Google CTO Visits Taiwan to Line Up Partners for TPU Mass Production

Google's push to line up Taiwanese manufacturers for its next-generation TPUs, including Zebrafish and Sunfish, points to a significant ramp-up in its proprietary AI hardware strategy. This initiative aims to reduce Google's reliance on Nvidia's GPUs by boosting its in-house ASIC production. Foxconn, already seeing its ASIC server revenue share rise above 10% last year, stands to gain substantially, with a stated goal of over 40% of the ASIC market. Compal, a newer entrant in AI servers, also secured TPU server orders in the first half of this year, which analysts expect to impact its revenue and operating profit by next year. The real story for Asia is the deepening integration of Taiwan's contract manufacturers into Google's core AI infrastructure. This is not just about server assembly; it reflects a strategic shift where Taiwanese firms are becoming critical to Google's ability to scale its AI compute. The increased capital expenditure from Alphabet, projected to be between $195 billion and $205 billion this year, underscores the scale of this commitment and the potential for long-term revenue streams for partners like Foxconn and Compal. The thing to watch is how quickly these new TPU server orders translate into tangible revenue growth for the Taiwanese partners and whether Google's TPUs can truly challenge Nvidia's dominance in the broader AI chip market. Foxconn's aggressive target of 40% of the ASIC market suggests confidence in this shift, but execution on volume and performance will be key.