Lead: Anthropic launched its newest flagship model, Claude Opus 5, reclaiming the top spot on industry benchmarks — and in the same stretch, struck a computing partnership with AMD worth up to $5 billion and up to 2 gigawatts of capacity, reshaping the competitive landscape.

Claude Opus 5 reclaims the benchmark lead

On July 24, Anthropic released its new flagship model, Claude Opus 5, simultaneously across all platforms. According to reports, the model outperformed rival models that had previously held the top benchmark spots, swinging the generative-AI performance race back in Anthropic's favor. The industry has seen the lead change hands every few months in an unusually fast development race, and Opus 5 is the latest move in that contest.

A major AMD partnership, up to 2 gigawatts

Around the same time as the Opus 5 launch, Anthropic and semiconductor giant AMD announced a strategic partnership. AMD is reported to be making an equity investment in Anthropic worth up to $5 billion. At the same time, Anthropic plans to deploy AMD's newest AI accelerators — the Instinct MI450 series, packaged into data-center rack-scale systems called "Helios" — at a scale of up to 2 gigawatts to power its AI infrastructure. A gigawatt is an unfamiliar unit for most people, but it's roughly the output of a large power plant; committing that much capacity to a single AI company's computing needs is a striking figure. Reports indicate the first gigawatt of that capacity is expected to come online in the first half of 2027.

The deal has a second layer, too. AMD and Anthropic plan to work together on using Claude to accelerate workload optimization for AMD's GPUs and development of ROCm, AMD's software platform for GPU computing, and AMD itself is reportedly planning to roll Claude out broadly across its own internal design and engineering work. Some outlets describe it as a "circular" deal — one where money, hardware, and software expertise flow back and forth to reinforce both companies at once, rather than simple investment in one direction.

Why AMD?

NVIDIA GPUs have long been the default choice for AI computing, but Anthropic is reported to be deliberately diversifying its supply chain. According to reports, Anthropic is now working with AMD alongside NVIDIA, Amazon, and Google for chips and cloud infrastructure, and has secured more than 10 gigawatts of new computing capacity in total this year alone. Relying too heavily on a single supplier carries risk — from delivery delays to weaker pricing leverage — so spreading commitments across multiple partners helps secure the computing power AI development depends on. For AMD, which has been chasing NVIDIA in the AI chip market, landing a marquee deployment with a leading AI lab is seen as a significant win in its own right.

What it means for the competitive landscape

Anthropic and OpenAI sit at the front of the frontier-AI race, and the fight to lock down computing capacity has become just as intense as the fight over model performance. The AMD deal shows Anthropic isn't just competing on how smart its models are — it's also working to secure the computing foundation it will need for years to come. With a benchmark-topping model and a multibillion-dollar infrastructure bet landing in the same week, the AI race increasingly looks less like a pure contest of intelligence and more like a test of who can lock down — and sustain — the largest, most reliable supply of computing power.

Two announcements, one message

What makes the timing notable is that these were not two unrelated stories that happened to land in the same week — they reinforce each other. A new top-ranked model demonstrates that Anthropic's research team can still out-execute its rivals on raw capability. The AMD deal demonstrates that Anthropic has also lined up the physical infrastructure — chips, racks, and power — to keep training and serving models at that level for years, not just months. Taken together, reports frame the announcements as a signal to the rest of the industry, and to Anthropic's investors, that the company intends to compete on both fronts simultaneously rather than trading one for the other.

What comes next

The practical rollout will take time: the first gigawatt of AMD-powered capacity isn't expected online until the first half of 2027, meaning the computing benefits of the deal are still mostly ahead of Anthropic rather than already banked. In the meantime, the more immediate story is Opus 5 itself, and how long its benchmark lead holds before a competitor — most likely OpenAI or Google — answers with its own release. Given the pace the industry has kept up over the past year, few expect Opus 5 to hold the top spot for long, even as Anthropic's expanded computing base gives it a stronger platform for whatever comes after it.