As for market-leader Nvidia, it too is readying a next-generation AI GPU, the Blackwell Ultra, set to debut in the second half of this year. As a leader in AI-specific chips, Nvidia is greatly positioned to extend its lead on the competition, especially AMD. Industry pundits are predicting very different outcomes for Nvidia and AMD in terms of performance. This would be the case with the introduction of the upcoming Blackwell and Blackwell Ultra GPUs.
So far in 2024, Nvidia sold every one of its 1.3 million Hopper GPUs completely out to the big hyperscalers, including Amazon, Microsoft, Alphabet/Google and Oracle. The firm has realized an astounding commercial breakthrough by selling 3.6 million of its Blackwell GPUs to numerous AI companies. This success is especially impressive given that the product only formally launched last autumn. 2023’s Hopper line, announced back in March, has set the stage for Nvidia’s domination of the AI GPU market.
Nvidia’s Roadmap for AI Dominance
Nvidia’s overall strategy right now is to continue extending its lead in AI-specific chips, as seen with the forthcoming Blackwell Ultra. Assuming these new GPUs outperform their predecessors by the same margin that past generations have, this new release will continue to entrench Nvidia’s lead in the market. Intel’s aggressive company roadmap promises a strong, supportable way forward to stay ahead of AMD and other competition.
“Our proprietary benchmarking report suggests real-world throughput of NVDA’s H200 across a range of open-source models is substantially higher than AMD’s MI300x, despite MI300x’s higher advertised TFLOPs and memory bandwidth,” – Blayne Curtis
Industry analysts have noted that while AMD’s MI300x boasts impressive specifications, Nvidia’s H200 outperforms in real-world applications. This performance advantage will only increase with the upcoming launch of our Blackwell and Blackwell Ultra GPUs.
Hyperscalers and Nvidia’s Market Reach
The expected sale of 1.3 million Hopper GPUs in 2024 further proves Nvidia’s dominating footing with the top hyperscalers. These companies, including Amazon, Microsoft, Alphabet, and Oracle, rely heavily on advanced AI capabilities for their cloud services and data centers. Nvidia’s prowess in producing leading-edge technology positions it as the obvious partner of choice for these developer behemoths.
That indicates an aggressive rollout of the Blackwell GPUs by the very same firms, an indication of their confidence in the robustness of Nvidia’s offerings. Blackwell GPUs Blackwell has only been on the market for less than a year, but already they are jumping off the shelves. After all, they’ve sold 3.6 million of them already! However, success has put the spotlight on Nvidia’s extraordinary ability to make good on unparalleled demand for high-performance AI solutions.
Widening the Performance Gap with AMD
Nvidia has left AMD in the dust when it comes to producing chips specifically designed for AI. The very new Blackwell Ultra, which drops on Sept. 2, is poised to further exacerbate this gap. The speed and performance among the two companies’ respective offerings have created a great deal of debate and confusion among industry experts.
“A valid pushback would be that the AMD solution is not as well optimized for the models we tested, but we view this as exactly the point,” – Blayne Curtis
Blayne Curtis emphasizes that while AMD may not be as well-optimized for certain models, this serves to highlight Nvidia’s strengths in delivering superior performance across a range of applications. With the Blackwell Ultra expected to be their next move, Nvidia is primed to further widen the gap in the race for AI GPUs.