Key point: Google invests billions in TPU chip production and data center financing to threaten Nvidia’s 90 percent AI market share, copying Nvidia’s proven infrastructure lock-in strategy in the process.
Google is massively financing the expansion of its Tensor Processing Units (TPUs) and the development of AI data centers to attack Nvidia’s 90 percent market share in AI chips. With investments such as the $3.2 billion Lake Mariner project, Google is copying Nvidia’s strategy of securing sales through infrastructure financing.
Google is systematically expanding its TPU marketing. The Lake Mariner project in New York is the central example: Alphabet is providing $3.2 billion to rent thousands of Google TPUs to Anthropic through a data center. This strategy mimics Nvidia’s proven model – binding customers through financial commitments and hardware availability.
Google’s TPU development dates back to 2013. Jeff Dean, now Chief Scientist at DeepMind, diagnosed the problem at the time: if Google wanted to deploy a language model for 100 million users, this would have doubled server capacity. So specialized hardware had to be created. For years, Google used TPUs only internally for web search and speech recognition. Since AI demand exploded, Google has been selling the chips through its cloud division and direct sales. Anthropic is already training with the seventh generation of TPUs.
Concrete successes are emerging: Google closed a $5 billion partnership with Blackstone to build AI-specialized cloud services. Customer Citadel Securities reports 30 percent cost reductions and four times higher processing speed with TPU migration. In parallel, Google is planning global infrastructure expansion with $85 billion in equity – in addition to Lake Mariner, also the $7 billion River Bend project in Louisiana and a $1.4 billion project in Texas.
Nvidia maintains its dominant market position nonetheless: over 90 percent of the global AI chip market. The reason is the established CUDA software architecture, to which developers remain bound. Nvidia CEO Jensen Huang remained skeptical: “I would like to hear how they demonstrate the cost advantage of TPUs” (April interview). Since December, Amin Vahdat as Chief Technology Officer coordinates Google’s global infrastructure expansion. Vahdat views the competition pragmatically – Google continues to operate Nvidia GPUs alongside TPUs in its data centers.
Source: www.it-daily.net · Published June 22, 2026
Lumi AI News — AI-assisted curation in accordance with Article 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.7.1.