Nvidia leads AI chip revenue as rivals trail in reported market figures

Nvidia accounted for an estimated 76 per cent of global server GPU revenue in the fourth quarter of 2025, according to figures shared by The Kobeissi Letter, highlighting the company’s strong position in the market for chips used in artificial intelligence infrastructure.
The figures put Broadcom in second place with 10 per cent of revenue, followed by AMD with 3 per cent.
The post said Nvidia’s share was more than three times the combined revenue attributed to its competitors. However, it did not identify the research provider or explain the methodology behind the estimates.
The comparison also requires some context. Nvidia and AMD sell GPUs used in data centres, while Broadcom is a major supplier of custom application-specific integrated circuits (ASICs) and networking technology for AI infrastructure. These products serve related markets but are not all conventional GPUs.
Nvidia’s position has been built through years of investment in graphics processing technology, software and computing platforms beyond its traditional gaming business. Its CUDA software ecosystem has also helped establish its products across AI development and data centre applications.
The company’s financial results reflect the scale of its data centre business. Nvidia reported US$193.7 billion in data centre revenue for its fiscal year 2026, compared with US$115.2 billion in fiscal 2025.
Competition is continuing, with AMD developing its Instinct accelerator range and major technology companies investing in custom chips to meet their own AI computing requirements.
While the reported figures point to Nvidia’s strong position, the lack of a disclosed source and methodology means the precise market shares should be treated with caution.