This article first appeared on GuruFocus.
Aug 28 – Elon Musk’s latest forecast for artificial intelligence places compute power at the heart of the industry’s race forward. In a post on X, Musk said his AI startup xAI could require the equivalent of 50 million Nvidia (NASDAQ:NVDA) H100-class chips within five years, eventually reaching billions.
By framing his projection around Nvidia’s flagship accelerator, Musk anchored expectations in a familiar benchmark while leaving room for generational improvements in chip technology. Even at today’s prices of $25,000 to $40,000 per unit, that vision suggests trillions in potential spending. While actual costs will decline as hardware efficiency improves, the numbers still underline the massive infrastructure buildout required.
The forecast points to parallel growth in networking, memory, packaging, and power systems needed to support dense clusters of AI hardware. Musk’s perspective carries weight: xAI trains large language models, Tesla (NASDAQ:TSLA) pushes autonomy and robotics, and SpaceX operates at the edge of hardware and networking demands.
The bottom line is straightforward to investors. With stores of consistent accelerators and power demand, then doors and semiconductors, data centres and utilities might open, but risks are around the corner should the supply chain or energy constraints sting.
Afficher les commentaires