Hyperscalers and Distributed Energy Capacity: What Data Centre Demand Means for US DG Platforms in 2026

Data current to 13 October 2026.

Hyperscalers are now paying for distributed capacity, not only signing utility-scale power purchase agreements. In six weeks, Google funded a utility-run virtual power plant in California, Voltus raised $225 million to sell distributed capacity to data centres, and more than 140,000 home batteries dispatched 580 MW in a single event. For US distributed generation (DG) platforms, that creates a buyer that purchases very differently from a community solar subscriber. This briefing sets out the deals, the market rules and programme risks behind them, and the commercial capability a platform needs to sell to this buyer.

In brief

• Data centre demand is reaching distributed assets through virtual power plants (VPPs) and capacity products, alongside continued utility-scale PPAs.
• The near-term route is bilateral contracting. Aggregated distributed resources cannot take part in PJM's capacity market until 1 February 2028.
• Much distributed capacity still depends on state programmes with uncertain funding, such as California's DSGS.
• In VSP's 2026 Leadership Pay Review, origination leaders with large-load or wholesale experience earned a median base of $290,000, 14% above the median for all origination leaders. Fewer than one in five DG commercial leaders in the review had led a contract with a large-load buyer.

How are hyperscalers buying distributed capacity?

A virtual power plant is a fleet of small distributed assets, such as home batteries, smart thermostats or commercial solar and storage, controlled together so it can respond to the grid like a single power plant. Voltus's BYOC product applies that idea to data centres: it aggregates distributed capacity so a large new load can bring its own supply to the grid.

Voltus's round was led by Generation Investment Management, with Activate Capital and Vitol co-leading. Google became the first customer for BYOC in June. Voltus targets a 20 GW platform by 2030, which is a company target. PJM, the grid operator covering 13 states and Washington DC, does not open its capacity market to aggregated distributed resources until 1 February 2028, so the near-term route is bilateral contracting.

Each figure needs its qualifier. The 21,000 in SHARE is a device target, not enrolled capacity. Madison's targets are its own, and no hyperscaler offtaker has been named. The 580 MW figure, and the description of it as a record, come from Sunrun and Tesla. The dispatch was triggered by California's Demand Side Grid Support (DSGS) programme and PG&E's Emergency Load Reduction Program.

Is utility-scale offtake still growing alongside?

Yes. This is additive, not a replacement. In the same window, Zelestra closed $350 million for its 203 MW Reclamation Solar project in Indiana, backed by a Meta PPA. Apex Clean Energy and Meta signed their seventh deal. ContourGlobal broke ground on a 450 MWac solar and 360 MW storage project in Arizona, with Tesla taking about 90% of the output.

The Michigan Public Service Commission also approved a DTE contract on 1 October under which Google funds up to 1,600 MW of renewables and 480 MW of storage for a 1 GW data centre. That model is utility-owned, and it competes with independent power producers for the same buyer.

What are the programme and market risks?

Distributed capacity still leans on fragile state money. California's budget left DSGS, the programme behind the September dispatch, with no new funding for 2027. Canary Media reports that a $27 million transfer keeps it running through 2026.

The longer-term route is resource adequacy, the system regulators use to make sure enough capacity is contracted to meet peak demand. California's SB 913, signed on 30 September, directs regulators to let aggregated distributed resources count towards resource adequacy, with a deadline of 30 June 2028 in the bill text. In Texas, ERCOT's aggregated distributed energy resource pilot has 293 MW qualified against a 500 MW cap, and a Phase 4 proposal would remove the cap for aggregated storage. It remains a proposal.

The buyer is changing from thousands of subscribers to a handful of data centre operators. The business behind the asset has to change with it.

What changes for distributed platforms?

The commercial model shifts from many small customers under state programme rules to a few large counterparties buying capacity.

Not every platform will make that shift, and the evidence does not say how many will. Programme-led community solar remains the core business for most distributed developers. But the platforms that want the new buyer will need to sell, contract and operate in a different way.

What does this mean for leadership and capability?

None of this is evidence of hiring demand. What these developments do change is the capability mix a distributed platform needs if it wants this buyer:

• Enterprise and large-load origination, rather than subscriber acquisition
• Wholesale market and capacity rule expertise across PJM, MISO, ERCOT and CAISO
• Dispatch, software and fleet performance management
• Regulatory engagement on resource adequacy and large-load tariffs
• Contract structuring for capacity products that sit between a PPA and a demand-response programme

That mix is rare in DG commercial teams today. In VSP's 2026 review, only 18% of commercial leaders at DG and community solar platforms had led a contract with a utility, data centre developer or other large-load buyer. Those who had earned a median base of $290,000, against $255,000 across all origination leaders. For the full pay ranges, see US community solar and DG leadership pay 2026.

For boards, the question is whether a commercial team built for state programmes and subscribers can sell to a data centre. For candidates, experience across both worlds, distributed assets on one side and large-load or wholesale commercial work on the other, is an unusual combination worth making explicit.

Frequently asked questions

Are hyperscalers buying capacity from distributed energy resources?

Yes. Google funded PG&E's SHARE virtual power plant in California in September 2026 and became the first customer for Voltus's Bring Your Own Capacity product in June 2026. Most hyperscaler procurement is still utility-scale PPAs, so distributed capacity is an addition, not a replacement.

What is Voltus's Bring Your Own Capacity product?

It aggregates distributed energy resources so that a data centre can bring its own capacity to the grid. Voltus raised a $225 million Series D on 8 October 2026 to fund it in PJM and MISO.

When can aggregated distributed resources join PJM's capacity market?

From 1 February 2028. Until then, distributed capacity sold to data centres in PJM moves through bilateral contracts.

What capabilities do DG platforms need to sell to data centres?

Large-load origination, wholesale and capacity market expertise, fleet dispatch and performance management, regulatory engagement on resource adequacy, and contract structuring for capacity products.

Do commercial leaders with large-load experience earn more?

In VSP's 2026 Leadership Pay Review, origination leaders who had led a contract with a large-load buyer earned a median base of $290,000, 14% above the $255,000 median for all origination leaders.

About the data

VSP figures come from the VSP 2026 Leadership Pay Review: 142 senior offers, counter-offers and pay discussions at 63 US distributed generation, community solar and IPP platforms, January to September 2026. "Large-load experience" means the candidate led a signed contract with a utility, data centre developer or other customer above 50 MW. Figures are median base salary.

For why contracts now decide value in storage, see contracted storage is pulling ahead of merchant. For where distributed solar capital is going, see community solar capital is backing operators.

Read next: Interconnection Certainty Just Slipped Into 2027. Someone on Your Team Owns That Risk.

Further reading: Market Reports → · FAQs →