Meta and Amd Agree to 6‑Gigawatt GPU Deal; AMD to Issue 160 Million Shares
Meta and AMD announced a multiyear agreement on Tuesday under which Meta will purchase upward of 6 gigawatts' worth of AI chips as part of an expansive AI build‑out. The deal includes a 160 million share stock component tied to performance milestones, and amd's market reaction was immediate as trading moved sharply on the news.
Amd stock reaction and Jean Hu comments
Shares of AMD jumped as much as 10% in premarket trading before giving back some gains in the regular session. AMD Chief Financial Officer Jean Hu said the partnership is expected to drive "substantial multi‑year revenue growth" and be accretive to non‑GAAP earnings per share, adding that the performance‑based structure "tightly aligns AMD and Meta around execution and long‑term value creation. "
MI450 GPUs and Helios rack deployment timeline
The first GPUs under the agreement will be AMD's MI450 line, which Meta plans to deploy in its Helios rack‑scale data center systems. Those MI450 GPUs will be paired with EPYC CPUs and are slated for deployment in the second half of the year.
160 million shares and the 1‑gigawatt milestone
As part of the agreement, AMD will issue Meta 160 million shares of common stock. Those shares will vest in a series of tranches if AMD meets specified milestones; the first tranche vests when AMD ships its first 1 gigawatt of chips under the deal. The structure ties equity issuance directly to chip shipments, linking financial consideration to measurable operational progress.
CPU purchases: Venice, Verano and the rising role of processors
Meta also committed to purchasing a large additional set of CPUs, including AMD's Venice chip and its next‑generation Verano processor. The company framed CPUs as increasingly important for AI data centers as deployments expand toward agentic AI services and as teams focus on inference workloads, not only model training.
Nvidia agreement, Grace CPUs and GB200/GB300 superchips
Last week, Meta struck a separate multiyear agreement with Nvidia that will see Nvidia supply Meta with millions of its Blackwell and Rubin GPUs. Nvidia said Meta will host the first large‑scale deployment of its Grace CPU servers. Nvidia commonly pairs a Grace CPU with two Blackwell or Blackwell Ultra processors to form GB200 and GB300 superchips that fit into its rack‑scale platforms, and that architecture will be part of Meta's multi‑vendor strategy.
Capital spending, market moves and industry implications
The AMD agreement comes as Meta plans to spend upwards of $135 billion in capital expenditures throughout 2026 to support data center construction, chips and model training. Meta, Amazon, Google and Microsoft are expected to spend approximately $650 billion collectively on AI initiatives this year; that scale of investment has raised questions among some market participants about whether the outlays will pay off.
Since Meta disclosed its investment plans and reported earnings on Jan. 28, its shares have fallen 2. 5%. Over the same period, Google shares have dropped 8. 7%, Amazon 11. 9% and Microsoft 15. 5%. Wall Street has cooled on many chip stocks amid concerns about an AI investment bubble and the prospect that custom chips developed by Amazon, Google, Meta and Microsoft could challenge GPU incumbents in the data center.
In November, Meta was in talks with Google about using TPU chips for AI applications. Analysts cited in industry commentary say it will be difficult for those specialized chips to supplant the more general‑purpose AI processors that Nvidia and AMD produce. What makes this notable is that Meta is pursuing both large purchases from established GPU makers and exploration of alternative architectures, underscoring a hedged procurement approach as it scales AI infrastructure.
Unclear in the provided context: any additional commercial terms, precise shipment schedules beyond the first 1 gigawatt milestone, and the full tranche schedule for the 160 million shares.
