Liftr Insights, a pioneer in market intelligence driven by unique data, caught the first appearance of the MI300 that places AMD alongside NVIDIA GPUs in the public cloud.
While the data show that NVIDIA holds the dominant lead, AMD is positioned to capture revenue in markets where NVIDIA can not fulfill demand. This has effects on both companies’ ability to command price and placement while shaping the architecture underpinning Artificial Intelligence. Like the NVIDIA Hopper series, the AMD Instinct series are essential components to support increasing demand for AI accelerated instance types.
Liftr® tracks data for nine cloud providers, which represent over 75% of the public cloud space. Liftr data show the appearance of the MI300 in three countries and, starting out, it represents 0.3% of the accelerator market, but that will certainly grow.
Liftr also tracks specialty cloud providers CoreWeave and Lambda. This is notable because these smaller providers offered the NVIDIA H100 GPUs prior to the major cloud providers. While the MI300x was first expected in Azure, it’s helpful to see what effect it will have on other cloud provider offerings.
“The expansion rate over the next six to eight months reveals a lot about the adoption and success,” says Tab Schadt, CEO of Liftr Insights. “The data will provide an objective means to measure overall demand and see if there is appetite for Intel’s accelerator.”
Liftr has yet to see any instances of the latest Intel Gaudi GPU appear in any of the nine clouds. There is also the impact of home-grown accelerators, such as AWS’s Inferentia GPUs. As of the end of 2023, AWS and GCP accelerated instances represented 9.6% of the market, which is still dominated by NVIDIA.
“Liftr provides a view of the entire market which is more than NVIDIA, AMD, and Intel,” says Schadt. “That completeness helps market intelligence analysts get the full view—on a regular cadence—that they need to do their job well.”
For more such updates, follow us on Google News Martech News