Anthropic's IPO filing discloses a four-chip strategy; analysts say it could shake Nvidia's pricing power

Anthropic's IPO Prospectus Discloses Chip Strategy; Analysts Say It Could Shake Nvidia's Pricing Power

Phoenix Tech Download the App for Exclusive Early Access October 28, 2026 06:19:17 From Beijing Updated October 28, 2026 06:44:51

Phoenix Tech News, October 28 (Beijing Time) — According to Business Insider, the prospectus Anthropic filed ahead of its listing shows that the company manages compute from multiple strategic partners as a unified resource pool and uses software to allocate tasks based on factors such as price and performance. Analysts believe this multi-chip strategy could weaken Nvidia's pricing power over the long term.

The prospectus shows that Anthropic uses four types of chips to train Claude and provide related services: Amazon AWS Trainium, Google TPU, Nvidia GPUs, and AMD GPUs. The company says its self-developed software scheduling layer can optimize hardware selection based on task requirements, and that this flexibility took years and significant investment to build.

Anthropic said the multi-chip architecture expands the scale of compute the company can use, reduces its dependence on a single supplier, and enables it to continuously assign tasks to the appropriate hardware.

Patrick Moorhead, founder and CEO of Moor Insights & Strategy, believes this capability could allow large AI companies to gradually reduce their dependence on Nvidia and gain more negotiating leverage.

"What could really be affected is Nvidia's pricing power, because a customer that can move tasks across four platforms is in a completely different negotiating position from one that cannot," Moorhead said.

IDC analyst Ashish Nadkarni said the multi-chip strategy can also reduce Anthropic's risk of being affected by fluctuations in Nvidia's chip supply.

Limited Short-Term Threat

Moorhead noted that Anthropic's use of multiple chips is not new, but the company is now pitching it to investors as a core advantage.

However, he said switching between chips remains difficult, especially when training AI models, and only the largest AI companies have the resources to make such an investment. He also said Anthropic is still spending billions of dollars on Nvidia compute.

Gil Luria, head of technology research at D.A. Davidson, said other large companies such as OpenAI may also have adopted multi-chip strategies, but for smaller companies, doing so may not be worth the corresponding effort and investment.

Brad Gastwirth, global head of research and market intelligence at Circular Technology, believes Anthropic's strategy will not pose a major threat in the short term. He said having more chip options does not eliminate Nvidia's advantages in performance, networking, and software.

Business Insider noted that if AI compute gradually becomes more like a commodity, Nvidia could face broader risks. Ken Griffin, founder and CEO of hedge fund Citadel, told Bloomberg last week that if compute supply exceeds demand, it could lead to falling rental prices.

"GPUs are like airlines; the marginal cost of running them is very low. Once there is overcapacity, prices will go into free fall," Griffin said.

As of the time of publication, Nvidia had not immediately responded to Business Insider's request for comment. (Editor/Yu Lei)

For more first-hand news, please download the Phoenix News app and subscribe to Phoenix Tech. For in-depth reports, search "Phoenix Tech" on WeChat.