Google Speeds Up Next-Gen TPU Mass Production
Google is accelerating the mass production of its proprietary artificial intelligence (AI) chip, the Tensor Processing Unit (TPU). Following the initiation of mass production for the eighth-generation inference TPU in the second half of this year, the company is actively expanding its supply chain i
Google is accelerating the production of its AI chip, the Tensor Processing Unit (TPU), specifically the eighth-generation models, TPU 8i (Zebrafish) and TPU 8t (Sunfish). These new chips are being manufactured in collaboration with major Taiwanese server manufacturers, including TSMC. The TPU 8i, with 216GB and 288GB of high-bandwidth memory, boasts improved performance per dollar compared to its predecessor, the Ironwood.
Google's Vice President and Chief Technology Officer for AI Infrastructure, Amin Vahdat, predicts that despite potential advancements in AI model efficiency, the overall demand for computing power will not diminish. He stresses the importance of reducing data movement costs and power consumption in future chip designs. Vahdat plans to hold closed-door meetings with major Taiwanese companies to discuss next-generation TPU server orders.
The expansion of the TPU market is anticipated to benefit the domestic semiconductor industry, as Samsung Electronics is reportedly considering a potential $200 billion partnership in memory and foundry sectors by 2030.
Written by urgent.news from BusinessKorea's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
