Google Reportedly Plans to Deploy 12-15 Million TPU v9 Chips by 2028
According to a market analysis report, Google plans to deploy 12 to 15 million of its custom TPU v9 AI ASIC chips by 2028. This projected volume could potentially match or exceed NVIDIA's estimated shipment of 12.4 million AI GPUs in the same timeframe. If realized, this massive deployment would challenge NVIDIA's dominance in the AI hardware market and significantly shift the competitive landscape. It also highlights the growing industry trend of tech giants relying on custom ASICs over general-purpose GPUs for large-scale AI workloads. The TPU v9 is expected to feature a 4-compute-die chiplet architecture, which will sharply increase demand for advanced semiconductor manufacturing and packaging. To meet these needs, Google is reportedly looking beyond TSMC to secure production capacity from Intel and Samsung.
## BACKGROUND
Tensor Processing Units (TPUs) are application-specific integrated circuits (ASICs) custom-designed by Google specifically for accelerating machine learning workloads. Unlike general-purpose GPUs, TPUs are optimized for tensor and matrix operations. As chip design hits physical limits, manufacturers are increasingly adopting chiplet architectures, which combine multiple smaller dies (like compute and I/O dies) onto a single package to improve yields and performance.