Amazon and Google Challenge Nvidia’s AI Chip Dominance: Implications for Investors and Tech Entrepreneurs
SAN FRANCISCO: Since the artificial intelligence (AI) boom began three years ago, it has been widely acknowledged that nearly every major AI initiative starts with chips supplied by Nvidia. However, in the past year, two of the technology sector’s most influential companies — Amazon and Google, who are also significant customers of Nvidia — have made noteworthy inroads into Nvidia’s formidable market position.
Amazon has recently integrated thousands of its proprietary AI chips into an extensive network of data centers in Indiana, which are being utilized by Anthropic, one of the leading AI firms globally. Concurrently, Google has formed multiple agreements with Anthropic, supplying chips as the company establishes its own data centers in New York, Texas, and other locations.
Numerous chip manufacturers have spent years attempting to rival Nvidia, including established firms like Advanced Micro Devices (AMD), startups such as Cerebras, and tech giants like Microsoft and Meta. Nonetheless, the burgeoning chip enterprises at Amazon and Google represent the most serious competition Nvidia has encountered.
“While Nvidia commands a significant portion of the market, even small shares can translate into billions in revenue,” noted Jordan Nanos, an analyst at SemiAnalysis, a data center research firm.
In 2025, Amazon reported that revenue from its AI chip, Trainium, exceeded “multiple billions,” according to CEO Andy Jassy during a recent investor earnings call. Google’s tensor processing units, or TPUs, generated tens of billions, as revealed by Broadcom CEO Hock Tan shortly thereafter.
Despite being valued as the world’s most valuable publicly traded company, Nvidia still dominates 92 percent of the lucrative market for specialized chips essential for developing online chatbots, image generators, and various AI technologies. In 2025, its AI chip revenue neared $200 billion.
Amazon and Google are navigating a precarious path as they compete against Nvidia while still relying on it as their primary supplier. Anthropic, like most AI companies, is heavily dependent on Nvidia’s chips but has grown increasingly critical of the firm for its sales to China and is working on minimizing that reliance. This dynamic has fostered tens of billions in chip revenue for both Amazon and Google, Anthropic’s largest investors.
Interest in Google’s chip business surged when Tan disclosed that the company had sold $10 billion worth of Google chips to Anthropic, which subsequently placed an additional order for $11 billion.
Anthropic is installing these chips in data centers being constructed in collaboration with a company called Fluidstack. As Fluidstack incurs debt for this project, Google has agreed to support the deal by ensuring repayment if Fluidstack is unable to do so.
Since 2017, Google has leased TPUs through its cloud services but established a new paradigm with Anthropic, allowing the installation of these chips in data centers not owned by Google. This decision is driven by Anthropic’s desire to maintain strict oversight of its AI technologies. CEO Dario Amodei and his team are concerned about potential dangers posed by AI, leading them to insist that even close partners refrain from accessing the source code of the company’s software. By operating its own data centers, Anthropic aims to secure this sensitive code.
This unconventional arrangement highlights the growth of Google’s chip business, signaling that revenue is soaring into the tens of billions. Andrew Feldman, CEO of Cerebras, remarked, “That is a lot of money. They are selling an extraordinary volume of chips.”
Amazon commenced its chip development approximately three years after Google. In the fall of 2023, the company announced a $4 billion investment in Anthropic. An Amazon executive later revealed that Anthropic had secured the partnership by committing to develop its AI using Amazon’s chips, with the aim of establishing a viable alternative to Nvidia.
The following year, in collaboration with Anthropic, Amazon customized a new version of its Trainium chip specifically for data centers dedicated to training and deploying AI. While these chips may not match the power of those from Nvidia or Google, Amazon is installing double the quantity in each data center to maximize computing power while conserving energy.
With the expansion of chip deployments across Indiana and other facilities, Amazon’s chip revenue is reportedly growing at a rate of 150% every three months, according to David Brown, a vice president in Amazon’s cloud computing division, Amazon Web Services. “Growth is limited by how quickly we can get the chips out there,” he stated in an interview.
While Anthropic currently drives much of Amazon’s chip revenue, experts predict that such a high-profile partnership could encourage broader market changes. Anthropic’s choice to utilize chips from Amazon or Google demonstrates that Nvidia’s offerings are not the sole option available in the marketplace.
— The New York Times
Special Analysis by Omanet | Navigate Oman’s Market
The shift towards competition in AI chip development by Amazon and Google signals a potential transformation in the tech industry, creating new opportunities for local businesses in Oman to innovate and possibly collaborate in AI applications. However, this also presents risks for companies overly reliant on Nvidia, prompting smart investors and entrepreneurs to diversify their partnerships and explore alternatives in chip sourcing. Strategic investment in emerging tech and local AI initiatives could yield significant advantages as the market evolves.
