Data Center Physical Infrastructure Market Grows 27 Percent Y/Y to $14 Billion in 2Q 2026, According to Dell'Oro Group

Worldwide data center physical infrastructure revenue rose 27% year over year to $14 billion in 2Q 2026, extending growth above 20% as AI capacity demand continued to outpace supply. Thermal management grew more than 40%, while UPS revenue increased about 20%; AI-specialized cloud grew roughly 2x year over year. Dell’Oro identifies siting and permitting, power availability, and grid interconnections as the most serious near-term constraints, citing New York’s statewide moratorium, Pennsylvania’s permitting change, and a Texas ERCOT queue audit.

Analysis

Permitting risk changes the timing and location of capacity, not necessarily the underlying demand. The near-term market mechanism is schedule slippage: delayed site approvals and grid studies can push equipment shipments and revenue recognition across quarters, while leaving supplier backlogs intact. That distinction favors tracking orders, backlog conversion, and cancellation rates over headline growth. It also raises execution and working-capital risk if equipment is built or staged ahead of delayed projects.

The likely second-order response is geographic and design substitution: developers may shift projects toward jurisdictions with clearer approval paths, use behind-the-meter generation, or prioritize brownfield expansions. That supports power and heat-rejection solutions over a multi-year horizon, but creates uneven quarterly winners. VRT’s power/cooling exposure and planned UtilityInnovation Group deal offer potential participation; ECL’s CoolIT and SLB’s Kelvion add relevant capabilities, but the article provides no basis to size their earnings contribution. Do not treat any acquisition as proof of material near-term profit uplift.

Contrarian point: statewide restrictions are consequential but do not establish that national projects are being cancelled. Conversely, strong demand claims from a market-research release do not establish when permits, interconnections, or shipments convert to revenue. The revised customer segmentation also limits clean historical comparisons. Over 1–3 months, watch queue-study timelines, permit outcomes, and supplier order/backlog commentary; over 6–18 months, monitor whether projects migrate or disappear. The thesis weakens if cancellations rise, supplier orders roll over, or guidance cuts cite lost projects rather than timing. It strengthens if approvals clear and backlog converts without discounting.

Key Decisions for Investors

• No broad AI-infrastructure trade on this release alone. Treat it as a timing-risk signal; verify supplier order growth, backlog conversion, and cancellation disclosures before adding exposure.
• Watch VRT for a staged long on weakness only if orders/backlog remain resilient and management indicates delayed rather than lost projects. Upside is delayed shipment conversion plus potential behind-the-meter demand; downside is that permitting slippage pushes deliveries beyond the investment window or the acquisition fails to add measurable economics. Reassess on the next earnings update.
• Track ECL and SLB as secondary, unquantified beneficiaries—not standalone earnings catalysts. Require acquisition contribution, order, or guidance evidence before attributing material growth to CoolIT or Kelvion.
• Monitor state-level permitting and ERCOT large-load study progress over the next 1–3 months. A widening set of moratoria or rising cancellations would falsify the ‘delay, not cancellation’ view; approvals and interconnection progress with stable supplier orders would support it.