Global technology firms are treating computing power as a strategic commodity, with hyperscalers projected to spend $765 billion on AI-related capital expenditure in 2026, according to Goldman Sachs forecasts.
Microsoft leads with an estimated $190 billion, followed by Amazon at $200 billion, Alphabet at $175–185 billion, and Meta at $125–145 billion. Combined, these companies’ 2026 CapEx represents a sharp increase from approximately $410 billion in 2025.
BlackRock CEO Larry Fink said on May 5, 2026, that surging compute demand could create a new asset class around “compute futures,” citing power, chip, and capacity shortages similar to historical commodity markets.
The 2023 NVIDIA H100 GPU shortage highlighted supply constraints, with lead times of 8–11 months and spot prices reaching $40,000 per unit, compared with a $30,000 list price.
TSMC now produces roughly 92% of the world’s advanced foundry capacity below 10 nm, concentrating global risk in Taiwan.
Market infrastructure is emerging, with cloud spot markets run by AWS, Azure, and Google Cloud, alongside public listings such as CoreWeave’s March 2025 IPO.
Experts suggest that as autonomous AI agents drive continuous 24/7 compute demand, infrastructure ownership will determine long-term value capture, particularly in emerging markets such as Pakistan.
