Nvidia's Dominance in Data Center GPU Market

TapTechNews June 11th news, according to the latest research report from semiconductor analysis company TechInsights, Nvidia has achieved explosive growth in the data center GPU shipments in 2023, with a total shipment of approximately 3.76 million units, an increase of over one million units compared to 2.64 million in 2022.

Of course, Nvidia remains an almost monopolistic overlord in the data center GPU market in 2023, and its market share is similar to that in 2022, about 98%.

Also, Nvidia also occupies 98% of the market share in terms of data center GPU revenue, with a total revenue of $36.2 billion (TapTechNews footnote: currently about 262.69 billion yuan), which is more than three times that of its 2022 revenue ($10.9 billion).

According to TechInsights' data, AMD shipped approximately 500,000 data center GPUs in 2023, while Intel shipped 400,000.

If the data of AMD and Intel are counted, the total data center GPU shipments in 2023 is 3.85 million units, an increase from 2.67 million units in 2022.

Analyst James Sanders said: Google's TPU, AMD's GPU, Intel's AI chips and even CPUs are all expected to become AI alternatives to Nvidia's GPU. Due to the growth of AI, I suspect the market will inevitably move from Nvidia to diversification.

Due to the good sales of AMD's MI300 series GPUs and the receipt of orders from companies such as Microsoft, Meta, and Oracle, it is expected that AMD's data center GPU shipments will further increase this year.

Suzanne Yang said in the earnings conference call in April: We expect the data center GPU revenue in 2024 to exceed $4 billion (currently about 29.027 billion yuan), higher than the $3.5 billion predicted in January.

In addition, after the release of the Blackwell series GPU, Nvidia also stated that it will adjust the release rhythm to once a year, and plans to launch an upgraded new product in 2025, and to launch new GPUs from the new Rubin family in 2026 and 2027.