AI data center rollouts face project delays
The massive capital expenditure build-out for AI infrastructure is highly likely to experience budget overruns and timeline delays based on mega-project base rates.
The argument
The guest cited Bent Flyvbjerg's database of 16,000 mega-projects, noting that less than 9% are completed on time and on budget, and only 0.5% deliver on their initial promises. While modularity helps data center construction, real-world constraints like permitting, energy grid capacity, and cooling are already causing delays, with estimates suggesting 30% to 50% of AI data centers could face delays by 2026.
The thesis, stress-tested
✓ What validates it
- ✓Industry reports confirming 30% to 50% of planned data centers in 2026 are delayed
- ✓Hyperscalers revising capital expenditure guidance upward due to cost overruns
▸ Risks discussed
- ▸Increased modularization of data center components could streamline construction faster than historical infrastructure projects
- ▸Hyperscalers have unprecedented capital resources to brute-force supply chain bottlenecks
Hear it yourself
"Again, I don't know if this is right or wrong, but it said about 25% of AI data centers in 2025 were delayed for some reason, and the estimate for 2026 is somewhere around 30 to 50%."
00:00 / 00:12
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE