AI Chip War Heats Up as Rubin, Helios, and Maia 300 Battle for Market Supremacy
Nvidia's Rubin platform has entered lab samples at hyperscalers, while AMD's Helios rack-scale system is shipping in volume to cloud providers. Microsoft is preparing to unveil its next-generation Maia 300 chip as soon as next month.
The three-way scramble is reshaping how much compute costs and who controls the supply chain. Nvidia's near-total grip on AI training may not survive another two years, according to industry analysts.
Nvidia's Rubin platform pairs a new Arm-based CPU with a next-generation GPU inside Nvidia's NVLink 6 interconnect fabric. The Vera Rubin NVL144 rack is expected to deliver three to four times the FLOPs per chip of Blackwell Ultra, driven by streaming multiprocessor changes and a new interconnect topology.
AMD's Helios system combines sixth-generation EPYC 9006 series CPUs with the Instinct MI455X GPU, Pensando networking silicon, and the ROCm software stack. AMD is betting that hyperscalers will trade some software maturity for lower per-token cost and supply diversification away from a single vendor.