Seattle’s 369‑MW Data‑Center Moment: The One Clause Utilities Are Adding That Could Let AI ‘Factories’ Cut Power First (and Your Neighborhood Pay the Price)
Five proposed data centers could add 369 MW of maximum demand—enough to reshape staffing, infrastructure, and contracts. The real fight: who funds upgrades, and whether new “AI factory” loads are interruptible before households when the grid is stressed.
Key Points
- 1Five proposed data centers could add 369 MW of maximum demand, a peak load big enough to reshape City Light’s planning and operations.
- 2Watch the contracts: a curtailment/interruptibility clause could decide whether “AI factory” loads get cut before households during grid stress.
- 3Follow the money: proposed policies push upfront upgrade payments and “bring your own energy,” but cost-shift questions remain for upstream work.
Seattle is staring at a number that rarely shows up in local politics until something breaks: 369 megawatts.
That is the combined maximum demand of five proposed large-scale data centers now under review by Seattle City Light, according to reporting published April 10, 2026. City Light told Axios the proposals come from four companies, though the utility declined to identify them publicly at the time. The headline isn’t simply “more data centers.” It’s that a handful of new customers could arrive with an appetite that changes the rules of the system.
Axios described the 369 MW figure as about one-third of Seattle’s typical daily electricity use—a comparison that’s striking even if the underlying daily-load figure isn’t published in that item. City Light officials also warned the proposed step-change could strain staffing and infrastructure used to serve existing customers. In utility language, that is not hand-wringing. It’s a feasibility problem.
“In grid planning, 369 MW isn’t a talking point. It’s a re-write.”
— — TheMurrow Editorial
Behind the scenes, the fight isn’t only about whether Seattle wants data centers. It’s about who pays for the wires and substations, and whether the contracts will include a simple but consequential condition: when the grid is stressed, who gets curtailed first.
Why 369 MW hits like breaking news in a hydro city
A data center proposal is not like a new apartment tower. The load is concentrated, continuous, and unforgiving. When a utility sees a large new customer, it must ask whether the local network—feeders, substations, protective equipment—can deliver power reliably at the requested maximum. The check is not philosophical. It’s engineering and operations.
Maximum demand vs. average use: the part that confuses everyone
For readers, the key point is simple: the grid must be sized to the peaks, because the peak is when things fail.
Why a few customers can dominate the conversation
- New substations or substation expansions
- Heavier distribution feeders
- Protection and control upgrades (to keep faults from cascading)
- Staffing needs (design, construction management, operations)
City Light’s warning to Axios about potential strain on staffing and infrastructure is a tell. Even if the region has plenty of energy in the abstract, the local system still needs the hardware and people to deliver it.
“Energy is not the same thing as deliverability. The wires are the city.”
— — TheMurrow Editorial
The grid reality: you can’t “just add” 369 MW
Start with interconnection. Bringing on a major load is not only about whether generation exists somewhere on the map. It’s also about whether the local grid can carry that power without voltage problems, equipment overloads, or creating a single point of failure.
What planners have to prove before they say yes
- Can the existing substation serve the requested load, or does it need new transformers?
- Are the feeder lines capable of carrying the current year-round without overheating?
- Will fault currents exceed what breakers and protection systems can safely handle?
- How many months—or years—will design, permitting, and construction take?
Axios reported City Light is reviewing the five proposals now. The review phase itself hints at the scale. When a single project might require new infrastructure, utilities slow down and verify everything twice.
The operational side: not every risk shows up on a map
That point deserves attention in a city that already asks its public agencies to do more with less. A public utility can’t hire and train specialized engineering talent overnight. Reliability is built with time and repetition.
“Reliability isn’t a vibe. It’s headcount, hardware, and lead times.”
— — TheMurrow Editorial
The “one clause” question: will data centers be curtailable when the grid is stressed?
Many utilities already have some form of emergency authority. The policy fight is more precise: whether large new loads will be placed, explicitly, into a category that can be interrupted before firm residential and small business load—and what the customer receives in exchange.
What we can document in Seattle right now
- Secure their own energy, rather than relying on City Light’s portfolio
- Pay for grid upgrades needed to serve them
Separately, City Light’s City Light Review Panel materials (meeting packet dated June 18, 2025) outline a “Proposed NLL (New Large Load) Rate Policy.” The packet states that it would apply to:
- New load larger than 10 MVA, or
- Growth of 10 MW+
Under that proposed policy, the customer would sign a Power Service Agreement and make upfront payment of all costs of building new infrastructure.
Those are not small conditions. They signal a utility trying to protect existing ratepayers while still keeping the door open to new development.
What we cannot yet prove—and why it matters
That gap matters because curtailment is political in practice. If an agreement explicitly says “interruptible,” residents will want to know when and how often interruption can occur, and what happens if interruption threatens public safety, housing, or critical services.
The next step for real accountability is straightforward: obtain and publish the draft Power Service Agreement template or the tariff/ordinance text that defines the priority stack.
Key Insight
The cost-shift fear: who pays for the wires that serve “AI factory” loads?
When a utility builds expensive upgrades sized for a giant new customer, a key question follows: are the costs assigned to the customer causing them, or socialized into rates for everyone?
City Light’s stated direction: pay upfront, bring your own supply
- Large new loads may have to pay upfront for the infrastructure that connects and serves them.
- Large customers may be required to secure their own energy.
Those are classic tools to prevent cost shifting. They also impose real discipline on developers: if a project only pencils out when the public subsidizes grid expansion, maybe the project doesn’t pencil out.
But “upfront” is not the end of the story
- Who pays for upstream upgrades that benefit multiple customers?
- What happens if the data center doesn’t materialize after upgrades are started?
- If the load grows again, who pays the second time?
These questions are the difference between “fair” and “politically explosive.” Seattle doesn’t need to be anti-data-center to insist on clean accounting and enforceable commitments.
What “cost shifting” really means here
Reliability, equity, and the social contract of a public utility
Axios reported City Light officials warned the proposals could strain the utility’s ability to serve existing customers. That is an equity issue as much as an engineering one. Outages and service constraints are rarely distributed evenly. The people least able to tolerate disruptions—low-income households, medically vulnerable residents, small businesses—often have the fewest alternatives.
“Curtailment first” sounds simple; implementing it isn’t
- What counts as a grid emergency?
- Who decides, and how transparent is that decision?
- How frequently can curtailment be used without turning “interruptible” into “unreliable”?
- What protections exist for critical community infrastructure nearby?
If City Light does not make them interruptible, the city must answer a different question: why should households carry any risk for a private load that is large enough to shape the system?
The honest answer may be that Seattle will need a mix of tools—pricing, interconnection requirements, and emergency rules—rather than a single silver bullet clause.
The economic case for data centers—and why it doesn’t settle the debate
Yet electricity is not like other inputs. The grid is shared. A decision to add 369 MW of maximum demand is a decision to commit physical capacity—and organizational attention—to serving that demand.
What supporters will argue
- Seattle’s clean-energy reputation and the idea that growth should happen where power is cleaner
- The potential for large customers to pay substantial revenue into the public utility
- The possibility of negotiated terms that protect residents while allowing development
Those points deserve a hearing. A public utility must plan for a future economy, not only the past.
What skeptics will keep pressing
- Reliability risk and the burden on City Light staffing and project delivery
- Cost shifts, especially if upgrades become “system” costs
- The basic optics of dedicating scarce capacity to private computing demand
The unresolved question is whether Seattle can create a framework that makes large loads pay their way, operate with meaningful flexibility, and avoid undermining reliability for everyone else.
How the debate tends to split
Pros
- +Cleaner grid reputation; major utility revenue potential; negotiated terms could protect residents while enabling growth
Cons
- -Reliability and staffing strain; risk of cost shifts via “system” upgrades; optics of scarce capacity serving private computing load
What readers should watch next: the documents, the deadlines, the deal terms
The three documents that will tell the real story
- Especially any provisions on curtailment/interruptibility, performance requirements, and termination
2. The finalized New Large Load (NLL) Rate Policy mechanics
- What triggers it (the Review Panel packet cites >10 MVA new load or 10 MW+ growth)
- What costs are paid upfront, and what costs become “system” costs
3. Interconnection and construction timelines
- How long City Light says upgrades will take
- Whether the projects demand new substations or major feeder work
Practical implications for Seattle residents
- Reliability: If City Light staffing is stretched, restoration times and maintenance schedules can suffer.
- Transparency: City Light declined to identify the companies in the Axios reporting. If the projects move forward, public disclosure will become harder to avoid.
Seattle doesn’t have to choose between welcoming growth and protecting the public. It has to insist on terms that make those goals compatible.
What to track for accountability
- 1.1. Get the draft Power Service Agreement template (or tariff/ordinance text) that defines the curtailment priority stack.
- 2.2. Verify the NLL policy triggers and how “upfront” costs are separated from “system” costs.
- 3.3. Follow City Light’s interconnection timelines and whether upgrades require substations, feeder work, or major protection/control changes.
Frequently Asked Questions
What does “369 MW of maximum demand” actually mean?
Maximum demand is the highest amount of power a customer could draw at once, measured in megawatts. It’s not the same as energy used over time. Utilities must build and reserve enough capacity to meet that peak safely and reliably, which is why maximum demand can drive costly upgrades even if average use is lower.
How big is 369 MW in Seattle’s context?
Axios reported that 369 MW is about one-third of Seattle’s typical daily electricity use. Even as a rough comparison, it signals scale: five proposed facilities could represent a major new block of demand concentrated in a small number of connections, with real consequences for infrastructure and operations.
Who is proposing the data centers?
According to Axios (April 10, 2026), the five proposals come from four companies, but Seattle City Light declined to identify them publicly at the time of reporting. If the projects proceed, public processes and permitting often make corporate identities easier to confirm.
Can Seattle City Light require data centers to pay for upgrades?
City Light’s June 18, 2025 City Light Review Panel materials describe a proposed New Large Load (NLL) Rate Policy requiring a Power Service Agreement and upfront payment of all costs of building new infrastructure for qualifying loads (new load >10 MVA or growth of 10 MW+). Axios also reported City Light is rewriting large-user contracts that could require payment for upgrades.
What does “secure their own energy” mean?
Axios reported City Light’s rewritten contracts could require large users to secure their own energy. In practice, that generally means the customer contracts for supply rather than relying on the utility’s existing portfolio. The specific obligations—how much, what type, and how verified—depend on the final contract terms.
Will data centers be the first to lose power during an emergency?
That’s the key unresolved detail. Utilities often use curtailment frameworks, but the publicly available City Light materials cited here describe policy direction rather than the precise “priority” clause. The decisive evidence would be the actual Power Service Agreement language or tariff text specifying whether the load is interruptible and under what conditions.
