Nvidia shipped 3.76 million data center GPU units in 2023 according to a study from semiconductor analyst house TechInsights, representing a year-on-year increase of over 1 million units shipped and a growth in sales of 42 percent when compared to 2022.

As reported by HPCwire, the study also showed that Nvidia retained a 98 percent market share of the data center GPU market last year, a figure that continued the dominance displayed by the company in the year prior.

– Sebastian Moss

As a result, Nvidia saw its revenue grow by 126 percent in 2023, hitting $60.9 billion. Earlier this month, the company’s market cap hit $3 trillion, making it the third company after Apple and Microsoft to ever receive such a valuation.

When AMD and Intel were factored into the study, total GPU shipments in 2023 totaled 3.85 million, up from 2.67 million units in 2022.

However, Nvidia's rivals are also expected to see a growth in shipments this year, with AMD having already reported strong sales of its MI300-series GPUs. Speaking after the company reported its Q1 2024 financials, CEO Lisa Su told analysts AMD expected data center GPU revenue to exceed $4 billion in 2024.

Furthermore, the study also noted that traditional semiconductor companies are starting to face competition from hyperscalers such as Microsoft and Google, which have started developing their own homegrown chips.

Microsoft’s in-house AI accelerator is called the Azure Maia 100 and is used in the company’s cloud data centers. The company has also developed its own Arm-based Azure Cobalt 100 CPUs, designed to support general workloads, with a focus on performance per watt.

Meanwhile, Google unveiled the sixth generation of its custom Tensor Processing Unit (TPU) AI chip dubbed Trillium at the company’s annual I/O developer conference last month. The latest iteration of the TPU has been specifically designed to support large language and recommender models, including Google’s Gemini, Imagen, and Gemma, with reduced latency and lower costs.