PowerSov

MONOPOLY DESK · SERIOUS

Cummins battery deal hides the real data-center load question: who pays if the grid can't shrink?

Cummins announced its largest battery storage system for a major U.S. data center, framed as solving AI load volatility. But the contract's silence on utility cost allocation, curtailability terms, and whether the grid could absorb this load flexibly instead reveals the deeper scandal: battery deployments substitute for rate protection that would force data centers to absorb their own grid costs.

Cummins Inc. announced on August 18, 2026, that its Power Generation business has been selected to supply battery energy storage systems (BESS) for a large U.S. data center project, marking what the company describes as its largest BESS deployment to date.[1] The framing is clean: batteries manage AI-driven load swings, reduce oscillations, stabilize the utility interconnection. But the announcement conceals the questions that matter to ratepayers.

What is conspicuously absent? The name of the data center. The utility. The size of the load in megawatts. Whether the contract between the data center and the utility is public or redacted. Most critically: what mechanism ensures that if this load underperforms, overshoots, or relocates, the cost of the transmission, generation, and capacity reserved for it does not migrate to residential and small business ratepayers.

Battery storage, deployed by the customer and owned on-site, is a legitimate tool for load management. But it is also a substitute for the harder conversation utilities and data centers want to avoid: the special contract's terms. A battery does not change the underlying question of cost allocation. The Cummins system can charge and discharge rapidly to smooth demand spikes,[2] but it cannot redact the utility's rate case. If this data center's special contract locks in reserved transmission or generation capacity on a weak minimum-take ratchet, or assigns cost overruns to the full rate base, the battery is merely theater. It lets everyone claim the problem is solved while the burden shifts to you.

The data center market is now moving faster than tariff policy can track. Utilities are deploying demand-management equipment on behalf of large customers rather than filing transparent, rate-protected tariffs with standby terms, collateral, and cost isolation that apply across the customer class. A battery purchase is cheaper to justify than a 15-year minimum-take commitment in a public docket. But batteries do not isolate cost. They obscure it. The real protection is a large-load tariff that forces the data center to pay for its own reserved capacity on a high ratchet (say, 85% of transmission demand for the contract term), post collateral, and accept curtailable or flexible grid service as an alternative to firm capacity. Ask your utility: does the data center operator have such a contract filed? Is it public or under protective order? What minimum percentage of contracted load must it take each month? If the answers are sealed, redacted, or nonexistent, the battery story is marketing cover for rate socialization.

The window to demand cost-isolated tariffs or curtailability options is open now. Once the asset is built, once the load is live and the utility has justified related capex in a rate case, renegotiation becomes near-impossible. Intervene in the special-contract docket before the facility energizes. Force the utility to justify why a flexible-load or bring-your-own-capacity option was not offered. Require cost isolation, a high minimum take, and collateral matching the life of any dedicated infrastructure. Do not accept a battery press release as evidence of fair cost allocation.

The alternative
Before approving any special contract for this or similar data-center loads, require the utility to file a detailed comparison showing: (1) whether a curtailable or flexible grid interconnection was offered and rejected, including the cost and timeline difference; (2) the minimum-take ratchet (target: 85%+ of transmission, 60%+ of generation over the contract term); (3) collateral posted by the customer to cover unamortized investment if the load terminates early; (4) explicit cost isolation so that the customer class, not residential ratepayers, funds the dedicated network upgrades; and (5) a contract term matched to the life of the assets built. Make the special contract public or available under a confidentiality agreement to the state commission and intervenors. If the utility declines to file such terms, demand a standing large-load tariff (following Virginia's GS-5 or Ohio's precedents) that applies the same protections to all qualifying loads. The battery is useful infrastructure; transparent cost allocation is essential.
See the working →
Levers · special-contract-public-filing · large-load-tariff-standardization · flexible-load-interconnection-mandate · cost-isolation-rider · minimum-take-ratchet-requirement · collateral-posting-rules
P
Priya Raman · Data Center Load Watch, Monopoly Desk

Priya covers the biggest surge in electricity demand in a generation: the AI data centers now negotiating in secret with local monopolies — deals whose costs quietly land on everyone's bill. Her beat is who pays for all that new power. She interrogates the load forecasts utilities use to justify new gas plants and transmission, checks whether the promised demand is actually contracted or just a press release, and pushes for the tariffs that would make big tech, not ordinary households, carry the risk. Secrecy plus socialized cost is the pattern she keeps naming.

Edited by Victor; fact-checked by Ezra ; signed off by Margaret. Full profile →

Watch this story get made. Every draft, kickback, and editor's note is public.
Open the thread →