Executives from Meta and AMD shaking hands to announce a multibillion-dollar deal for custom AI chips, with logos and tech visuals on a backdrop screen.
Executives from Meta and AMD shaking hands to announce a multibillion-dollar deal for custom AI chips, with logos and tech visuals on a backdrop screen.
Billede genereret af AI

Meta agrees to multibillion-dollar AI chip deal with AMD

Billede genereret af AI

Meta has announced a multi-year partnership with AMD to purchase up to six gigawatts of custom AI chips, potentially gaining a 10% stake in the chipmaker through performance-based shares. The deal, valued in double-digit billions per gigawatt, aims to support Meta's expanding AI infrastructure across its platforms. This arrangement mirrors a similar agreement AMD made with OpenAI last year.

Meta and AMD revealed their partnership on February 24, 2026, under which Meta will buy billions of dollars' worth of AMD Instinct GPUs based on the MI450 architecture, optimized for Meta's workloads. The agreement covers up to 6 gigawatts of computing power, equivalent to the energy needs of 5 million US households for a year or nearly 2,000 large solar panels per gigawatt. Deployment of the first gigawatt is scheduled for the second half of 2026.

As part of the deal, AMD is issuing Meta a performance-based warrant for up to 160 million shares of its common stock at an exercise price of $0.01. This could result in Meta owning 10% of AMD if milestones related to shipments, stock price thresholds up to $600, and technical achievements are met. The warrant expires in February 2031, with the first tranche vesting upon the initial gigawatt shipment.

AMD CEO Lisa Su described the arrangement as expanding their strategic partnership, stating, "We are proud to expand our strategic partnership with Meta as they push the boundaries of AI at unprecedented scale." She added that each gigawatt of compute is worth double-digit billions and noted, "In some sense, Meta is taking a big bet on AMD, and we are also giving Meta a chance to participate if AMD shareholders do well."

Meta CEO Mark Zuckerberg called AMD "an important partner for many years to come." Santosh Janardhan, Meta's head of infrastructure, emphasized diversification: "We don’t believe that a single silicon solution will work for all of our workloads. There’s a place for Nvidia, there’s a place for AMD and… there’s a place for our own custom silicon as well. We need all three."

The chips will primarily handle inference workloads for AI models on platforms like Facebook, Instagram, and WhatsApp. Meta also plans to deploy millions of AMD EPYC CPUs and serve as a launch customer for the sixth-generation EPYC processors. This deal follows a similar one with OpenAI announced in October 2025, which also offered a 10% stake for six gigawatts of GPUs.

Meta intends to nearly double its AI infrastructure spending to $135 billion this year, amid tightening supplies and rising prices for components like RAM. Analysts note such "circular" transactions could create dependencies between AI firms and chipmakers, potentially amplifying risks if AI demand falls short of expectations. Meta stated, "By diversifying our partnerships and technology stack, we’re building a more resilient and flexible infrastructure."

Hvad folk siger

X discussions highlight excitement over Meta's massive 6GW AI chip deal with AMD, valued at tens to hundreds of billions, positioning AMD as a strong Nvidia alternative and accelerating AI infrastructure buildout. Users note the scale's equivalence to powering millions of homes, strategic equity warrants for Meta, and bullish implications for AI capex, with minor skepticism on dilution but overwhelmingly positive sentiment from analysts, investors, and officials.

Relaterede artikler

Illustration of Oracle's OCI Zettascale10 AI supercomputer in a modern data center, highlighting AMD chips and advanced GPU technology.
Billede genereret af AI

Oracle unveils largest AI supercomputer with AMD chips

Rapporteret af AI Billede genereret af AI

Oracle has announced OCI Zettascale10, described as the world's largest AI supercomputer, featuring six times more GPUs than its 2024 predecessor. The company continues to rely on AMD chips for its superclusters despite Nvidia's industry dominance. This move aligns with Oracle's strategy to integrate AI directly with users' data wherever they operate.

Meta has struck a major deal with Nvidia to purchase millions of Blackwell and Rubin GPUs, marking a new phase in AI infrastructure. The partnership includes deploying Nvidia's Confidential Computing technology in WhatsApp to enable AI features while protecting user data. This expands on Meta's ambitious AI investments amid growing demand for integrated computing solutions.

Rapporteret af AI

Meta has announced a collaboration with Nvidia to develop a hyperscale AI infrastructure designed to support billions of users globally. The initiative integrates Meta's production workloads with Nvidia's hardware and software ecosystem. The partnership aims to enhance AI capabilities at an unprecedented scale.

Nvidia CEO Jensen Huang has dismissed reports suggesting the company is backing away from a significant investment in OpenAI. Speaking in Taipei, Huang emphasized his support for the AI firm and indicated the deal could be Nvidia's largest ever, though not approaching the previously discussed $100 billion figure. This comes amid earlier announcements and recent speculation about the partnership's progress.

Rapporteret af AI

The smart glasses market experienced a 139% year-over-year growth in the second half of 2025, driven largely by Meta's AI-enabled products. Meta captured 82% of shipments, with AI glasses making up 88% of the total. This surge highlights the rising popularity of advanced eyewear, though prices remain a barrier.

Elon Musk announced that Tesla's next-generation AI5 chip is almost complete, with early work already underway on AI6 and plans for a nine-month development cycle for future iterations. The chips are expected to become the highest-volume AI processors globally, powering vehicles, robots, and more. This update highlights Tesla's aggressive push in AI hardware for autonomy and beyond.

Rapporteret af AI

Elon Musk met with SpaceX's Brent at midnight to oversee xAI's new Colossus 2 supercomputer, set to become the world's largest and most powerful data center. The facility is incorporating over $375 million worth of Tesla Megapacks for sustainable energy to support advanced AI workloads.

 

 

 

Dette websted bruger cookies

Vi bruger cookies til analyse for at forbedre vores side. Læs vores privatlivspolitik for mere information.
Afvis