Advanced Micro Devices (AMD) just threw down the gauntlet in the AI arena, setting its sights directly on Nvidia, the reigning champion. With a flurry of hardware reveals, strategic partnerships, and bold predictions, AMD is making it clear they’re ready to claim a bigger share of the rapidly expanding AI market. But will the underdog be able to dethrone the leader? JPMorgan analyst Harlan Sur believes there’s significant potential in AMD’s AI arsenal but advises a cautious approach, maintaining a Neutral rating for the stock.
AMD’s AI event wasn’t just about showcasing new products – it was a declaration of intent. The company significantly revised its AI accelerator total addressable market (TAM) forecast, increasing it from $400 billion in 2027 to a whopping $500 billion by 2028. This dramatic upward revision reflects the surging demand for AI model deployments and the increasing number of applications across industries. While the TAM growth might slow down by 2028, AMD’s strong presence with its CPUs, GPUs, and DPUs positions them perfectly to capitalize on this growing wave. The recent acquisition of ZT Systems further strengthens their position, streamlining production and accelerating customer adoption – a clear signal that AMD is committed to scaling rapidly.
The event’s star attraction was undoubtedly the MI325X, AMD’s latest GPU designed to challenge Nvidia’s H200 HGX in AI training. The MI325X holds its own in training, but truly shines in AI inference, outperforming Nvidia’s H200 by a significant 1.4x. Production shipments for the MI325X are set to begin soon, and with the even more powerful MI350 slated for 2025 – promising a 1.8x boost in AI compute performance – AMD is clearly not backing down from its GPU battle with Nvidia. However, Sur points out that this race is a marathon, not a sprint. AMD is making impressive strides, but Nvidia’s lead remains substantial, and gaining significant market share will require sustained effort and time.
AMD is not neglecting its core CPU business either. The company launched its next-generation EPYC processor, codenamed “Turin,” built on its powerful Zen 5 architecture. This new processor boasts a remarkable 40% performance advantage over Intel’s Granite Rapids, further solidifying AMD’s position in the server CPU market. Sur notes that AMD’s server CPU market share surged from 31% to 34% in the first half of 2024, and Turin is poised to drive even greater gains in the coming quarters.
Beyond hardware, AMD’s progress on the software front is equally crucial. The ROCm 6.2 software stack has achieved a significant 2.4x improvement in AI inference performance – a vital factor in a market where software optimization can be a game-changer. Coupled with deep partnerships with cloud giants like Meta Platforms, Microsoft, and Oracle, AMD has secured strategic backing to compete with the industry leaders.
AMD’s AI event showcased a company ready for a fight, but unseating Nvidia won’t be easy. Sur’s price target of $180 by the end of 2025 reflects cautious optimism, acknowledging the challenges and opportunities ahead as AMD continues its aggressive push into AI accelerators and servers. Keep an eye on AMD stock as it enters its most ambitious AI campaign yet. This is a battle worth watching.
