The Contract Stack & Commercial/Legal Framework

Everything so far in Part 2 has been about getting the project built: the schedule (Chapter 2.1), the owner's organization and delivery model (Chapter 2.2) and the long-lead supply chain (Chapter 2.3). What binds all of it together, and decides who pays when something goes wrong, is the contract stack. An AI data center is procured under a dozen agreements, not one master: an interconnection agreement with the utility or ISO, a power-purchase agreement, an engineer-procure-construct (EPC) contract for the shell and power chain, equipment supply agreements and slot reservations for long-lead gear, a colocation or master-services agreement if you rent space, operations-and-maintenance contracts, and GPU-cloud service-level agreements if you sell compute. Each one allocates a slice of risk. The project's commercial integrity is the sum of those allocations and, critically, the seams between them.

The chapter maps the stack, traces how schedule, performance and force-majeure risk move through it, and shows what makes those allocations enforceable. It then covers equipment-supply and slot-reservation contracts, the grid-side agreements, and the tenant-facing terms that turn a built facility into a revenue contract. One principle runs through all of it: risk does not disappear when you sign it away. It moves to whoever the contract names, and the expensive mistakes land it on a party that cannot control the underlying event.

Picture the stack as a column from the grid down to the customer. At the top sit the interconnection agreement (the utility's large-load service agreement for load; a generator LGIA or ISA only where generation interconnection is in scope) and the power-purchase agreement. They define the permitted electrical service and price the contracted energy, with any remaining delivery and transmission arrangements identified, and they are the two hardest to unwind once signed. Below them, the EPC contract (or the set of trade packages that replaces it) builds the physical asset. It is fed by equipment supply agreements and slot-reservation contracts for the transformers, switchgear, chillers and gensets that have lead times of months to years. Wood Mackenzie reported US substation-transformer lead times above 160 weeks in May 2026, which is why the slot, not the price, is what the reservation contract buys.

If the developer is not the end user, a colocation agreement or master-services agreement (MSA) sits between the facility owner and the tenant. Once the facility is live, O&M contracts govern operations and a GPU-cloud SLA governs what the compute customer is owed. Each layer is a separate counterparty, a separate negotiation and a separate risk-allocation regime.

Contract interfaces matter more for an AI campus than for a legacy data center because of concurrency and the seam count. A traditional facility could be built in sequence: secure power, build, fit out, lease. An AI build runs all of these in parallel under a speed-to-power clock. The interconnection study is still open while the EPC mobilizes, the GPUs are being allocated and the anchor tenant is in diligence. Contracts are therefore signed before their dependencies are resolved, so the interfaces between them carry far more risk than the body of any single agreement. Those interfaces are the handoff dates, the owner-furnished-equipment delivery commitments and the conditions precedent. The recurring failure is not a bad clause but a gap between two contracts that no clause covers. Chapter 3.2 tests grid access; this chapter assigns the resulting obligations. In a co-location structure, a generator interconnection agreement alone does not establish the campus's accepted load-service arrangement.

The principle of risk allocation is older than data centers and rarely followed: assign each risk to the party best able to control or absorb it. Schedule risk on the concrete belongs to the EPC. Schedule risk on the GPU allocation belongs to whoever holds the OEM relationship. Schedule risk on energization belongs to nobody the contract can name, because it sits with the utility and the queue. The art is matching allocation to control. The recurring error is giving a risk to a party with no lever over the event, who then either prices it punitively or defaults on it, and either way the owner pays.

Schedule risk dominates because the speed-to-power clock dominates the economics. It splits three ways: construction schedule (the EPC's, capped by delay LDs), equipment-delivery schedule (the OEM's, often weakly capped or excused) and energization schedule (the utility's, almost never backed by enforceable damages). The three are serially dependent: a slipped transformer delays energization, which strands a finished building full of depreciating GPUs. Yet the contracts that govern them are independent, and the owner, who suffers the loss, holds recourse against none of the others for the full amount.

Performance risk asks whether the facility delivers the contracted power, PUE and availability at completion. It belongs to the EPC, backed by performance guarantees and performance LDs, and it is the cleaner allocation because the contractor controls the design and build. The complication is the owner-furnished-equipment line. When the owner furnishes the GPUs, the CDUs or the switchgear, the EPC's performance guarantee carves them out, and a facility that underperforms because the owner's chillers cannot reject the heat is the owner's problem. The guarantee is only as good as the equipment it covers.

Force-majeure risk is where 2026 has rewritten the playbook. A force-majeure clause excuses performance for events outside a party's control, and what counts as 'outside control' has become the most heavily negotiated definition in the stack. A factory fire, an HBM allocation shortfall, a tariff or export-control change and a grid-upgrade delay have different contractual triggers. Define which causes qualify for time relief, which allow price adjustment and which stay with the supplier, and require notice and mitigation evidence. Foreseeability follows the agreed clause and governing law. Contractor relief does not decide insurance recovery: delay-in-startup (DSU) insurance requires an insured physical-damage trigger under the placed policy, including any required supplier or utility extension. DSU does not cover a schedule miss without that trigger, so fund the loss that remains after contractual recoveries. See Chapter 2.6.

Risk allocation is a promise. Liquidated damages, guarantees, warranties and bonds make the promise collectable. Liquidated damages (LDs) are a pre-agreed sum payable per day of delay (delay LDs) or per unit of performance shortfall (performance LDs), set at signing so the owner need not prove actual loss in court. Both kinds are capped, and they need separate caps and triggers. As a renewable-EPC comparator, a 2025 solar-and-biogas review put performance-LD caps at 10-15% of contract price in most projects, across a 5-25% range (Husch Blackwell, May 9, 2025). The number that matters is the gap between that cap and your real loss.

A one-month energization slip on a 100 MW AI campus strands a building whose GPUs are depreciating at a multiple of any LD a contractor will accept. The LD cap recovers a fraction of the loss and the residual sits on the owner, by design. LDs are a risk-sharing instrument, not a make-whole one: the contractor will not price unlimited delay exposure, and if you force it, the bid price absorbs the risk premium anyway. Treat the LD cap as the contractor's contribution, then own the residual explicitly, through schedule buffer, delay-in-startup insurance (Chapter 2.6) and a financing structure that survives a slip (Chapter 2.5). Compare recoveries with retained loss, and never collect twice for the same damage.

Read the consequential-damages waiver and the delay-remedy clause together. AIA A201-2017 section 15.1.7 waives the owner's rental expenses and losses of use, income, profit and financing, while preserving applicable liquidated damages. Negotiate explicit exceptions for the delay-loss categories you intend to claim. If LDs are also the sole and exclusive remedy for delay, model their cap as the maximum contractual delay recovery; the waiver alone does not remove direct-damage remedies.

Performance guarantees are the technical promises the LDs enforce: a guaranteed PUE at a defined load and ambient, a guaranteed power capacity delivered to the rack, a guaranteed availability demonstrated at integrated systems testing. They are proven at commissioning, so the commissioning program (Chapter 13.1) and the acceptance test plan (Chapter 13.2) are commercial documents as much as technical ones. The test scripts define the conditions under which the guarantee is met, and a loosely written acceptance test is a guarantee the owner cannot enforce. Warranties cover defects after acceptance, typically 12-24 months on workmanship and longer on specified equipment. Their value depends on not being voided by the O&M contractor's maintenance or by an OFE interface the warranty excludes.

Bonding and security backstop the whole regime. A performance bond (commonly 10-100% of contract value, depending on jurisdiction and counterparty) secures the contractor's performance of the bonded obligations for the owner, subject to the surety's default and remedy provisions. A payment bond secures qualifying labor and material claimants, not the owner's damages. A parent-company guarantee, retention withholding and letters of credit make an LD or warranty claim collectable against a contractor thinly capitalized at the project level, and an on-demand LC pays on presentation, where a surety can contest. The security package is the difference between a contractual right and a recoverable dollar. Where the EPC is an SPV or a JV, it is the only thing standing between the owner and an uncollectable judgment.

The long-lead equipment that dominates the schedule (Chapter 2.3) is procured under its own commercial regime, and three contract types govern it. An equipment supply agreement is the conventional purchase: spec, price, delivery date, warranty, and LDs for late delivery or non-conformance. A slot-reservation agreement is the newer and more interesting instrument. It is a distinct contract that buys a place in the manufacturer's production queue for transformers, switchgear or accelerators, typically with a substantial non-refundable deposit and price-and-spec terms fixed later. In a supply-constrained market the slot itself is the scarce asset, and the reservation contract secures it before you have finalized the design. The deposit is the price of optionality on a delivery date you cannot otherwise guarantee.

A slot reservation inverts the usual leverage. A normal supply agreement lets the buyer hold delivery risk over the seller through LDs. In an allocation-driven market the seller holds the buyer, because the deposit is sunk and the alternative is the back of a multi-year queue. This is the GPU and HBM allocation game in contract form, and it bleeds into financing: vendor-financed and prepay structures exist because the slot is worth more than the cash. The owner's defense is to write delivery-date firmness and a remedy for slippage into the reservation itself, rather than accepting a deposit that buys a place in line but no enforceable date.

Then comes the seam that quietly destroys performance guarantees: the owner-furnished-equipment (OFE) interface. When the owner procures equipment directly (GPUs and AI servers when the owner holds the allocation, and sometimes switchgear, transformers and CDUs) and furnishes it to the EPC for installation, the contract must define a clean line where the OEM's warranty ends and the EPC's installation warranty begins. If that line is fuzzy, a failure at the interface falls into a no-man's-land. The OEM says it was mis-installed, the EPC says the equipment was defective, and the owner, who sits on both sides of the handoff, owns a loss neither counterparty will accept. The OFE interface warranty is among the most under-negotiated clauses in the AI-build stack, and it is where the performance guarantee you thought you bought silently disappears.

Put a recipient and a record on each custody transfer. The OEM hands the owner the serial-numbered factory test and preservation instructions. The warehouse signs receipt condition and keeps the required storage logs. The EPC records condition at site receipt, handling and installation. The OEM field engineer witnesses the agreed tests without relieving the EPC of workmanship. When the dielectric test fails, preserve samples and test records before repair, notify the counterparties under their contracts and fund the work that protects the schedule. Proof of a manufacturing defect selects the OEM remedy, breached warehouse preservation selects the warehouse remedy, and installation damage selects the EPC remedy. That finding decides who owes the cure, not who must urgently protect the project. If the cause stays unresolved, the owner's temporary cash exposure remains real. Reconcile contractual recoveries with 2.6 so the same loss is not collected twice; the schedule and acceptance dates remain in 2.1.

The interconnection and power agreements are usually filed under 'regulatory' and treated as a permitting hurdle. That framing is costly. The LGIA, ISA or large-load service agreement is the most commercially binding instrument in the stack. It carries the project's hardest dates, its largest unhedged cost-allocation exposure and obligations you plan around rather than renegotiate, because amendments run through the tariff and the regulator on their timetable, not through a bilateral negotiation. Read it as a contract, not a permit. Three commercial dimensions dominate: cost allocation, milestones and security.

Cost allocation answers who pays for the network upgrades the interconnection study identifies. It follows the applicable utility tariff and transmission study. Separate upfront construction contributions, recurring demand charges, refundable collateral and cancellation liabilities in the model. AEP Ohio's Letter of Agreement requires reimbursement of 100% of buildout costs if the customer cancels or delays its project by more than 12 months before target energization; that obligation expires once the project is completed. As of September 2026, EEI counts 25 states with at least one approved large-load tariff. The upgrade cost is identified during the study, so the number is uncertain at the moment you must commit.

Milestones impose a schedule the owner must hit, and post security against, or forfeit its place. The EPC and tenant contracts must be built around these dates, because a missed interconnection milestone cascades through the entire stack. Security is the cash at risk, and it is tariff-specific. Tri-State's rejected 2025 proposal paired a $2.7M/MW deposit with a 15-year term. Its revised High Impact Load Tariff (HILT) received conditional approval on August 14, 2026, followed by a September 9 compliance filing. Model the operative agreement's security formula and minimum term.

The newest commercial wrinkle is the take-or-pay floor. To protect other ratepayers from a data center that contracts for capacity it never uses, state tariffs now impose minimum-payment obligations. For New Load, AEP Ohio Schedule DCT sets a ramp-period billing floor at 85% of the agreed Load Ramp Contract Capacity, and the ramp cannot exceed four years. During the ramp that floor is a fixed cost the project carries while utilization is lowest. Afterward, demand billing uses the month's measured peak, subject to the greater of 85% of the previous eleven months' highest established billing demand and the tariff's tiered minimum. A standalone 50 MW contract has a tier floor of 15 MW plus 85% of any capacity above 25 MW: 15 MW + 85% × (50 − 25) = 36.25 MW, or 72.5% of contract capacity. Affiliated New Load is aggregated. The minimum initial term is the ramp plus eight years. These floors directly underwrite the lender's view of contracted versus merchant exposure (Chapter 2.5).

On the supply side, FERC's June 18, 2026 PJM co-location order specifies three transmission-service categories: NITS, including Interim NITS pending required upgrades; Firm Contract Demand; and Non-Firm Contract Demand. The contract must identify the applicable category and its curtailment obligations. Interim NITS is a transition for customers seeking NITS, not a separate permanent service choice. → energy-supply strategy and the PPA-vs-BYOP fork in Chapter 3.4; site-side interconnection strategy in Chapter 3.1.

The bottom of the stack faces the customer, and it splits by who the customer is. If you are a colocation operator, the tenant contract is a colo agreement or MSA that sells space, a power-capacity commitment (in kW or MW, not square feet, because power is the product) and a facility SLA on uptime and PUE. The critical term is the power-capacity ramp. The tenant's density ramp (Chapter 1.6) must be matched by a contracted capacity-delivery schedule, or the tenant outgrows the contract and the operator either strands capacity or breaches. Tie rent, service commencement and longstop remedies to separate acceptance boundaries. If you are a GPU-cloud operator, the customer contract is a service-level agreement on uptime, goodput and support, backed by service credits. SemiAnalysis reported in November 2025 that CoreWeave and Oracle were beginning to offer 99% rack-level uptime guarantees with penalties attached (ClusterMAX 2.0).

The seam to watch at the bottom of the stack is the SLA pass-through: when a facility outage breaches the cloud SLA, who owes the service credit? If the cloud operator leases from a colo, the facility SLA and the cloud SLA must be back-to-back, or the operator owes its compute customers a credit it cannot recover from its landlord. The SLA metric itself is shifting. A raw uptime percentage understates what an AI customer buys, because a cluster that is 'up' but delivering poor goodput (failing jobs, stragglers, fabric congestion) breaches the spirit of the contract while meeting its letter. The frontier of tenant terms is moving from availability to goodput-based SLAs, the same reframing the reliability chapters apply to facility design (Chapter 12.2). The full productization of consumption and SLA tiers, the menu of firm, burstable and spot compute the customer buys, is built out in Chapter 10.9.