NVIDIA’s 2025 Blackwell Ultra chip marks a major leap forward in the AI market, far surpassing its predecessor in performance. The chip delivers exceptional computational capacity for AI tasks: it operates 7.5 times faster than the previous generation and is equipped with significantly larger memory, enabling the training and deployment of the largest and most complex AI models. This technology is particularly crucial for so-called AI factories – large-scale, high-performance computing centres that can build, train, and run AI models continuously at scale.
Demand for NVIDIA chips remains extremely strong. By the first quarter of 2025, the company’s share of the AI accelerator market had reached 92–94 per cent. According to CEO Jensen Huang, the global AI infrastructure market could be worth as much as 3–4 trillion dollars over the next five years. The new Blackwell platform enables AI training at four times the previous speed and inference thirty times faster, while using twenty-five times less energy. This dramatically reduces both the costs and environmental footprint of running AI systems. A key factor behind NVIDIA’s dominance is the lock-in effect: developers have already invested substantial time and money into mastering the CUDA software ecosystem, making them less inclined to switch to alternative platforms.
Meanwhile, technology giants such as Google, Amazon, Microsoft, and Meta are developing their own AI chips to reduce reliance on NVIDIA’s products, which could pose competitive challenges in the longer term. At the same time, NVIDIA is expanding its global presence, including a £11 billion investment to establish Europe’s largest computing centre in the United Kingdom. Geopolitical factors – especially technological tensions and export restrictions between the US and China – continue to pose obstacles. Nevertheless, the Blackwell Ultra consolidates NVIDIA’s leadership in AI infrastructure and is expected to have a profound impact across industries, from scientific research to creative sectors.
Sources:


