On Thursday, October 8, the Federal Energy Regulatory Commission officially designated Google Energy as a Category 2 seller of wholesale electricity in PJM. The order is worth more attention than it will probably get: it is the regulatory acknowledgment that one of the world’s largest power buyers now controls more than 500 megawatts of generation capacity inside the nation’s largest wholesale market — and that the market’s rules for watching large sellers now apply to a company whose core business has nothing to do with selling electricity.
The designation matters less as a headline than as a mechanism. To see why, it helps to know what “Category 2” means in the architecture of market-based rate authority — the permission slip FERC grants companies that want to sell wholesale power at market-negotiated prices rather than cost-based rates. This is the bureaucratic plumbing beneath the buyer-to-seller transition reshaping power markets, and Google just became its most prominent example.
What Category 2 status actually means
FERC does not hand out market-based rate authority unconditionally. Every seller seeking it must demonstrate it lacks the ability to exercise market power — horizontally, by controlling enough generation to move prices, and vertically, by controlling transmission that could block rivals. Sellers are then sorted into two categories that determine how much paperwork they owe the Commission going forward.
Category 1 covers wholesale power marketers and producers owning or controlling 500 MW or less of generation in aggregate per region, owning only minimal transmission, and not affiliated with a franchised utility in the region. Their main privilege: exemption from filing updated market power analyses every three years. Category 2 covers everyone else — all other power sellers — who must continue filing those triennial studies, submitting to periodic reporting, and proving on a regular cycle that their activities are not stifling competition or exercising market power. Both categories must file change-in-status reports when their circumstances change.
Google Energy has held market-based rate authority for years as essentially a power marketer — a buyer and reseller of wholesale electricity, not an owner of plants. What changed is the second half of that definition: with Google’s contracted generation in PJM now exceeding 500 MW, the company no longer qualifies for the lighter-touch category. The reclassification is, in that sense, automatic. The number crossed the threshold; the status followed. The order went through its public notice period with no formal protests or adverse comments from competing utilities or consumer groups, according to the Commission’s order.
For market watchers, the interesting part is what Category 2 obliges going forward. Google Energy must now file periodic reports on its wholesale activities and pass the triennial market-power review — the same screen every large independent power producer faces, testing whether its generation position in PJM (including affiliated capacity) could be used to withhold supply or manipulate prices. In the piece that opened this series on wholesale markets, we described the independent market monitor as the permanent watchdog function of organized markets; the Category 2 designation adds Google to the list of sellers that watchdog is watching.
The 500-megawatt threshold — and how Google crossed it
The capacity that pushed Google over the line did not arrive by accident. It is the accumulation of a procurement campaign that has accelerated sharply over the past year and change, as Google moved from buying clean energy credits to contracting for utility-scale projects connected directly to the regional grid.
The largest single piece landed two days before the FERC order. On October 6, Google and Constellation Energy announced a 20-year power purchase agreement under which Constellation will uprate 11 of its existing nuclear units across six sites in Illinois, Pennsylvania, and New Jersey — adding 890 MW of new nuclear output to the PJM grid, with the first additional capacity expected in 2028. Constellation said it will invest more than $4.3 billion in the uprate program, sustaining roughly 4,400 existing jobs and creating about 7,200 construction jobs. In parallel, the two companies signed a separate 15-year supply agreement covering 2,700 MW from Constellation’s existing PJM fleet — bringing the combined package to 3,590 MW. Constellation described the deal as a direct response to PJM’s Bring Your Own Power proposal, a connection we will return to.
An uprate is not a new reactor. It swaps in new turbines, steam generators, and digital controls at already-licensed, already-connected plants so each produces more power — sidestepping the site hunt and the interconnection queue, and arriving at a cost (just over $4,800 per kilowatt across the program, per Constellation) well below new build. The companies framed the 890 MW of incremental output as equivalent to several small modular reactors or one large conventional reactor. Within that program, the biggest single site addition is a 348 MW boost at Constellation’s Limerick plant near Philadelphia, with a 97 MW increase at the Salem plant in southern New Jersey, according to company figures.
The nuclear package sits alongside earlier deals that already made Google a major PJM capacity holder: a $3 billion, 20-year arrangement with Brookfield for up to 3,000 MW of hydropower capacity across the U.S., with the first contracts covering Brookfield’s Holtwood and Safe Harbor plants in Pennsylvania. Taken together — the 3,590 MW Constellation package, the hydro contracts, and other arrangements — Google’s generation footprint in PJM territory comfortably exceeds the 500 MW threshold that triggered the Category 2 designation.
One precision worth insisting on: neither Google nor Constellation disclosed the pricing of the power contracts themselves. The dollar figures in the announcements — $4.3 billion, $3 billion — describe investment in upgrades and facilities, not the price per megawatt-hour Google will pay. PPA prices in these hyperscaler deals are routinely kept private, and they should not be confused with the capital budgets.
Why a hyperscaler is building generation
The question that should interest power market analysts is not whether Google can afford this — it plainly can — but why a technology company chose to assemble a generation portfolio at all, rather than simply buying more electricity as a customer.
The proximate answer is policy: PJM’s Bring Your Own Power proposal. The concept would require large new loads — data centers above a 50 MW threshold entering PJM after mid-2027, under the proposal’s reported terms — to bring their own supply or face curtailment. Constellation explicitly framed the Google deal as a “direct response” to that proposal, and said it aligns with the White House’s ratepayer protection pledge. FERC action on the proposal is expected imminently. For a hyperscaler with gigawatts of data center load planned across PJM, building a contracted generation book is the cheapest way to guarantee compliance with a regime that may soon require it.
The deeper answer is the supply picture. PJM, serving more than 65 million people across 13 states and D.C., faces a reported shortfall of roughly 6.8 GW of new supply against surging data center demand, according to industry analysis. Capacity market prices in the region have climbed sharply since 2024 as large buyers — Microsoft, Amazon, Meta, and now Google — have moved to lock up reactor output. When the forward price of reliability is rising that fast, a buyer with Google’s balance sheet has every incentive to convert price exposure into contracted assets. Self-supply is a hedge, and at this scale it is also a procurement strategy.
Google has also made a point of the cost-allocation angle. Around the announcement, the company said it is covering Constellation’s costs to invest in new capacity, fund necessary transmission upgrades, and manage ongoing operations — paying the full cost of the generation its demand requires rather than socializing those costs to other ratepayers. That framing is aimed squarely at the consumer-advocate concern, raised around the FERC designation itself, that hyperscaler load growth could land on residential bills. Whether the structure fully achieves that separation will be tested in rate proceedings; the commitment is notable regardless.
What changes for PJM’s markets — and what doesn’t
The honest assessment is that the Category 2 designation itself changes little about PJM’s day-to-day market mechanics, while the commercial activity behind it may change a great deal over time.
What doesn’t change: PJM’s market power mitigation machinery operates exactly as it does for any other seller of this size. Offers from Google Energy’s affiliated generation, to the extent it bids into PJM’s energy and capacity auctions, will be subject to the same structural tests and offer caps — typically cost-based — that constrain every large supplier. The independent market monitor’s surveillance, described in our RTO explainer, applies regardless of whether the seller’s parent makes search engines or software. The triennial filings add transparency; they do not create a special hyperscaler exemption.
What might change: a Category 2 seller is a seller. Google Energy is now a market participant entitled to sell wholesale power — energy, capacity, and ancillary services — into PJM’s markets, not merely a buyer hedging its own load. Today its portfolio is clearly assembled for self-supply: matching contracted generation to data center consumption. But the optionality exists. If Google’s contracted output at any point exceeds its load — or if it chooses to optimize its book — the company can sell the surplus at market prices like any merchant generator. The transition from “buyer” to “potential seller,” as the FERC reporting framed it, is real, and it represents a genuinely new kind of participant in the wholesale market: a load-serving buyer whose primary business is neither generation nor retail.
The precedent matters for cost allocation too. If BYONC-style rules take hold and hyperscalers broadly self-supply, the capacity market’s demand curve — the mechanism by which PJM procures reliability for the system — will reflect large loads that have pre-purchased their own adequacy. That is not a failure of the capacity market; it is arguably the design working. But it shifts who bears the residual risk, and it raises questions about whether capacity accreditation and deliverability rules treat self-supplied loads fairly relative to everyone else. Those debates will play out in stakeholder proceedings over the next year.
The buyer-to-seller trend is bigger than Google
Google’s reclassification is the largest and most symbolically potent instance of a pattern now running across the hyperscaler set — and its timing makes the pattern unmistakable. Constellation signed a 20-year deal with Amazon the week before the Google announcement, underpinning a 190 MW expansion of the Calvert Cliffs plant in Maryland, part of a broader 690 MW arrangement. A year earlier, Constellation and Meta announced a smaller uprate deal adding 30 MW at one of its plants. Georgia Power and Google reached a separate arrangement adding 96 MW at two nuclear stations. Each deal follows the same template: long-term contract, existing fleet, direct grid connection, uprate rather than new build.
This is a structural shift in how the wholesale market acquires new supply. In the organized markets we described in this series, new generation historically entered through merchant developers responding to forward price signals or utilities building rate-based plants. Now the entrants are the load itself — the largest buyers in the market, bypassing the merchant middleman and contracting generation directly, at volumes that move the needle on regional supply. Uprates have become a repeatable product because they are fast, licensed, and connected; the interconnection queue, the great bottleneck of the transition, is simply irrelevant to them.
For the independent power producers that have been PJM’s traditional sellers, the arrival of hyperscaler capital is a double-edged development. On one side, it underwrites investment — $4.3 billion here, $3 billion for Amazon’s Calvert Cliffs expansion there — that might otherwise struggle for financing. On the other, the buyers are no longer price-takers in the capacity and energy markets; they are assembling portfolios that give them structural positions in those same auctions. The market monitor’s job just got more complicated, and so did the strategist’s.
What to watch next
Three things will tell us how consequential this designation turns out to be. First, Google Energy’s filings: the periodic reports and the first triennial market-power analysis will reveal exactly how the company is positioning its PJM generation — self-supply, merchant sales, or some blend — in a way no press release will. Second, the fate of Bring Your Own Power at FERC: if the proposal is approved, every other hyperscaler with PJM ambitions will face the same self-supply arithmetic that drove Google’s deals, and the buyer-to-seller pipeline will accelerate. Third, the cost-allocation proceedings: whether Google’s full-cost-coverage commitments actually keep ratepayer bills insulated will be settled in the unglamorous language of tariffs and allocation factors.
The FERC order itself is a formality — the paperwork catching up to a commercial reality. But formalities are how markets learn to see new things. For the first time, the nation’s largest wholesale market has a hyperscaler on its seller roster. The question for everyone else in the market is what that roster looks like in five years.