⚡ THIS WEEK'S SIGNAL
Seven hyperscalers signed a pledge in March promising they wouldn't push their power costs onto your grandmother's electric bill. Six weeks later, state legislators filed more than 300 bills suggesting they'd like that in writing, notarized, and ideally with collateral. The federal-voluntary, state-mandatory split is the central question now for anyone planning generation, financing infrastructure, or quietly hoping their next rate case doesn't become a ballot issue.
━━━━━━━━━━━━━━━━━━━━━━━
📰 MAIN STORY: The Pledge and the Paperwork
Voluntary commitments tend to age in two directions. They either become the floor of a regulatory regime — or they become the thing the regulatory regime was built to replace. The "ratepayer protection pledge" Amazon, Google, Meta, Microsoft, Oracle, OpenAI, and xAI signed on March 4 is now firmly in the second category, and the speed of that shift is the actual story. In six weeks, statehouses across more than thirty states filed over three hundred bills addressing data center load growth. That's not a coincidence. That's a posture.
The hyperscalers want one thing above all else: speed. Every quarter a gigawatt slips on a campus build is a quarter of training runs, customer contracts, and earnings narrative. The pledge buys them political cover while they keep negotiating bespoke arrangements with utilities — special rates, capacity reservations, behind-the-meter generation that doesn't quite make it into anyone's IRP. State legislators want something different: a paper trail. They've watched Virginia's residential rates climb, watched Ohio approve gas plants for unnamed loads, and concluded that "we'll pay our full share" needs a definition before the next rate case, not after.
Here's the contrast that frames the policy fight. Microsoft and Constellation last year inked a roughly $1.6 billion deal to restart Three Mile Island Unit 1 — 835 megawatts dedicated to a single customer. In the same window, Dominion filed for a residential rate increase in Virginia, citing transmission build-out partly driven by data center growth in Loudoun County. One customer gets a reactor restarted in their name. The other customer gets a line item.
The data is the part you'll want to screenshot. Three hundred-plus bills in thirty-plus states in roughly forty-two days — about seven bills a day, with more than half clustered in states already hosting hyperscale campuses: Virginia, Georgia, Texas, Ohio, Oregon, Iowa. Hyperscale data center load is on track to reach roughly 9% of U.S. electricity demand by 2030, up from about 4% today. Combined capex commitments from the seven signatories run north of $320 billion across 2025-2026. The pledge itself has no enforcement mechanism, no audit requirement, and no definition of "full share."
Two scenarios, both trackable:
Scenario A: Virginia (or Georgia, or Ohio) passes a bill mandating a separate rate class for loads above a 100 MW threshold, with cost-causation language attached. Watch the Virginia General Assembly's reconvene calendar and the SCC's docket through Q3. If one hub state gets this through, the voluntary pledge converts into a regulatory floor and the template spreads to peer states inside twelve months. Hyperscalers absorb the cost shift; utilities get the cleanest rate-case story they've had in a decade.
Scenario B: The bills get watered down in committee, the pledge holds as the de facto standard, and the next stranded-cost or capacity-shortage event becomes the forcing function instead. In that world, the question isn't whether ratepayers pay. It's which ratepayers find out first.
One thing to watch: Dominion's next IRP filing with the Virginia SCC, expected in Q3 — specifically whether it carries a separate hyperscale tariff or treats large loads inside the existing GS-4 framework. That single document tells you which scenario we're in.
━━━━━━━━━━━━━━━━━━━━━━━
PRESENTED BY [SPONSOR]
[Sponsor copy — one-line value proposition, two sentences of context, CTA.]
━━━━━━━━━━━━━━━━━━━━━━━
⚡ QUICK HITS
PJM Capacity Auction Clears at Another Record: The 2027/28 base residual auction cleared at price levels that, depending on the zone, will flow through to residential bills as a noticeable line item by next summer. For utilities, this is the second consecutive auction making it easier — not harder — to justify new gas peakers in rate filings. [Source: PJM auction results, verify exact clearing prices before publishing]
xAI's Memphis Site Adds Another Phase: xAI filed for additional gas turbine capacity at its Memphis facility, expanding the behind-the-meter footprint that initially drew air-quality scrutiny from local regulators last year. The pattern — first the load, then the generation, then the permit conversation — is now the operating template for any campus that can't wait for an interconnection queue. [Source: TVA filings / local press, verify]
Georgia Public Service Commission Opens Cost-Allocation Docket: Following Georgia Power's request to recover transmission costs tied to large-load growth, the GPSC opened a generic docket on how hyperscale costs should be allocated across customer classes. Outcome here matters beyond Georgia — it's the cleanest test case in the Southeast for whether the pledge framework holds without statutory backup. [Source: GPSC docket filings, verify docket number]
EPRI Releases Updated Load Forecast Methodology: EPRI published revised guidance on how utilities should incorporate hyperscale load into their long-term planning, including treatment of speculative versus contracted load. For planners staring at IRPs due this year, the methodology change is the difference between a defensible forecast and a regulatory rejection. [Source: EPRI publication, verify release date and content]
First Energy Storage Tariff for Data Center Co-Location: A regulated utility (likely in the MISO footprint based on recent filings) is proposing a tariff specifically for paired storage-plus-data-center configurations, allowing capacity contribution to count against demand charges. Worth tracking because it's the first attempt to price the storage-load interaction inside a formal tariff rather than a special contract. [Source: verify utility and state]
━━━━━━━━━━━━━━━━━━━━━━━
🔧 TOOL / RESOURCE OF THE WEEK
NCSL Energy & Utility Legislation Tracker: The National Conference of State Legislatures maintains a searchable database of state energy bills filterable by topic, state, and status. With three hundred-plus data center bills moving in parallel, this is the only practical way to see which ones are actually advancing versus which ones are press releases with bill numbers attached. → ncsl.org/energy
━━━━━━━━━━━━━━━━━━━━━━━
💬 CLOSING THOUGHT
The pledge isn't worthless — voluntary frameworks have moved markets before, and the fact that seven competitors agreed on anything in writing is itself a signal. But for the optimism to hold, we'd need to see two things in the next two quarters: a hyperscaler publicly accepting a separate rate class without litigation, and a state PUC issuing an order that uses pledge language as a benchmark. Until then, it's a press release with good intentions.
Here's what I'd actually like to know from someone in the seat: if you're at a utility right now structuring a special contract with a hyperscale customer, are you assuming the pledge gives you cover, or are you pricing in the regulatory backstop that's clearly coming? Genuinely curious which way the room is leaning.
