Nvidia CEO: $700B AI Capex 'Just the Start'
Nvidia CEO Jensen Huang stated that the tech industry's $700 billion in AI infrastructure capital expenditure is "just the start" of a much larger investment wave. The company's latest forecast points to accelerating growth as it prepares for renewed competition with rivals Intel and AMD in the AI chip market.
The spending surge is driven by hyperscale cloud providers and big tech, with companies like Microsoft, Meta, Google, and Amazon accounting for nearly 40% of Nvidia's revenue. In 2023, Microsoft and Meta alone purchased a combined $9 billion worth of H100 chips. This investment is flowing into a data center infrastructure market projected to hit $1 trillion in annual spending by 2030, a significant increase from $290 billion in 2024. Nvidia's CEO Jensen Huang argues this spending is not speculative but a rational response to monetizable demand for AI compute, with the build-out expected to continue for seven to eight years. This demand has propelled Nvidia to become the largest customer of chip manufacturing giant TSMC, surpassing Apple in 2025. The company maintains a dominant market share, controlling over 80% of the market for GPUs used in AI training. This boom isn't without its challenges, as the massive expansion of data centers strains local energy and water resources. The electricity demand from U.S. data centers alone is projected to potentially increase more than thirtyfold by 2035. Huang has stated that the global buildout required for generative AI will ultimately demand trillions of dollars in investment. Intel is positioning its Gaudi 3 chip as a more power-efficient and cost-effective alternative to Nvidia's offerings, claiming it can run AI models 1.5 times faster than the H100 GPU. The company is focusing on the market for AI inference and custom chips rather than directly competing with Nvidia in the high-end training market. Meanwhile, AMD is targeting the AI PC market with its Ryzen AI series processors and competing in the data center with its MI series of AI accelerators. The company's MI300 accelerator is projected to generate significant revenue as it aims to secure a stronger foothold in the AI hardware space.
