---
type: "post"
id: "google-and-voltus-announce-one-gigawatt-byoc-agreement-e45876"
slug: "google-and-voltus-announce-one-gigawatt-byoc-agreement-e45876"
title: "Google and Voltus Announce One Gigawatt BYOC Agreement"
description: "Google and Voltus have expanded a PJM demand-response agreement into a Google-funded virtual power plant targeting one gigawatt of flexible capacity."
retrieval_nugget: "Google and Voltus have expanded a PJM demand-response agreement into a Google-funded virtual power plant targeting one gigawatt of flexible capacity. For teams building AI infrastructure, the announcement shows that securing data-center power can also mean procuring controllable reductions in demand, rather than only adding generation or obtaining a grid connection."
published_at: "2026-10-27"
updated_at: "2026-10-27"
record_date: "2026-10-27"
date_kind: "published_at"
topics: ["google","ai-infrastructure"]
entities: ["Google","Voltus","PJM Interconnection","Bring Your Own Capacity"]
related_patterns: []
novelty_label: "new_fact"
what_changed: "Google and Voltus announced a 1 GW Bring Your Own Capacity agreement in PJM, paying businesses to reduce or shift electricity use for data-center power needs; no prior covers this deal."
prior_refs: []
source_url: "https://www.voltus.co/press/google-voltus-1gw"
source_urls: ["https://www.voltus.co/press/google-voltus-1gw"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:google-and-voltus-announce-one-gigawatt-byoc-agreement-e45876"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/google-and-voltus-announce-one-gigawatt-byoc-agreement-e45876/","markdown":"https://newruntime.com/posts/google-and-voltus-announce-one-gigawatt-byoc-agreement-e45876.md","json":"https://newruntime.com/posts/google-and-voltus-announce-one-gigawatt-byoc-agreement-e45876.json"}
---

# Google and Voltus Announce One Gigawatt BYOC Agreement

## Retrieval answer

Google and Voltus have expanded a PJM demand-response agreement into a Google-funded virtual power plant targeting one gigawatt of flexible capacity. For teams building AI infrastructure, the announcement shows that securing data-center power can also mean procuring controllable reductions in demand, rather than only adding generation or obtaining a grid connection.

## Summary

Google and Voltus have expanded a PJM demand-response agreement into a Google-funded virtual power plant targeting one gigawatt of flexible capacity. For teams building AI infrastructure, the announcement shows that securing data-center power can also mean procuring controllable reductions in demand, rather than only adding generation or obtaining a grid connection.

## What happened

On October 22, 2026, distributed-energy platform and virtual-power-plant operator Voltus said it and Google had expanded their agreement in the PJM grid region. PJM serves 67 million customers across 13 US states and the District of Columbia. The companies describe the arrangement as a one-gigawatt Bring Your Own Capacity agreement that will create a Google-funded virtual power plant.

The agreement follows an initial deal the companies signed in June 2026 for 100 megawatts of flexible power capacity in PJM. The new announcement describes an additional 1,070 MW of capacity. Voltus says the resulting portfolio is one of the largest known distributed-capacity portfolios deployed by a hyperscaler. The release does not state when the added capacity must be delivered, name particular Google facilities, disclose a contractual term, or give the price Google will pay.

Voltus plans to aggregate new capacity from homes and local businesses. When the grid is under strain, it will signal participating sites. Those sites can lower or shift their use of electricity through batteries, smart thermostats, industrial-process adjustments, and other flexible assets. The arrangement is therefore not a proposal to build a conventional new power station. It is a plan to make a distributed group of existing customer-side resources responsive during periods when capacity is valuable.

The company says participants may receive financial benefits, including bill savings and direct payments. Google describes the program as a way to bring its own capacity to the grids serving its operations. Voltus describes it as a way to turn data-center growth into direct economic benefits for homes and businesses. The companies also present the project as supporting grid reliability and affordability in PJM.

Operationally, the announcement adds an instrument to the usual discussion of data-center power. Instead of relying solely on supply procurement, interconnection, generation, and transmission, Google is funding a capability that can reduce or move other customers' load at constrained times. The source does not explain exactly how capacity will be dispatched, how it will be allocated relative to Google's operations, or what performance standard will determine whether enrolled resources count toward the target.

Voltus says its Bring Your Own Capacity program connects flexible load, distributed generation, energy storage, energy efficiency, and electric-vehicle resources to electricity markets. In this agreement, batteries and thermostats are examples, rather than an exhaustive specification of the portfolio. The release does not provide an asset mix, participant count, geographic distribution within PJM, or a record of past dispatch performance for the new capacity.

The scale cited in the release should also be read carefully. The headline calls the agreement one gigawatt, while the text says the expansion is an additional 1,070 MW after the earlier 100 MW agreement. The source presents the expanded arrangement as a gigawatt-scale portfolio; it does not reconcile that wording into a single disclosed contracted total or establish measured output at a peak event.

## What the archive adds

We previously covered [a Finnish order requiring Google to halt work at two data-center sites over missing environmental reviews](https://newruntime.com/posts/finland-ordered-google-to-halt-work-at-two-c6cdd6/). That Field Note concerned an execution gate on physical data-center construction. This is a new fact on the broader Google infrastructure story: it concerns flexible grid capacity once facilities seek power, rather than environmental review before construction proceeds.

Our earlier Telegram coverage also treated AI infrastructure as a financing and physical-capacity problem, including energy and grid constraints alongside compute. This announcement adds a specific procurement mechanism: funding demand-side flexibility in the same regional grid where capacity is needed.

## Trend signal

No listed New Runtime pattern directly fits this event. It is an infrastructure-procurement announcement, not evidence about how agent systems, software interfaces, or AI development workflows are changing.

## Why now

The timing in the source is explicit only in relation to the earlier agreement. Google and Voltus signed for 100 MW in June 2026 and announced the expanded portfolio on October 22, 2026. The release frames the tenfold expansion as an effort to bring capacity to the grids that serve Google's operations, but it does not identify a particular PJM reliability event, data-center opening, or policy change that prompted it.

## Connections to other events

The related Google Field Note about Finland shows another constraint on data-center expansion: physical construction can be delayed by environmental review. This Voltus agreement concerns a different layer of the same broader buildout. A site can clear construction hurdles yet still face limits on dependable electricity capacity. The two items should not be read as evidence of a shared project or common schedule; the archive supplies no such link.

## Why it matters

AI builders do not choose PJM capacity contracts in a weekly development cycle, but large-scale model deployment depends on decisions like this one. Usable data-center capacity is constrained by local electricity conditions as well as hardware availability. A virtual power plant is a different route from waiting for a new generator or transmission upgrade: it pays for aggregated, potentially dispatchable flexibility from many distributed assets.

For teams operating, buying, or forecasting AI infrastructure, the practical implication is planning discipline. Compute roadmaps should distinguish installed hardware from power that can reliably be served at constrained hours in a specific location. Procurement conversations should ask what grid assumptions, curtailment arrangements, and flexible-load resources sit behind promised capacity. The agreement does not prove demand response solves the wider power problem. It does show a hyperscaler treating it as a material part of its capacity stack.

## Limits and uncertainty

This is a joint company press release, not an independently audited account of delivered capacity. It does not disclose when the 1,070 MW will be enrolled, how much capacity will be available simultaneously, the frequency or duration of dispatches, or the price Google will pay. It also does not identify the Google data centers that benefit, quantify payments to participants, or establish whether the portfolio reduces the need for other PJM investments.

The source uses capacity language, not observed operational results. A target of one gigawatt does not demonstrate one gigawatt of measured performance during a particular peak event. Nor does the release provide a methodology for its “clean” characterization, emissions accounting, customer savings, or reliability contribution. The portfolio's future value depends on enrollment, availability, dispatch, and verification that the announcement does not yet document.

## Skill to build now

Add power availability to infrastructure due diligence for any AI workload that may require dedicated or colocated capacity. This week, separate nominal hardware supply from deliverable power in capacity plans, and ask prospective providers what they can serve during local peak periods. Record whether an offer relies on grid interconnection, on-site resources, contractual demand response, or another mechanism; those are different operational assumptions.

## Reader takeaway

Google is expanding its PJM power strategy beyond buying electricity or building facilities: it is funding a large pool of customer-side load flexibility through Voltus. For AI infrastructure, the binding resource is increasingly dependable capacity at the relevant hour and location, not simply more installed compute or nominal grid supply.

## Open question

Can the portfolio enroll and deliver the stated flexible capacity at the same times that Google's PJM data-center demand is highest?

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Google](https://newruntime.com/topics/google.md): Explore the google topic hub.
- topic hub: [Ai Infrastructure](https://newruntime.com/topics/ai-infrastructure.md): Explore the ai-infrastructure topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)