This post was originally published on this site
The big chip and software product launch unveiled by Advanced Micro Devices Inc. on Wednesday represents the first real competitive threat to Nvidia Corp.’s dominance in artificial intelligence.
AMD
AMD,
unveiled a slew of products, including the official launch of its graphics processor units (GPUs) in the Instinct MI300 family for accelerating AI applications, and its software ecosystem, called ROCm, to compete with Nvidia’s
NVDA,
CUDA software platform.
No matter how dominant Nvidia remains with its popular GPUs, AMD plans to win share in the market by being the No. 2 company to attack this specific arena, a tried formula that has worked in its rivalry with Intel Corp.
INTC,
in the core processor space for both PCs and servers.
Also read: Take a bow Lisa Su, AMD’s data-center business is a true rival to Intel.
“They no longer have the empty playing field that they have had until now,” said Chirag Dekate, an analyst at Gartner, of Nvidia. “Frankly speaking, in this multi- vendor market ecosystem, you will see end users benefit, and diverse-channel partners benefit.” Nvidia will no longer be able to command the same level of high prices — and profit margins — that it has in the past for its GPUs.
“Nvidia is obviously getting a very disproportionate share of the market,” said Daniel Newman, principal analyst and founding partner of Futurum Research. “While they have been partnering with everyone, companies are struggling to get product… We need competition, we need supply.”
AMD said its new chips are available now, and the chip maker trotted out several hardware companies, such as Dell Technologies Inc.
DELL,
Lenovo
992,
and Supermicro
SMCI,
that will be offering them in their systems. Some cloud customers are also committing, including Microsoft Corp.’s
MSFT,
Azure and Oracle Corp.’s
ORCL,
Cloud business.
Analysts said that they believe AMD’s chips will be able to make traction in the AI market because they will be able to offer customers lower costs and in some cases higher performance than Nvidia right now. Dekate of Gartner noted that the MI300X chips have a much larger memory capacity, which will save customers money in many workloads.
“You can run the same size models faster, and in many cases, using less resources,” he said. “That is a big difference. That memory capacity is really important, and really a differentiator.”
AMD’s presence will not change things overnight, and the market certainly is not expecting that to happen. Nvidia’s shares fell 2.3% on Wednesday, while AMD’s shares fell 1.3%. Nvidia will also move the goal post again, Dekate noted, with its next offerings. “But today, AMD does have a compelling offering.”
