The capital is there. The capacity to turn it into finished data centers on schedule often isn't. Nearly half of the new U.S. data centers planned for 2026 are likely to face delays or cancellations, according to a Bloomberg report, as supply chain issues and limited U.S. manufacturing capacity create a bottleneck in construction. As one infrastructure lead put it in that reporting, a single delayed piece of the supply chain can stall an entire project.
That's true whether the entity building is a hyperscaler standing up its own campus, a colocation provider leasing space to a dozen different tenants, or an operator managing both. Servers, power gear, cooling systems, and fiber all move through different vendors, different lead times, and different tiers of visibility—and when one link of the chain slips, the whole build slips with it.
What's changed for infrastructure teams
Yesterday’s data center supply chain was built for steady, forecastable demand: a handful of SKUs, predictable refresh cycles, and long planning horizons. Today’s leading edge AI-centric buildouts broke that model for everyone in the value chain. Power availability is now a sourcing constraint as much as a utility negotiation. GPU and memory allocations shift week to week. A single rack redesign can ripple back through six tiers of suppliers before anyone downstream even knows what’s happened—and for a colocation provider who has committed to a tenant's energization date, that kind of ripple can put contractual timelines at risk.
Most supply chain planning systems in this space were never built for this kind of volatility. They were built for a world where lead times moved in months, not weeks, which is exactly the gap hyperscalers, operators, and colocation providers are all running into now, just from slightly different angles.
The capability gap
Three architectural gaps show up again and again, regardless of which side of the build a company sits on:
- Fragmented visibility. Power, compute, and facilities teams plan using separate systems, so no one sees the full picture until a constraint has already caused a delay. For colocation providers juggling multiple tenants, that fragmentation complexity multiplies with every lease.
- Static forecasting. Demand signals move faster than quarterly or even monthly planning cycles can absorb, whether the signal is a hyperscaler's internal capacity plan or a colocation provider's pipeline of prospective tenants.
- Thin tier coverage. Most planning tools stop at tier-one suppliers, which is typically not where component and power constraints actually originate.
None of these are solved by adding more dashboards, reports, or alerts. They're solved by giving supply chain planning teams a common view of the network and the ability to act on it, not just observe it.
What decision-centric planning looks like
The question is how can data centers achieve a resilient multi-party supply chain during a rapid build-out phase? Here are five components for success:
1. Think of supply chains as living systems
2. Model the physical, informational and partner network
3. Ensure networked master data management can scale
4. Integrate the existing broader architecture to the network model
5. Enable a shift to automated sensing, escalations and adaptive execution with AI and ML intelligence
The critical shift is moving from visibility for its own sake to a decision-centric planning approach—the ability to seamlessly sense a disruption, evaluate the trade-offs, and act, all before the delay reaches the build. Across hyperscalers, colocation, and hybrid operating models, this means you need:
- Multi-tier network visibility (including multi-enterprise, multi-party, multi-mode, multi-legs), so a shortage at a sub-tier power component supplier surfaces before it becomes a missed energization date or an at-risk tenant commitment.
- Connected planning across functions, so procurement, construction, facilities—and, for colocation providers, leasing—can work from the same demand signals instead of reconciling after the fact.
- AI-driven scenario evaluation, so teams can weigh sourcing alternatives in hours, not weeks, whenever a supplier slips.
- A common data layer, so the answer doesn't depend on which system, or which tenant contract, someone happens to be looking at that day.
Where this plays out first
High-tech supply chains have lived with this kind of volatility before, just at a smaller scale—semiconductor manufacturers and OEMs have spent years managing cyclical demand swings and multi-tier component shortages. Hyperscalers, data center operators, and colocation providers are now facing the same problem at hyperscale, compounded by power constraints that don't show up in a traditional bill of materials at all.
What's next
As delays keep stacking up across the industry, the hyperscalers, operators, and colocation providers who adopt a decision-centric planning approach now will be the ones still hitting energization dates and honoring tenant commitments two years from now, when lead times are even less forgiving. The Supply Chain Command Center is where that capability starts to take shape—providing a single layer of visibility and control across your multi-tier data center supply network.
