cd /news/artificial-intelligence/anthropic-to-buy-up-to-two-gigawatts… · home topics artificial-intelligence article
[ARTICLE · art-69167] src=siliconangle.com ↗ pub= topic=artificial-intelligence verified=true sentiment=↑ positive

Anthropic to buy up to two gigawatts of GPU capacity from AMD

Anthropic PBC will purchase up to two gigawatts of graphics cards from Advanced Micro Devices Inc. as part of a multibillion-dollar deal announced today. AMD will invest up to $5 billion in Anthropic and use Claude to speed up its software development efforts. The AI developer is adopting AMD's MI450 series GPUs, which will be installed in Helios rack systems providing up to 1.4 exaflops of FP8 performance.

read3 min views1 publishedJul 22, 2026
Anthropic to buy up to two gigawatts of GPU capacity from AMD
Image: Siliconangle (auto-discovered)

Anthropic to buy up to two gigawatts of GPU capacity from AMD

Anthropic PBC will purchase up to two gigawatts worth of graphics cards from Advanced Micro Devices Inc. as part of a multibillion deal announced today.

The partnership also has several other components. Notably, AMD will invest up to $5 billion in Anthropic and use Claude to speed up its software development efforts.

The artificial intelligence developer is adopting AMD’s MI450 series of graphics processing units. It will install the GPUs in appliances based on the latter company’s Helios rack design. According to AMD, each Helios system (pictured) includes 72 graphics cards that together provide up to 1.4 exaflops of performance when processing FP8 data.

Another core selling point of the appliance is that it includes 31 terabytes of HBM4 memory. That gives Helios 50% more capacity than Nvidia Corp.’s flagship Vera Rubin systems. AMD’s appliance also provides 43 terabits per second of scale-out bandwidth, a metric that measures how fast it can move data to and from the other racks in an AI cluster.

Helios includes not only GPUs but also central processing units and Pensando data processing units. The latter chips perform more than a dozen infrastructure management tasks. They track hardware reliability metrics such as latency, perform load balancing and encrypt traffic.

AMD and Anthropic will start deploying the hardware in the first half of 2027. Additionally, they will launch a multi-year effort to make the former company’s software development workflow more efficient. AMD plans to “broadly adopt” Claude as part of the effort.

The initiative will place particular emphasis on speeding up the development of ROCm, a software platform that ships with Helios. It includes a compiler that turns customers’ AI models into an AMD-compatible format. Additionally, ROCm provides tools that boost the performance of training and inference workloads.

“Together, we will accelerate AI adoption at scale and establish Helios as a major platform for the next generation of AI infrastructure,” said AMD Chief Executive Officer Lisa Su.

The deal is the latest in a series of high-profile hardware contracts inked by AMD since the start of the year.

Earlier this week, the chipmaker revealed that Microsoft Corp. is deploying Helios in its cloud data centers. In February, Meta Platforms Inc. announced plans to purchase up to 6 gigawatts worth of graphics cards from AMD. The Facebook parent’s initial order is for a custom GPU based on the MI450 chip series that Anthropic is adopting. Meta also plans to expand its use of AMD’s CPUs.

Meta, Anthropic and other major players in the AI market are purchasing chips from multiple suppliers to avoid supply chain bottlenecks. The latter company’s deal with AMD comes a few months after it committed to buying billions of dollars worth of AI accelerators from Google LLC and Broadcom Inc. Around the same time, Anthropic announced plans to purchase up to 1 gigawatt of additional GPU capacity from Nvidia.

Image: AMD

Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.

15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more** 11.4k+ theCUBE alumni**— Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network.

About SiliconANGLE Media

SiliconANGLE,

theCUBE Network,

theCUBE Research,

CUBE365,

theCUBE AIand theCUBE SuperStudios — with flagship locations in Silicon Valley and the New York Stock Exchange — SiliconANGLE Media operates at the intersection of media, technology and AI.

Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @anthropic pbc 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/anthropic-to-buy-up-…] indexed:0 read:3min 2026-07-22 ·