AI Cloud

Google and Constellation add 890 MW of nuclear uprates to PJM

Google and Constellation signed a 20-year power deal that funds uprates at 11 nuclear units across six sites in Illinois, Pennsylvania, and New Jersey. The package is meant to put 890 MW of new capacity on the PJM grid, with the first uprate expected in 2028. PPA price and which Google campuses take the power are undisclosed.

By Cole Harlan · 7 Oct 2026 · 13 min

On 6 October 2026, Google and Constellation Energy said they had struck a long-term clean-energy collaboration that will fund nuclear uprates at Constellation plants serving the PJM Interconnection. The centerpiece is a 20-year power purchase agreement. The companies say that agreement will unlock 890 megawatts of new nuclear capacity on the PJM grid.

Cowry Tech News is covering this as a power-and-compute story. Hyperscalers need firm megawatts for data-center growth. Grid operators need new supply that does not wait a decade for greenfield plants. Buyers and lenders need to know what the announcement actually locks in, and what it still leaves out.

This piece is not the Oct 2 CTN Brief on Google’s space-satellite work, and it is not the Oct 6 Brief on Applied Digital’s Finland power-access arrangements. Those are separate items. Today’s story is the Google–Constellation nuclear-uprate package for PJM.

What Google and Constellation announced

The facts in the Google Cloud Press Corner release and Constellation’s matching news post are the same on the commercial core.

  • Parties: Google and Constellation Energy Corporation.
  • Structure: a 20-year power purchase agreement that funds equipment and technology investments at 11 Constellation-owned nuclear units.
  • Geography: Illinois, Pennsylvania, and New Jersey, across six nuclear sites.
  • Capacity claim: 890 MW of new, firm nuclear capacity added to the PJM footprint. The companies call that capacity entirely new generation available to the grid.
  • First delivery: first uprate expected by 2028.
  • Investment: more than $4.3 billion of new investment by Constellation (company figure).
  • Jobs: helps sustain roughly 4,400 existing jobs and creates approximately 7,200 construction jobs during the building period (company figures).
  • Separate supply deal: a 15-year energy supply agreement for an additional 2,700 MW from Constellation’s operating PJM fleet.
  • Technology side: an expanded five-year Google Cloud and Gemini Enterprise alliance framed as an “AI for Energy” blueprint.
  • Framing: response to PJM’s “Bring Your Own Power” proposal and alignment with a White House Ratepayer Protection Pledge, as stated in the release.

Amanda Peterson Corio, Google’s global head of energy and power, tied the deal to grid scale and jobs. “We’re committed to meeting our growth responsibly by actively investing in clean, reliable power that brings new capacity to our nation’s grids,” she said in the release. “Our agreement with Constellation to fund nuclear reactor uprates will strengthen the PJM grid, which serves 67 million people, all while protecting energy affordability and supporting local union jobs.”

Joe Dominguez, Constellation’s chairman, president and CEO, framed it as a model for private funding of grid-wide benefits. “Preserving and expanding the output of the existing nuclear fleet and developing new clean generation is critical to meeting our nation’s growing energy needs while keeping costs affordable for homes and businesses,” he said. “This long-term clean energy collaboration with Google can serve as a model for how technology companies and the energy industry can work together to responsibly develop the digital economy and invest in our nation’s energy infrastructure in a way that delivers grid-wide benefits for all, funded by private entities.”

What a nuclear uprate is

A nuclear uprate raises the output of a reactor that is already built and licensed to operate. Keep the plant. Modernize the parts that limit how much electricity it can put on the wire. Get more megawatts without pouring a new containment building.

Google’s blog by Amanda Peterson Corio describes an uprate as modernizing turbines, steam generators, and digital control systems to boost power output. Constellation’s release uses the same hardware list: turbines, steam generators, and digital control systems, plus thermal and electrical efficiency gains.

That path differs from a greenfield reactor or a small modular reactor (SMR) project. Those start from site selection, licensing, construction, and interconnection queues that can run many years. An uprate still needs regulatory approval and outage windows. It does not start from a bare field. The companies say the 890 MW package is roughly equivalent to the output of a large new conventional reactor, or several SMRs, delivered years sooner than building from scratch.

The Press Corner release also ties the work to the federal UPRISE program, which the companies describe as encouraging public-private collaboration to maximize clean electricity from existing U.S. nuclear infrastructure. CTN is reporting that framing as company language, not as an independent federal scorecard for this deal.

The 890 MW package and the separate 2,700 MW supply deal

Readers should keep two numbers distinct.

The first is 890 MW. That is the incremental capacity the companies say will come from uprates at the 11 units. It is the “new” megawatts. The release presents it as new baseload supply on PJM, not as a reallocation of existing plant output.

The second is 2,700 MW. That is a separate 15-year energy supply agreement covering power from Constellation’s existing operating fleet in PJM. The release says this financial structure is meant to keep those assets delivering energy and capacity to the PJM market, and to keep them economically viable without taking resources off the grid.

Mixing the two into one headline total would misstate the deal. 890 MW is the uprate capacity claim. 2,700 MW is a supply agreement on existing generation. Different products, different terms, different timelines.

The Press Corner release also lists demand-flexibility pieces inside the commercial package: operational load-shaping and demand-response capabilities so non-critical consumption can be curtailed during high-stress grid events. Google and Constellation also say they have a strategic framework to evaluate further clean generation, storage, and demand-response opportunities in the United States, with near-term priority in PJM. That framework is exploratory language. It is not a second 890 MW commitment.

Timeline: 2028 first uprate through end-2032

Two dates appear in the company materials.

The joint release says the first uprate is expected to be delivered by 2028. Google’s blog says the uprate projects will add the 890 MW to PJM before the end of 2032. Together, that describes a multi-year construction and outage program, not a single switch-flip in 2028.

Neither the Press Corner release nor the Google blog publishes a plant-by-plant schedule, a year-by-year MW ramp, or NRC filing dates for each unit. ANS / Nuclear Newswire reported that Constellation spokesman Paul Dempsey named plants involved as including Braidwood, Byron, LaSalle, and Quad Cities in Illinois; Limerick in Pennsylvania; and PSEG Nuclear’s Salem in New Jersey (with Constellation owning a minority stake in Salem). CTN treats that naming as secondary reporting attributed to ANS and the spokesman. The primary releases themselves stay at “11 units” and “six sites” without a public unit list in the Press Corner text CTN used.

Buyers who need firm dates for interconnection studies or campus power planning should ask Google and Constellation for a delivery schedule in writing. The public timeline today is: first uprate expected 2028; full 890 MW package targeted before end-2032, per Google’s blog.

Jobs, investment, and PJM framing

Constellation and Google put three economic figures next to the uprates.

More than $4.3 billion of new investment by Constellation. Roughly 4,400 existing jobs sustained. Approximately 7,200 construction jobs during the building period. All three are company figures from the joint release. CTN is not independently verifying headcount or capital spend.

PJM, in the release’s language, serves 67 million people across a Mid-Atlantic and Midwestern footprint. Google’s blog says the grid spans 13 states. The companies pitch the deal as adding new capacity alongside hyperscaler load growth, rather than shifting existing megawatts away from other customers.

The release links that pitch to PJM’s “Bring Your Own Power” proposal and to a White House Ratepayer Protection Pledge. In company terms, major power consumers bring new supply as they grow, and do so without passing uprate costs onto residential ratepayers. That is framing from the announcement. Whether rate cases, capacity-market outcomes, and retail bills move as claimed will show up later in PJM and state dockets, not in a press release.

The Google Cloud / Gemini Enterprise side of the deal

Alongside the power contracts, Constellation selected Google Cloud and Gemini Enterprise under an expanded five-year technology alliance. The release calls it an “AI for Energy” blueprint for next-generation grid operations.

Three focus areas are named in the release:

  1. Speed to power: AI for site selection, power-flow modeling, permitting, and interconnection planning, aimed at shortening engineering and regulatory work on uprates.
  2. Generation optimization: asset health monitoring, outage and turnaround management, and scenario optimization so plants stay at full output.
  3. Protecting critical infrastructure: Google Cloud platforms for securing operational technology (OT) networks, including air-gapped systems that connect to the bulk electric system.

Google’s blog adds that Constellation will deploy Gemini Enterprise agentic workflows across core operations to accelerate uprate planning, streamline construction, and optimize plant output. That is Google’s product framing for the alliance. CTN is not treating it as a measured performance claim. No latency, outage, or cost-reduction metrics appear in the materials CTN used.

For readers who buy cloud or sell into utilities, keep the lines straight. The 20-year PPA and the 15-year supply deal are energy contracts. The five-year Google Cloud / Gemini deal is a technology alliance. They sit in one press package. They are not interchangeable.

What the disclosure leaves out

Several buyer-relevant numbers are missing from the primary materials.

Soft hold on offtake campuses: Neither the Press Corner release nor Google’s blog names which Google data-center campuses will take power under the PPA, or how many megawatts attach to any named Google site. CTN will not invent campus lists or campus-level MW.

Soft hold on PPA price: The $/MWh or other contract price is undisclosed. CTN will not invent a price.

Also absent from the primary releases CTN used:

  • a binding plant-by-plant MW breakdown for the 890 MW total,
  • a year-by-year delivery schedule beyond first-uprate-by-2028 and package-by-end-2032,
  • the share of the $4.3 billion that is equipment versus labor versus digital systems,
  • measured ratepayer savings for PJM residential customers,
  • independent verification of the job figures.

ANS / Nuclear Newswire noted that plant names were not in the formal announcement and relied on a Constellation spokesman for a list. That gap matters for local reporting and for anyone modeling transmission constraints around specific sites. It does not change the company-level claim of 890 MW across six sites and 11 units.

The accurate frame is a long-term commercial package announced on 6 October 2026: uprate-backed new capacity, a separate existing-fleet supply agreement, and a multi-year cloud/AI alliance. It is not a named map of Google campus offtake, and it is not a priced PPA you can plug into a model without asking the companies.

How this sits next to Google’s other nuclear moves

Google’s blog places the Constellation deal inside a wider nuclear portfolio. CTN is using that section as attributed context only. It is not today’s lead.

Google says that with this PPA, it has “enabled over 1.5 GW of new nuclear capacity for the U.S. energy system from increasing the power output of existing reactors and bringing dormant facilities back online.” That “>1.5 GW” figure is Google’s company framing. Soft hold: CTN attributes it to Google and does not treat it as a third-party audited total.

The same blog cites other Google nuclear work:

  • Georgia Power: agreement to support nuclear uprates across four reactor units, which Google says is expected to enable approximately $900 million in ratepayer benefits.
  • NextEra Energy / Duane Arnold in Iowa: collaboration to restart the Duane Arnold Energy Center, bringing over 600 MW of carbon-free nuclear power online, per Google.
  • Kairos Power and TVA in Tennessee: partnership to deploy advanced nuclear reactors and reinvest in Oak Ridge, per Google.

Google also references a U.S. Department of Energy goal of enabling 5 GW of domestic nuclear expansion by 2030 through uprates and restarts, and says its agreements are expected to enable nearly 25% of that goal by the end of the decade. Again, that is Google’s projection in its blog, not an independent DOE scoring of this week’s deal.

For CTN’s beat, the Constellation package is the PJM-specific, 890 MW uprate story dated 6 October 2026. The Georgia, Iowa, and Tennessee items are prior or parallel Google nuclear moves that Google itself lists for context.

What buyers and grid watchers should ask

If you buy power for compute campuses, finance generation, or watch PJM capacity, the announcement raises practical questions the releases do not close.

  1. Offtake mapping: Which Google loads, in which PJM zones, sit behind the 20-year PPA? Soft hold: not disclosed publicly in the materials CTN used.
  2. Price and structure: What is the contract price path, and how do capacity payments, energy payments, and any curtailment rights split? Soft hold: price undisclosed.
  3. Delivery schedule: What is the MW-by-year plan from first uprate in 2028 through end-2032?
  4. Unit list and NRC path: Which of the 11 units file which uprate applications, and on what dates?
  5. Interaction with the 2,700 MW supply deal: How does the existing-fleet agreement interact with capacity-market offers and with Google’s load growth?
  6. Ratepayer tests: What filings will show whether residential customers are insulated from uprate costs as the companies claim?
  7. Demand response: What loads can Google curtail, under what notice, and how does that bind during PJM emergency procedures?
  8. Cloud/OT scope: Which Constellation OT networks move onto Google Cloud platforms, and under what air-gap and NERC CIP controls?
  9. Distinction from other Google power news: Confirm this package is separate from Google’s Oct 2 space-sat Brief item and from unrelated hyperscaler power stories the same week, including Applied Digital’s Finland power-access Brief on Oct 6.

None of those answers appear as hard public metrics in the Press Corner release or the Google blog. The right next step for operators is a direct ask, not a model built on silence.

What this piece is not

This piece is not a plant-level construction schedule. It is not a priced offtake model for a named Google campus. It is not investment advice on Constellation (Nasdaq: CEG) or Alphabet. It is not coverage of Google’s Oct 2 space-satellite Brief, and it is not coverage of Applied Digital’s Oct 6 Finland power-access Brief.

CTN does not invent campus MW, PPA prices, or independent job audits. Where Google claims a cumulative “>1.5 GW” of enabled U.S. nuclear from uprates and restarts, that claim is attributed to Google’s blog.

Disclosure: Lone Cowry / Cowry Tech News has no commercial relationship with Google or Constellation.

Sources

  • Google Cloud Press Corner, “Google and Constellation Announce Landmark Agreement to Bring 890 MW of New Nuclear Capacity to PJM Grid as Part of Long-Term Power Deal,” 6 October 2026: https://www.googlecloudpresscorner.com/2026-10-06-Google-and-Constellation-Announce-Landmark-Agreement-to-Bring-890-MW-of-New-Nuclear-Capacity-to-PJM-Grid-as-Part-of-Long-Term-Power-Deal
  • Amanda Peterson Corio, “Why we’re backing America’s existing nuclear plants,” Google blog, 6 October 2026: https://blog.google/company-news/why-were-backing-americas-existing-nuclear-plants/
  • Constellation Energy, matching news release, 6 October 2026: https://www.constellationenergy.com/news/2026/10/google-and-constellation-announce-landmark-agreement-to-bring-890-mw-of-new-nuclear-capacity-to-pjm-grid.html
  • ANS / Nuclear Newswire, “Google, Constellation reach 11-unit uprate plan; Constellation to invest $4.3 billion,” 6 October 2026: https://www.ans.org/news/2026-10-06/article-8467/google-constellation-reach-11unit-uprate-plan-constellation-to-invest-43-billion/
  • CNBC, “Google squeezes more power from existing nuke plants in Constellation deal,” 6 October 2026: https://www.cnbc.com/2026/10/06/google-constellation-energy-ceg-nuclear-power-data-center.html

Related reading: Vultr / HPE AMD Helios order · CoreWeave Vera Rubin NVL72 · Distinct from the Oct 2 Google space-sat Brief and the Oct 6 Applied Digital Finland power-access Brief.