Google could build more AI accelerators than Nvidia sells in 2028, analyst claims

According to a Fubon Research report, Google plans to produce 12 to 15 million TPU v9 chips by 2028, potentially surpassing NVIDIA’s AI GPU shipments in the same period. This marks a significant shift... Read original →

Google aims to produce 12 to 15 million TPU v9 chips by 2028, potentially exceeding NVIDIA’s current GPU shipments, signaling a fundamental shift in the AI hardware landscape. The TPU v9’s fourfold compute die capacity intensifies pressure on TSMC’s production capacity, pushing Google to rely on Intel Foundry, underscoring a trend toward vertical integration in AI infrastructure. This move undermines NVIDIA’s CUDA ecosystem dominance and accelerates the trend toward specialized, tailored AI chips. From a compliance standpoint, geopolitical fragmentation between the U.S. and China raises supply chain risks, especially in advanced nodes, forcing companies to reassess manufacturing strategies. If NVIDIA fails to scale production or strengthen its ecosystem, it risks losing market share. Over the next two years, competition between TPU and GPU architectures will intensify, with TSMC and Intel Foundry’s capacity dynamics becoming pivotal factors in shaping the industry.