Report: Oracle to Deploy AI Cluster with 30,000 AMD MI355X Accelerators

Nvidia, hot off the festivities of its big GTC conference last week in San Jose, dominates the GPU market with a roughly 95 percent share. But the question remains whether and how other chip suppliers will make inroads into the GPU sector.

The top potential inroad-making candidate is AMD with its Instinct MI AI accelerators. According to a recent story from TechRadar, AMD has received a cred boost from Oracle, which let it be known during its latest quarterly earnings call that it signed a multi-billion dollar contract with AMD to build a cluster with 30,000 of AMD’s upcoming MI355X GPUs.

The chips, built on TSMC’s 3nm node and based on AMD’s CDNA architecture, is expected to ship this summer, according to TechRadar. “It will feature 288GB of HBM3E memory, bandwidth of up to 8TB/sec, and support for FP6 and FP4 low-precision computing, positioning it as a strong rival to Nvidia’s Blackwell B100 and B200,” the publication said.

The Oracle news was disclosed by Larry Ellison, chairman and CTO.

During the earnings call, he also discussed Project Stargate, the big AI data center project announced at the White House in late January by President Trump, Ellison, OpenAI CEO Sam Altman and Softbank Founder and CEO Masayoshi Son. Stargate, which could amount to a $500 billion AI infrastructure investment, will include a 64,000 GPU liquid-cooled Nvidia GB200 cluster for AI training, according to Ellison.