Google and Amazon are stepping up efforts to challenge Nvidia’s dominance in the artificial intelligence chip market, moving beyond using their own processors internally and pushing toward broader adoption of custom AI hardware.
The moves represent one of the biggest challenges yet to Nvidia’s position as the leading supplier of chips powering the global AI boom. Nvidia’s graphics processing units (GPUs) have become the industry standard for training and running advanced AI models, supported by its powerful software ecosystem.
Amazon Web Services (AWS) is exploring plans to sell its custom AI chips, known as Trainium, directly to other companies for use in their own data centres, marking a significant shift from its previous strategy of keeping the chips mainly within its cloud platform.
Amazon AI chief Peter DeSantis said AWS has begun discussions around making Trainium available to external customers, although the company has not disclosed potential buyers. The move follows comments from Amazon CEO Andy Jassy that the company’s AI chip business could become a major standalone opportunity.
Amazon believes its chips can offer customers a lower-cost alternative for AI workloads while reducing dependence on Nvidia’s expensive hardware. The company has already deployed Trainium chips across AWS infrastructure and is positioning them as part of a broader effort to compete in AI computing.
Google is pursuing a similar strategy with its Tensor Processing Units (TPUs), which were originally developed to power Google’s own artificial intelligence systems and services.
The company is now expanding efforts to make TPUs available more widely, using its cloud business and infrastructure investments to attract AI developers and companies looking for alternatives to Nvidia-powered systems.
Google’s strategy mirrors Nvidia’s own playbook: combine specialised hardware with software, cloud services and partnerships to create a complete AI ecosystem rather than simply selling chips.
More Read
The battle comes as demand for AI computing power continues to surge. Companies developing large language models and AI applications require massive amounts of computing capacity, creating a market expected to remain one of the fastest-growing areas in technology.
For years, Nvidia has benefited from a strong competitive advantage through its CUDA software platform, which has become widely used by AI researchers and developers. This ecosystem has made it difficult for competitors to replace Nvidia chips even when alternative hardware becomes available.
Analysts say Google and Amazon’s challenge is not only about chip performance but also about convincing customers to adopt different software environments and development tools.
Amazon’s advantage comes from AWS, the world’s largest cloud computing provider, allowing it to bundle chips with storage, networking and other services. Google, meanwhile, has decades of experience designing specialised processors through its TPU programme.
The companies are also responding to growing concerns among businesses about the cost and availability of Nvidia hardware as AI demand expands.
While Nvidia remains the market leader, the rise of custom chips from major technology companies could reshape the AI infrastructure industry by giving customers more choices and increasing competition.
The shift also reflects a broader trend among technology giants seeking greater control over the AI supply chain, from semiconductor design to cloud computing and software development.
For Nvidia, the challenge comes as rivals attempt to weaken its dominance by targeting the very foundation of its AI empire: the specialised chips that power the next generation of artificial intelligence.
