A new CNAS report warns semiconductor manufacturing capacity cannot keep pace with AI demand. Microsoft, Amazon, Alphabet, Meta, and Oracle could collectively spend $700B+ on CapEx in 2026.
Key facts
- Hyperscalers could spend $700B+ on CapEx in 2026
- CNAS says chip production is 'binding constraint' on AI
- Shift from power shortage to silicon shortage in 2025-2026
- Silicon is short-term; power is long-term constraint
- Demand outpacing chip manufacturers' forecasts
The AI industry spent the last two years worrying about running out of electricity. Now a new report from the Center for a New American Security (CNAS) argues the next bottleneck is silicon — chip manufacturing, HBM memory, and advanced packaging are all falling behind hyperscale demand.
“The world’s leading AI companies cannot get enough chips,” the report states, describing AI chip production as a “binding constraint on the pace of the AI compute buildout.” The shift marks a reversal from 2024 and early 2025, when operators like Satya Nadella described holding GPUs they could not plug in due to power shortages [According to Data Center Knowledge].
The Two-Timeline Problem
Stephen Sopko, semiconductor analyst at HyperFrame Research, frames the issue as two physical problems on different timelines. “Silicon is the binding short-term constraint. Power is the binding long-term constraint,” Sopko told Data Center Knowledge. Power projects take years to decades; chip fabs expand faster but still cannot match hyperscale demand.
The report argues AI compute demand is now “outpacing many chip manufacturers’ forecasts.” The supply chain spans advanced logic, high-bandwidth memory (HBM), networking silicon, and packaging — all of which must scale together. A single weak link strands the rest [According to the CNAS report].
$700B Meets Reality
[Microsoft](slug: microsoft), [Amazon](slug: amazon), [Meta](slug: meta), [Alphabet](slug: alphabet), and [Oracle](slug: oracle) could collectively spend $700B or more on capital expenditures in 2026, with most tied to AI systems and data centers. That spending surge is colliding with a semiconductor supply chain that cannot expand at the same pace. The CNAS report and recent earnings commentary from TSMC, Micron, SK hynix, Nvidia, and Broadcom converge on the same message: chip supply is the new binding constraint.
What to watch
Watch TSMC and Micron earnings for forward guidance on capacity expansion timelines. Also track hyperscaler CapEx disclosures in Q3 2026 — if spending continues rising without matching chip supply growth, the silicon wall will tighten further.
