Dana Guernsey did not set out to raise money this year. Voltus had a different plan for 2026: buy more batteries, sign a few hyperscaler capacity contracts, and keep compounding from a position of quiet profitability. Then the contracts started landing faster than the operating plan could absorb them.
"By the spring, it became pretty apparent to us as we were signing these BYOC contracts that now was the time to accelerate," the chief executive said. The internal name for the round was Project Accelerate. It closed as a $225 million oversubscribed Series D, led by Generation Investment Management, with Activate Capital and Vitol as co-leads. Broadscale Group, Climate Investment, NGP, Ajax Strategies and All Aboard Fund participated. Voltus did not disclose its valuation.
The date matters more than the number. Voltus announced the round on October 8. The next morning, the Federal Energy Regulatory Commission rejected the core of PJM Interconnection's plan for connecting data centers and other large loads.
A close that landed one day before the rulebook changed
For a year, PJM's proposal had been the most concrete attempt in the United States to resolve a structural collision. Data centers wanted to connect before the generation to serve them existed. PJM wanted to let them in anyway, in exchange for being curtailed first during emergencies. That trade was bundled into an Interim Resource Adequacy Service, filed in docket ER26-3515 on August 13 with a requested October 12 effective date.
FERC kept the definitional scaffolding: a large load is now 50 MW or more at a single site, aggregating affiliated facilities within a one-mile radius, and a Large Load Registry became effective October 12. It rejected IRAS, the bring-your-own-capacity provisions, and the related tariff language. Commissioner David Rosner concurred, warning that excluding IRAS loads from the capacity auction's demand curve could signal surplus while the region shed thousands of megawatts, and arguing that curtailment should come only after voluntary demand response is exhausted.
The practical effect is that the fast lane is gone. A developer that had planned to connect under IRAS and accept curtailment risk now needs a different path, most likely a new filing or state-level action.
That is the gap Voltus is selling into. Its product, Bring Your Own Capacity, asks a hyperscaler to pay for accredited distributed capacity in the communities around its facilities rather than wait in an interconnection queue that has stopped clearing on schedule. Voltus aggregates the other side: batteries, cooling systems, industrial processes, enrolled at thousands of businesses and homes, dispatched into wholesale markets.
Voltus Series D round size
First raise since a $31 million Series C in 2021. Valuation not disclosed. · Voltus, October 2026
The company says it manages 8.5 GW across more than 22,000 participant sites in nine US and Canadian wholesale markets. In 2021, at the Series C, that figure was 1.7 GW. The stated goal is 20 GW by 2030, which means roughly doubling a portfolio that already took five years to assemble from a much smaller base.
What BYOC actually sells
The mechanism is easy to misread as demand response with better branding. Traditional demand response pays a customer to curtail load when the grid is stressed. Voltus's version packages that flexibility as capacity a large load can point to when regulators and utilities ask where its megawatts will come from.
The customer list has become the proof point. Voltus announced Google as its first official BYOC customer in June, aggregating up to 100 MW of accredited distributed capacity in PJM. In August, Sunrun agreed to supply capacity from residential solar-plus-storage systems into the same program. Earlier partnerships with Octopus Energy US and CenTrio extended the model into retail supply and district energy. The company also acquired energy storage startup Brightfield, largely for the team that scopes which existing Voltus sites can host batteries.
Guernsey frames the shift in market structure, not technology. "Over the course of the last year and even just months, more and more folks have understood that this is how we bring more capacity online," she said. "We have proven this thesis. This is an execution play."
Electricity is now the constraint behind almost every major ambition we have, but it doesn't have to be. The question is no longer whether the grid needs flexibility, it's how quickly we can make it foundational.— Dana Guernsey, CEO, Voltus
The 6,831 MW hole
PJM's numbers explain why capital showed up. In July, the 2028/2029 Base Residual Auction cleared at the $325/MW-day price cap and procured 138,318 MW, finishing 6,831 MW short of its reliability requirement. The prior auction was about 6,500 MW short. PJM says these were the first two auctions in its history in which the entire footprint fell short. Only 525 MW of new generation and uprates cleared.
| Delivery year | Clearing price ($/MW-day) | Context |
|---|---|---|
| 2024/2025 | 28.92 | Pre-data-center repricing |
| 2025/2026 | 269.92 | Record at the time |
| 2026/2027 | 329.17 | At or near cap |
| 2027/2028 | 333.44 | At or near cap |
| 2028/2029 | 325.00 | Cap; 6,831 MW short |
PJM RTO-wide capacity clearing prices, 2024/25 through 2028/29 delivery years. Source: PJM, via LandGate and MGRID summaries.
A capacity price at the cap for four consecutive delivery years changes the arithmetic of every alternative. LandGate's illustration is blunt: 300 MW priced at $325/MW-day works out to about $35.6 million a year before a single kilowatt-hour of energy is billed. Any mechanism that delivers accredited megawatts faster than a new gas turbine, and at lower capital cost, is now competing against a very expensive baseline.
PJM 2028/29 reliability shortfall
Auction cleared at the $325/MW-day cap while procuring 138,318 MW. · PJM, July 2026
The demand backdrop is not speculative. NERC expects peak demand to grow 24% over the next decade, more than double its forecast from two years ago. The pv magazine summary of NERC's estimate puts the need at more than 270 GW of new peak capacity by 2034. Rocky Mountain Institute argues that virtual power plants could account for up to 60 GW of that by 2034, alongside efficiency and grid-enhancing technologies.
Why the aggregation layer is the contested asset
If flexibility is the marginal unit, the valuable position is not the battery or the chiller. It is the contract with the customer, the accreditation with the market operator, and the dispatch software that turns thousands of small commitments into a block a utility will count.
That is a different business from installing hardware, and it explains the investor mix. Generation Investment Management brings the climate thesis, Vitol brings trading depth in US power, and Activate Capital has backed the company since the Series C. Each is buying exposure to a market-design outcome rather than a technology bet.
Voltus's BYOC economics depend on market rules that credit distributed capacity as substitutable for generation. FERC's rejection of IRAS removed a competing fast lane, but it also left the curtailment question unresolved and pushed it to states. A state-by-state patchwork would raise the cost of scaling the aggregation layer across nine markets.
Voltus has been unusually active in the PJM stakeholder process, and as we noted in our October analysis of the flexible-load rulebook, the proposed framework had already moved in its direction before FERC's order: sites that participated in demand response previously but skipped last year can be reactivated and counted as new capacity, and incremental capacity such as new batteries at existing locations will count. Those changes matter more to a 20 GW target than the headline round size.
The company's own framing is that grid capacity is now the binding constraint on AI buildout. That claim is supported by the auction data. The harder question is whether an aggregator can scale accreditation faster than hyperscalers can sign direct supply deals, build behind-the-meter generation, or wait for the next PJM reform.
What to watch over the next four quarters
Three markers will show whether the round bought a durable position or a cyclical one. First, the count of accredited BYOC megawatts in PJM after the registry matures. Second, whether additional hyperscalers follow Google into multi-year capacity contracts rather than one-off pilots. Third, whether Voltus converts its battery pipeline through the Brightfield team fast enough to lift the portfolio above the 8.5 GW base without buying it.
Stated portfolio target
Up from 8.5 GW under management in October 2026. · Voltus, October 2026
Does the flexibility thesis hold through 2027?
Probability: 65% — PJM capacity prices are pinned near the cap and FERC has removed the competing fast lane, but state-level curtailment rules remain unresolved.
🟢 Bull case (30%)
Consequence: the aggregation layer becomes the default procurement route for large loads, and Voltus enters 2028 as the category's price-setting intermediary.
🟡 Base case (55%)
Consequence: the business compounds, but the 20 GW target slips past 2030 and the IPO question stays open.
🔴 Bear case (15%)
Consequence: Voltus's accredited portfolio keeps growing while its per-megawatt economics compress, turning a capacity play into a margin business.
The round is best read as a market-design bet placed at the moment the design was contested. Voltus raised on October 8 that flexible capacity is the cheapest new supply. On October 9, FERC removed the alternative that would have let large loads bypass the question. Both facts point the same way for now. Neither settles who gets paid when the grid is short, and that is the part of the model the $225 million cannot buy.