Sources: Meta reviews some of its 11 in-development data center projects globally and lays off a Danish contractor, as part of a design change for AI workloads
Sebastian Moss / DCD :
Context & Ripple Effects
This December 2022 report was the first public signal of what became a costly pivot: weeks later Meta booked a $4.2B Q4 restructuring charge that explicitly included data center redesign costs, confirming the review of its 11 in-development projects had real balance-sheet consequences.
The arc since has only sharpened the stakes — by 2026, sources describe sweeping Meta layoffs of 20% or more of staff driven partly by mounting AI infrastructure costs, meaning the facility-design question raised here grew into a company-wide cost problem.
First-order effects
- The Danish contractor loses its engagement immediately, and any of the 11 reviewed projects still in early construction faces scope changes or pauses while Meta re-specs facilities for AI workloads.
Second-order effects
- Data center builders and equipment suppliers working with Meta absorb schedule volatility, and rival hyperscalers watching this review face pressure to audit their own pipelines before committing to pre-AI designs.
Third-order effects
- If AI workload requirements keep forcing mid-build redesigns, hyperscale construction becomes a recurring-restructuring business where contractors price in cancellation risk, and capital shifts toward flexible, retrofit-friendly facility designs.
The trend: Hyperscalers are redesigning data center fleets around AI workloads, and construction partners — not just employees — are bearing the resulting volatility.