Nvidia’s Lead
Nvidia is known for graphics processing units originally built for video games, technology that has become the bedrock of data centers because it can run massive calculations simultaneously to train and run AI models. Its most advanced chips, such as Blackwell and Rubin, are banned from export to China. In September, Nvidia also entered the PC market with its RTX Spark chip.
- Industry estimates put Nvidia’s share of high-end data center chips for AI training and inference at around 80%.
- In Singapore, Nvidia chips are used by Singtel, government agency HTX and data center operator Equinix.
- Singapore’s national supercomputer Aspire 2B at Nanyang Technological University runs on more than 1,500 H200 GPUs and serves over 9,000 researchers.
- Its proprietary CUDA software platform spares developers from coding complex functions from scratch, but also makes it hard for companies to switch to rivals’ hardware.
According to Su Lian Jye, chief analyst at Omdia, Nvidia’s strength comes from a comprehensive portfolio serving data centers, high-end PCs, medical devices, humanoid robots and self-driving cars.
Rivals Pick Up Speed
Nvidia faces competition on several fronts:

















