alone could cost more than $100 billion, based on projected chip volumes and pricing. This approach helps manage risks from depending on a single vendor in a market where demand outstrips supply.
Investors reacted with clear enthusiasm to the news. AMD’s stock surged around 14% in early trading today, a strong sign of confidence in its AI capabilities, though it pulled back some later in the session. Reports noted premarket gains between 10% and 15%, with commentary highlighting this as proof AMD can compete at scale. Meta’s shares saw a slight dip, but attention focused on the supplier dynamic. Such jumps reflect how partnerships like this validate a company’s tech in the eyes of Wall Street.
This development hints at shifts in the AI hardware world. Tech giants pour billions into infrastructure to stay ahead, and no one wants bottlenecks. Meta’s CEO Mark Zuckerberg emphasized the need for efficient inference compute, pointing to this deal as part of broader diversification. For AMD, under CEO Lisa Su, it rewards heavy investments in accelerators that rival Nvidia’s offerings. Competition may drive down costs and spur faster improvements for everyone involved.
Deals like this one show how interconnected tech and finance have become. Companies tie equity to performance to build trust and commitment. As AI demand grows, expect more firms to mix suppliers for reliability. Meta’s path forward relies on both proven leaders like Nvidia and challengers like AMD to power its vision.
