Power has always been an input to data center economics. In 2026, it is the underwriting variable.
Capital continues to chase AI and cloud demand. But the market is now pricing a different scarcity than it did five years ago. The constraint is not “data center supply” in the abstract; it’s deliverable megawatts on a timeline that matches customer demand.
This week offered a clean illustration of the shift: a White House–convened “Ratepayer Protection Pledge” signed by Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI commits signatories to fund new generation and pay for grid upgrades required to serve their data centers, with the stated goal of avoiding cost pass-through to existing utility customers (POWER Magazine). Whether you view the pledge as policy signaling or a meaningful framework, it’s a marker of where the market is headed: power cost allocation is becoming a board-level topic, and regulators are increasingly willing to require bespoke rate structures for large loads (POWER Magazine).
For buyers, sellers, and lenders, the implication is direct: M&A is no longer about buying buildings and leasing space. It’s about buying (or creating) a power position—and understanding the downside if that position is conditional.
The new underwriting stack: power-first, land-second
A decade ago, a credible “site” meant land, fiber adjacency, a substation nearby, and a path to permits. Today, those boxes are necessary but insufficient.
“Available power” is no longer a binary claim. A site can have: – a conceptual utility conversation – a position in an interconnection queue – a signed service agreement with enforceable milestones – a deliverability model with upgrades identified – a tariff that defines who pays for what
Those are not equivalent.
The premium has moved from shell to substation: the most valuable assets in many markets are the ones that have already converted complexity into certainty. In practical terms, certainty looks like defined upgrades, committed schedules, and enforceable payment obligations.
That’s why a seemingly “operational” topic—large-load tariffs—has become directly relevant to capital markets. As of late 2025, one tracking effort cited dozens of large-load tariffs across many states, reflecting how rapidly utilities and regulators are adapting to hyperscale demand (POWER Magazine).
This week’s signal: hyperscalers publicly commit to pay for upgrades
The pledge matters less as a standalone document and more as confirmation of an emerging deal reality:
1) Utilities and regulators are pushing cost responsibility toward large loads. – The pledge describes separate rate structures negotiated with utilities and state governments, with hyperscalers bearing costs of generation and grid upgrades required to serve them (POWER Magazine).
2) Minimum bills and “pay whether you use it or not” is becoming normal. – The pledge’s fact-sheet framing includes paying for contracted supply and associated delivery infrastructure regardless of utilization (POWER Magazine).
3) Public scrutiny is rising. – When a market starts arguing about who pays, timelines slow down and diligence standards rise.
There’s also a critical limitation: the pledge is voluntary and the reporting suggests it does not include clear enforcement mechanisms or independent auditing (POWER Magazine). For M&A, that means you cannot “assume compliance” or treat this as a substitute for local tariff and interconnection diligence.
What we’re seeing in market activity
Even in a single monthly roundup, you can see the shape of 2026’s market:
Large acquisitions and financing packages tied to platform expansion. For example, one industry roundup reported Equinix and Canada Pension Plan Investment Board acquiring atNorth for $4B, alongside a $4.2B financing package (DataCenterKnowledge).
Power partnerships embedded in growth stories. The same roundup highlights multiple instances where utility relationships, renewable procurement, and “energy as strategy” sit next to site announcements (DataCenterKnowledge).
The pattern: the market is converging toward an IPP-like mindset. You’re not just underwriting a lease-up curve; you’re underwriting a power delivery plan.
A buyer’s checklist for “power diligence” in any platform or asset deal
Here is the diligence lens we believe separates “good deals” from “regret deals” in the current cycle.
1) Interconnection and deliverability
- What exactly has been submitted: load letter, interconnection request, or executed agreement?
- What upgrades are identified, and who pays?
- What are the “stop points” that could reset schedule (restudies, re-clustering, transmission constraints)?
If the site is in a market with congestion or high queue volatility, treat schedule as a distribution, not a point estimate.
2) Tariffs, riders, and minimum bills
Large-load rate structures can materially change cash flows. – Does the tariff impose minimum demand charges? – Are there construction-work-in-progress riders or true-ups? – Are there provisions that shift upgrade costs to the customer regardless of load utilization?
This matters because “pay whether you use it or not” is increasingly part of the policy conversation (POWER Magazine).
3) Fuel, firmness, and curtailment rights
A signed PPA is not the same as firm deliverability. – Is the supply shaped, firmed, or exposed to intermittency? – What curtailment rights exist (utility side and customer side)? – Are there backup generation expectations or obligations?
The pledge itself references coordination with grid operators and making backup resources available during emergencies (POWER Magazine). That direction of travel should influence how you view resiliency as a contractual requirement, not a marketing feature.
4) Schedule realism and permitting pathways
Two schedules matter: your construction schedule and your power schedule. – What’s the critical path to energization? – Are there permitting or environmental approvals that historically run long in that jurisdiction?
The market is openly discussing permit timeline compression for new generation projects (POWER Magazine). Whether or not you believe the timelines, the existence of the discussion is a clue: buyers should plan for both upside and delay.
Deal structuring moves that reduce regret
When power is the constraint, deal terms need to reflect that reality.
Earnouts tied to power milestones
If the seller is marketing a growth narrative based on future energization, consider tying a portion of consideration to: – executed interconnection agreement – substation delivery – partial energization – final energization
Step-in rights and control over utility engagement
When a platform’s valuation is tied to interconnection outcomes, control matters. Buyers often underestimate how much value is created by disciplined utility engagement, escalation, and documentation.
Contracting strategies: PPAs vs sleeves vs behind-the-meter
There’s no universal “best.” But the wrong choice can create stranded cost or schedule risk. – A PPA can solve price risk without solving deliverability. – Behind-the-meter can solve timeline risk while introducing fuel and permitting complexity.
When to walk away
Walk away when: – the “MW story” is unsupported by documents – upgrades are undefined or cost allocation is speculative – the schedule assumes perfect execution across multiple external stakeholders
Where Stacked AI fits
Stacked AI works as a neutral, expert advisor focused on speed and risk reduction.
In practice, that means: – translating business requirements (MW, density, geography, timeline, risk tolerance) into sourcing criteria – validating power path assumptions early, before teams waste cycles on non-viable sites – structuring a procurement process that keeps leverage with the buyer and avoids “deal drift”
In 2026, winning is less about finding a building. It’s about buying a plan you can actually execute.