capexScore 80/100Add to thesis
Microsoft scaling compute for frontier models by 3 additional orders of magnitude over next few years, driving exponential capex in data center infrastructure
Mustafa Suleyman· Microsoft AI· AI· 2026-05-20· about Microsoft (MSFT)
“the compute that we use to train frontier models has increased by one trillionfold. That's 12 orders of magnitude of computation in just 15 years... in the next few years, we're going to see three more orders of magnitude of compute applied to train frontier models.”
Why it matters
Explicit commitment to 1000x scaling of training compute signals massive capex acceleration for GPUs, data centers, power infrastructure, and networking. This is among the most concrete forward-looking guidance on AI capex trends.
Investment implication
Significant upside for GPU suppliers (NVIDIA, AMD), data center infrastructure providers, power delivery systems, advanced cooling solutions, and high-performance networking vendors. Power infrastructure may become a critical bottleneck.