- AMD is targeted at Nvidias Blackwell with the upcoming instinct MI355X accelerator
- Oracle is planning massive 30,000-device MI355X cluster for AI working loading with high performance AI
- It’s beyond Stargate, Oracle’s 64,000 GPU NVIDIA GB200 Cluster
While AI Darling Nvidia continues to dominate the AI Accelerator market, AMD hopes with a share of over 90%, its closest rival, AMD, to challenge the Blackwell lineup with its new instinct MI355X series of GPUs.
MI355X, now expected to arrive in mid-2025, is manufactured on TSMC’s 3NM node and built on AMD’s new CDNA 4 architecture. It will contain 288 GB HBM3E memory, bandwidth of up to 8 TB/SEK, and support for FP6 and FP4 Low-precision computing that places it as a strong rival to NVIDIA’s Blackwell B100 and B200.
In 2024, we reported a number of major wins for AMD that included shipping thousands of its MI300X AI accelerators for Vultr, a leading privately owned cloud computing platform and for Oracle. The latter has now announced plans to build a cluster of 30,000 MI355X AI accelerators.
Stargate
This latest news was revealed during Oracle’s recent Q2 2025 earnings call, with Larry Ellison, chairman and heading technology officer, said to investors, “In the 3rd quarter we signed a contract of several billion dollars with AMD to build a cluster of 30,000 of their latest MI355X GPUs.”
Although he did not go into it, Ellison talked about Project Stargate and said, “We are building a gigantic 64,000 GPU fluid-cooled NVIDIA GB200 cluster for AI training.”
Later, he added, “Stargate seems to be the biggest AI training project out there, and we expect it to allow us to grow our RPO even higher in the coming quarters. And we expect our first big Stargate contract to be pretty soon.”
When asked further about Stargate by a Deutsche Bank analyst, Ellison gave an answer that could just as easily apply to the cluster of the MI355X AI Accelerators Oracle plans to build.
“The capacity we have is to build these huge AI clusters with technology that actually runs faster and more financially than our competitors. So it really is a technology advantage we have over them. If you run faster and you pay with the hour, you cost less. So that technology advantage is translated into an economic advantage that allows us to win a lot of these huge deals,” he said.
Ellison also affected Oracle’s data center strategy and said, “So we can start our data centers less than our competitors and then we grow based on demand. Building these data centers is expensive and they are really expensive if they are not full or at least half full. So we tend to start small and then add capacity as demand arises.”
