What should buyers normalize before comparing data center costs?
Normalize the definition of MW before comparing any cost number. A cost-per-MW benchmark can refer to critical IT load, total facility power, shell cost, powered shell cost, full AI fit-out, or a project budget that includes land, utility interconnection, soft costs, and contingency. Buyers should force every estimate into the same scope before comparing build, lease, or powered-shell options.
AI capacity adds extra pressure because rack density, liquid cooling, electrical distribution, commissioning, and utility timing can change the budget more than the building shell itself. A low cost per MW is not useful if it excludes the substation, transformer lead time, cooling plant, land, design fees, or the owner-furnished equipment required to make the site usable.
| Cost layer | Normalize before comparing | Buyer caveat |
|---|---|---|
| Shell and building | Building, yard, security, base MEP scope, and delivery standard | A shell quote may exclude the AI fit-out that drives the actual budget |
| Utility and interconnection | Substation, transformer, switchgear, utility upgrades, energization schedule | Time to power can matter more than headline cost per MW |
| AI fit-out | High-density electrical distribution, liquid cooling, containment, and controls | GPU-ready density can move cost above conventional enterprise assumptions |
| Land and permitting | Site cost, due diligence, entitlement, environmental review, and local fees | Markets with lower land cost can still lose if power or permitting stalls |
| Soft costs and contingency | Design, commissioning, owner representation, insurance, financing, and reserves | Exclusions can make two estimates appear comparable when they are not |
How do budget layers change the build-versus-lease decision?
The build-versus-lease question should be answered after the buyer separates capital cost, operating cost, time to power, utilization, and exit risk. A build can look attractive on long-term unit economics but fail if power delivery, permitting, or commissioning pushes the workload past the business deadline. A lease can look expensive on rate alone but still win when it gives the buyer a faster path to revenue, shorter commitment, or lower operating burden.
Treat lease economics and construction budgets as different risk bundles. Lease pricing often wraps facility capital, power infrastructure, operations, redundancy, and provider margin into a recurring payment. Building exposes the buyer directly to land, utility, labor, materials, cooling, design, contingency, and financing risk. Powered shell and build-to-suit structures sit between those endpoints.
| Decision factor | Build tends to fit when | Lease tends to fit when |
|---|---|---|
| Load durability | Demand is predictable over a long horizon | Demand is uncertain, bursty, or tied to fast-changing GPU cycles |
| Time to power | Utility and permitting milestones are credible | Business timing needs existing or near-term capacity |
| Operating capability | The buyer can manage facility operations or hire the right team | The buyer wants provider operations, remote hands, and facility SLAs |
| Capital strategy | Balance sheet and governance support owned infrastructure | The buyer prefers operating flexibility or shorter commitments |
| Exit risk | The site has durable strategic value | The workload, hardware, or region may change before full payback |
When do incentives, power contracts, and energy costs matter most?
Incentives matter when they change the net cost of construction, equipment, sales tax, property tax, or job commitments enough to affect site selection. They should not be treated as guaranteed savings until the buyer has checked eligibility, capital thresholds, employment requirements, clawbacks, local approvals, and timing. Incentives can improve a project, but they rarely fix a weak power or permitting case by themselves.
Power contracts and energy costs matter throughout the life of the deployment. Utility tariffs, demand charges, PPA terms, basis risk, renewable energy credit treatment, backup generation, and load factor can change the true cost of AI capacity after the construction budget is complete. Buyers should model energy cost, escalation, and reliability separately from the one-time development budget.
| Finance area | Why it matters | What to verify |
|---|---|---|
| Tax incentives | Can reduce eligible project cost or operating taxes | Eligibility, approval path, clawbacks, term, and reporting burden |
| Utility tariff | Sets recurring power cost and demand-charge exposure | Rate class, riders, demand charges, standby charges, and escalation |
| PPA or clean-energy contract | Can support sustainability goals and price planning | Strike price, shape risk, basis risk, REC ownership, and settlement terms |
| Backup and resilience | Adds capital and operating cost beyond base utility service | Generator, UPS, battery, fuel, runtime, testing, permits, and emissions constraints |
| Financing and contingency | Determines whether cost overruns can be absorbed | Interest, draw schedule, owner contingency, change orders, and schedule risk |
Which finance tools should buyers use next?
Use the construction cost guide when the first question is the benchmark: what is included in a cost-per-MW estimate, how AI fit-out changes the budget, and which exclusions can distort comparisons. Use the build-versus-lease calculator when the buyer needs to compare owned infrastructure against leased capacity across NPV, time to power, utilization, and exit risk.
Use the cost-per-MW calculator to test scenarios before asking providers for quotes. Use the tax incentive estimator when states, abatements, sales tax exemptions, or jobs commitments could affect site selection. Use the PPA-versus-utility comparator when recurring energy economics, term length, renewable claims, and price risk could change the total cost of the deployment.