Google is expanding its custom artificial intelligence chip business as competition grows across the global data centre market. The company is using financial guarantees, manufacturing deals, and software tools to attract more outside customers to its Tensor Processing Units, or TPUs.
Google has provided a $3.2 billion financial guarantee for the Lake Mariner data centre campus in western New York, according to reports. TeraWulf operates the site, which will host computing systems built around Google’s TPU chips. The project will supply capacity to Fluidstack, which will then provide resources to Anthropic for work on its Claude AI models.
The arrangement expands Google’s role beyond chip design and cloud services. It also places the company in the financing structure behind the infrastructure that uses its hardware.
NVIDIA has used similar support for GPU projects by helping customers secure leases and funding for large computing clusters. Google is now using financial support to raise demand for its processors.
Google first developed TPUs for its internal services. The company later opened access through Google Cloud and is now seeking more customers outside its operations. Broadcom CEO Hock Tan has said the TPU business has generated “tens of billions of dollars” in revenue, although Google has not released a separate figure.
Reports project Google TPU shipments could reach 4.3 million units in 2026 and rise to 35 million by 2028. Those estimates remain forecasts rather than confirmed sales. TrendForce expects custom AI chip sales to grow faster than standard GPU sales in 2026, as cloud companies build processors for selected workloads and seek more control over costs and supply.
Meanwhile, Google is expanding its production network. The company has reportedly placed an order with Intel to manufacture more than three million TPUs in 2028. It is also said to be in talks with Marvell Technology about new custom chip designs. These plans would add suppliers while increasing production capacity.
Hardware alone does not decide the AI chip market. NVIDIA’s CUDA software platform supports AI tools and developers. Google has introduced TorchTPU to help PyTorch developers run workloads on TPUs. The software aims to reduce the workload required when moving certain AI tasks off NVIDIA systems.
Google is also linked to a $35 billion infrastructure arrangement involving Broadcom, Anthropic, Apollo Global Management, and Blackstone across five US data centres. Reports describe the plan as a financing structure for sites using Google chips. The full commercial terms have not been disclosed, so the final scale and timing remain subject to agreements between the companies.
Other technology groups are also developing custom AI processors. Amazon, Microsoft, and Meta have invested in their own chips to secure additional supply options. Still, these companies use NVIDIA GPUs for AI workloads. Reports that Google and Amazon informed NVIDIA CEO Jensen Huang about their plans show that custom chips and NVIDIA hardware remain part of their current strategies.
Google’s approach combines chip design, cloud services, software support, and project financing. The company is using these tools to place TPUs in more data centres and reach customers who need large amounts of AI computing power.
Also Read: Micron Joins $1 Trillion Club After UBS Raises Stock Target on Strong AI Chip Demand