Four Giants Unleash AI Chip Fury as Market Heats Up

The AI chip market has heated up as four major players - Nvidia, Google, Intel, and Microsoft - announced new AI chips within a ten-day span. This sudden flurry of activity is unprecedented in the industry, with each company racing to claim its share of the lucrative AI compute market.

Nvidia's Rubin platform is at the forefront of this development, featuring six distinct chips designed to work together as an 'AI supercomputer.' The Rubin GPU itself is accompanied by a Vera CPU for host processing, NVLink 6 switch for chip-to-chip bandwidth, and other components. Nvidia claims that its Rubin samples are already being tested by select hyperscaler labs, with official availability expected in the first quarter of 2027.

Google's TPU 8i and TPU 8t chips are specifically designed for inference and training, respectively. The company is positioning these chips as key players in the 'agentic era' of AI, where autonomous agents execute multi-step workflows without human intervention. Google has a history of building TPUs for internal workloads and select cloud customers, but this split between inference and training silicon suggests a new focus on agentic workloads.

Microsoft's Maia 300 chip is the least confirmed of the four announcements, but its significance lies in Microsoft's commitment to secure manufacturing capacity from TSMC for over 300,000 units. This order volume indicates that Microsoft sees custom silicon as core infrastructure and not just an experiment. Intel's contribution to the AI chip market emphasizes power efficiency rather than raw scale.