Nvidia’s AI dominance challenged by AMD and Custom Silicon

Nvidia’s AI dominance challenged by AMD and Custom Silicon is profiled by BTW Media because published evidence links it to internet infrastructure, governance, operational dependencies, or market visibility.

• Nvidia’s Hopper GPU shipments triple, but AMD gains significant market share.
• Cloud providers and hyperscalers increasingly deploy custom AI silicon.

What happened: AMD gains traction as cloud giants embrace custom AI silicon

Nvidia retained its leadership in AI hardware throughout 2024, with shipments of its Hopper GPUs surpassing two million units among its top 12 customers. This milestone underscores Nvidia’s dominance in the AI acceleration space. However, AMD is quickly gaining traction, fueled by the success of its Instinct MI300 series GPUs.

Early adopters like Microsoft and Meta have been instrumental in AMD’s growth. Omdia reports that Microsoft procured 581,000 GPUs in 2024, with 1 in 6 being AMD-built, while Meta relied on AMD for an impressive 43% of its GPU shipments.

Despite these gains, Nvidia remains the dominant player, holding a substantial lead in overall market share. Nevertheless, AMD’s progress signals increasing competition in the AI hardware landscape. Additionally, the rise of custom AI silicon from cloud providers and hyperscalers is emerging as a significant challenge to Nvidia’s market position, marking a shift in how AI computing power is developed and deployed.

Also read: Nvidia boosts Jetson Orin Nano dev kits ahead of the holidays
Also read: Nvidia under fire: Anti-monopoly probe heats up in China

Why it is important

The competition between Nvidia, AMD, and custom AI silicon is pivotal for the future of AI infrastructure. Nvidia’s dominance has been a driving force in the AI market, but the emergence of AMD’s MI300X accelerators, which offer higher floating-point performance and memory bandwidth, signals a shift.

These advancements are particularly attractive for inference workloads, where memory capacity and speed are critical. The increasing deployment of custom AI silicon by major cloud providers indicates a diversification in the AI hardware market, potentially leading to more specialized and efficient solutions.

This competition is crucial as it pushes for innovation, potentially reducing costs, and could lead to more accessible AI technology. The market dynamics also highlight the importance of having alternative suppliers, which can influence pricing, performance, and the overall evolution of AI applications.