L5 · CHOKEPOINT
Data-Centre Physical
Once the power arrives, it still has to be distributed safely across a building full of servers and then removed again as heat. This layer covers the physical data-centre shell: site selection, construction, electrical distribution, and cooling. AI racks run far hotter than conventional servers, forcing a shift from air cooling to direct liquid cooling throughout the facility. Capacity is being added at historic speed, but liquid-cooling hardware, medium-voltage electrical gear and specialist commissioning teams are all supply-constrained.
WHY IT'S A CHOKEPOINT
AI rewrote the physical spec for data centres faster than the supply chain could follow. Facilities built for air-cooled racks now need liquid cooling throughout, and CDUs, cold plates, manifolds and commissioning capacity are all undersupplied. Permitting timelines add months on top.
Signals
- Direct liquid cooling grew ~156% YoY in 2025; Microsoft mandated it for all new Azure AI infrastructure (Dell'Oro; Microsoft 2025).
- Vertiv's backlog hit a record ~$15B at end-2025, up ~109% YoY, roughly a year of forward revenue (company filings).
- Schneider Electric and Eaton report record backlogs (~€25.4B and ~$22.8B), most due within a year (company filings).
- US construction needs ~349,000 more workers in 2026, adding up to ~8.5 months to data-centre builds (industry research).
The investment angle
Liquid-cooling hardware and MV electrical gear are the in-building version of the transformer bottleneck; specialist commissioning capacity is the final constraint that turns delivered kit into live compute.
Inside this layer, node by node
The atlas data behind this layer: 105 nodes, 6 of them chokepoints. Every node links back into the network map; market figures carry their source.
Acquires and permits land for data centres, advancing raw sites to shovel-ready status with secured power and fibre. Specialist land bankers and entitlement advisers earn premiums by cutting development risk; power-queue positions carry significant option value.
Owns and operates completed data centres providing space, power, cooling and connectivity across hyperscale, wholesale, retail, neocloud, REIT and edge models. They set the cost and availability of physical AI compute housing, capturing value through long-term leases and power mark-ups.
Site mobilisation through handover-ready shell and core: structural, civil, MEP rough-in and modular delivery. Construction speed limits AI data centre supply, giving general contractors and modular OEMs pricing power.
Power equipment from utility entrance to IT rack: MV/LV switchgear, transformers, UPS, batteries, busway, PDUs, gensets, emerging 800V DC. Long lead times here delay GPU cluster energisation; 12-18 month delivery is common. Siemens, Schneider, Eaton and ABB maintain margins through scarcity pricing.
Heat removal systems span air handlers, liquid cooling infrastructure, water treatment and humidity control. AI GPU clusters at 100+ kW per rack exceed air cooling limits, forcing liquid cooling adoption. CDU OEMs and coolant fluid suppliers command premium pricing amid supply shortages.
Everything installed in the data hall from floor slab to top-of-rack except IT hardware: racks, containment, cabling, and row-level power. Determines airflow, density potential, and upgrade flexibility. Fragmented market; high-density specialists command premiums.
Multi-level verification that facility systems perform to design intent under normal and failure modes, including load bank testing. Catches integration failures before live IT deployment. Commissioning agents and IST engineers are capacity-constrained and command premium rates.
Software for monitoring and managing data centre power, cooling, space, environment and assets. Excludes cloud orchestration and AI frameworks. SaaS and subscription models generate recurring revenue; integration creates high switching costs.
Operation, preventive and corrective maintenance, and lifecycle management of data centre systems, including critical spares, service contracts and staffing. Sustains uptime after handover. Critical-facilities specialists earn premiums over generalist FM; value capture splits between OEM service arms and independent TPM providers.
Controls physical access to data centres and monitors for unauthorised intrusion. Required for compliance and insurance; security downtime is extremely costly. Integrators earn recurring maintenance revenue, while AI video analytics SaaS layers subscription fees on hardware.
Detects fire at the earliest stage and suppresses it without damaging IT equipment or leaving residue. Rapid clean suppression is critical above 50kW per rack; false discharge is catastrophic. Clean-agent supply concentrates among two to three manufacturers, and 3M's Novec exit is forcing shift to FK-5-1-12 alternatives and inert gas blends.
Specialist logistics for data centre construction: heavy cranes, module transport, rigging, and on-site temporary works. Constrained crane capacity and modular build trends create bottlenecks. Heavy lift operators and precision rigging firms capture scarcity premiums in peak markets.
Technical, financial, legal, and strategic advisory across data centre project lifecycles. Complexity of power, cooling, and financing structures drives demand for sector-specific expertise. Specialist boutiques and lender's technical advisers charge above generalist consulting rates.
Standards bodies and certification schemes defining data centre technical and operational benchmarks. Compliance requirements create audit and interpretation demand. Certification auditors and standards consultants capture recurring fees from facility operators seeking market recognition.
On-site power generation physically installed at campus to bypass or supplement grid connections. Interconnection queues of 4-7 years make behind-the-meter plant attractive for speed. Genset and fuel-cell OEMs capture scarcity margins; developers gain time-to-market advantage.
Companies we track
Supply chain
Raw inputs
Key suppliers
Buyers