Advanced microdevices (AMD 2.55%) establishes itself as the No. 2 player in the artificial intelligence (AI) accelerator market, even if it remains far behind Nvidia. While Nvidia now generates $30 billion per quarter in its data center segment, AMD said it expects to produce about $5 billion from AI accelerator sales in all of 2024.
With demand still booming for AI accelerators and AMD’s low market share, the company should have plenty of room to quickly expand its AI business. However, 2025 could be a far from bright year for AMD if new predictions from a Wall Street analyst are to be believed.
A potential slowdown in growth
A Wolfe Research analyst on Thursday lowered AMD’s stock rating to the equivalent of “hold” and removed its price target entirely due to concerns about growth in graphics processing units (GPUs) at computer centers. data. The analyst pointed to plans drawn up by original design manufacturers to justify the increased pessimism.
Analyst Wolfe estimates AMD’s data center GPU revenue at $1.5 billion to $2 billion for the fourth quarter of 2024. For 2025, the forecast has been lowered to just $7 billion, although below the $10 billion generally expected by analysts. This annual forecast essentially predicts that AMD’s data center GPU revenue won’t increase much quarter-over-quarter this year.
Of course, any analyst opinion should be taken with a grain of salt. We are less than three weeks into the new year and the AI industry is evolving extremely quickly. AMD has compelling AI accelerators on the market, and they are clearly convincing customers. Oracle uses AMD’s MI300X GPUs for its cloud data centers, and Vultr, a smaller cloud provider, chose the same AMD GPU to build its AI offerings.
One obstacle that AMD and all other non-Nvidia competitors face is the maturity of Nvidia’s software ecosystem. Nvidia GPUs support a proprietary software layer called CUDA, which has been around for a long time and is widely used in industry and academia for accelerated computing tasks. Even with capable hardware from AMD, the path of least resistance for those building AI clusters remains Nvidia.
Software is a big reason Intels efforts to enter the AI accelerator market have largely failed so far. Intel has an added complication in that its Gaudi AI chips are not traditional GPUs, so the architecture also acts as a barrier. AMD is doing much better than Intel, but it is still far behind Nvidia.
New AI chips could help
The software situation is expected to improve over time and AMD’s new chips could help boost sales this year if they are well received by the market. The MI325X is expected to be widely available from system makers in the first quarter, although this chip uses the same architecture as the previous MI300X.
The next-generation Instinct MI350 family is expected to launch sometime in 2025, likely later in the year, and AMD is promising big performance improvements. Based on a new architecture, AMD claims the MI350 family delivers up to 35x better AI inference performance than the MI300 family. The MI400 family will then arrive in 2026, built on another all-new architecture.
Even with a rapid pace of product launches, AMD faces an equally rapid Nvidia and will continue to struggle with its software disadvantage. Although AMD’s AI chip sales this year are still to be seen, any shortfall compared to expectations could send the stock lower.
Timothy Green holds positions at Intel. The Motley Fool holds positions and recommends Advanced Micro Devices, Intel, Nvidia and Oracle. The Motley Fool recommends the following options: Short February 2025 $27 calls on Intel. The Motley Fool has a disclosure policy.
