Phased Energization:
Negotiating Partial Power

When the utility says a full service request cannot be met for years, most owners hear a verdict. It is better heard as an opening position: electric capacity is not delivered as a single event but in layers, and the layers can be negotiated separately.

Section 01Why partial power exists at all

A request for new or expanded electric service is not one question. It is a stack of questions, asked of different pieces of equipment with different amounts of spare room. Can the service drop and metering handle the load. Can the local circuit. Can the substation transformer bank behind the circuit. Does the transmission system behind the substation need reinforcement. Each layer has its own headroom, its own upgrade cost, and, critically, its own clock.

California's regulator has now put public numbers on those clocks. In September 2024 the California Public Utilities Commission adopted energization targets for the large investor-owned utilities, implementing the Powering Up Californians Act. For standard requests served under the line- and service-extension rules, the decision sets an average energization target of 182 calendar days. Where the request forces capacity work, the same decision recognizes how long that work takes: its maximum timelines run to 684 calendar days for a new or upgraded circuit, 1,021 for a substation upgrade, and 3,242 calendar days where a new substation must be built.¹

One clock is measured in months. The other is measured in the better part of a decade. The distance between them is the entire case for phased energization: a request that lands wholly in the slow bucket waits on the slow bucket, while a request engineered to split across the layers can draw real capacity from the fast layers while the slow ones are built. Utilities do this routinely for loads that genuinely ramp, because a data hall, a plant line, or a campus rarely needs its ultimate load on opening day. The practical question is whether the split is designed deliberately, by the owner, around the owner's build schedule, or arrives by default as whatever the utility's study happened to produce.

Transmission-level customers now have a parallel track with the same logic. In July 2025 the Commission approved interim implementation of Pacific Gas and Electric Company's proposed uniform rule for transmission-level retail service, conditioned on applicants agreeing to fund necessary transmission infrastructure up front, with the refund methodology deferred to a later decision.⁴ Speed is purchasable there; certainty, as Section 04 discusses, is not yet part of the product.

Section 02The phasing conversation, from the utility's chair

To negotiate partial power well, start with what the counterparty is solving for. The utility has an obligation to serve, but it plans against a pipeline in which announced load is a probability, not a fact. Some announced projects shrink. Some slip years. Some vanish. Network capital committed to a phantom becomes stranded cost with a regulatory audience, so planners discount aggressive ramp claims as a matter of institutional survival. When an applicant presents a nameplate total as a day-one requirement, the utility studies the nameplate, prices the nameplate, and quotes the slow clock. Both sides lose.

Credibility is therefore the real currency of a phasing conversation, and it is earned with evidence, not emphasis:

• A ramp you can document. Equipment delivery schedules, fit-out phases, hiring plans, occupancy dates. A load forecast with named milestones behind it gets studied differently from a round number with a press release behind it.
• The minimum viable tranche. The capacity below which the site cannot open at all. This is the most important number in the negotiation, and in our experience most applicants have never actually computed it; they know what they want, not what they need.
• A dated deadline per phase. With the business consequence of missing each date priced, so that trade-offs between speed and firmness can be evaluated rather than felt.
• A flexibility inventory. Which loads can shed, shift, or ride through on short notice. Flexibility is tradable for time, and programs now exist to formalize exactly that trade.⁵
• A bridge concept. What the site is prepared to self-supply, store, or defer. The tranche you must have from the grid on day one depends directly on what you can credibly do without it.

The machinery for this engagement is more formal than it used to be. The 2024 energization decision requires the large California utilities to track requests through a defined eight-step framework, maintain customer engagement plans, report progress publicly twice a year, and face a customer delay-reporting channel at the Commission.¹ Use that machinery. Ask early, ask in writing, and get the capacity finding for your site dated and documented, because everything else in the negotiation is built on it.

Section 03What is actually negotiable

One honest caveat first. Tariffs are filed rules, and a regulated utility cannot cut side deals against them; similarly situated customers are entitled to similar treatment. What is negotiable is everything the rules leave open: how the request is structured, sized, staged, secured, and sequenced. That space is larger than most applicants assume, and it is where the phasing outcome is decided.

Two levers deserve emphasis because they are the ones owners most often leave unpulled. The first is the study scope behind tranche one: the difference between a generic screen and a study that examines the specific circuit, the specific bank, and modest make-ready options can be the difference between a token allocation and an opening tranche the business can actually run on. The second is the sunset on flexibility terms: interim curtailment obligations that quietly survive the upgrades they were meant to bridge are a recurring, and avoidable, contract failure.

Section 04The money: contributions, allowances, and the unbuilt phase

The commercial architecture underneath a phased arrangement in California is set by the line- and service-extension rules. In simplified terms: the utility funds extension work up to an allowance keyed to the revenue the new load is projected to produce; costs beyond the allowance are the applicant's to pay; payments and contributed facilities are treated as taxable contributions in aid of construction and carry an income-tax component on top; and facilities that are dedicated to one customer or oversized at the customer's request are special facilities, funded by that customer.^2,3 The exact arithmetic is tariff arithmetic, applied to a specific design, and this paper deliberately deals in categories rather than quotes. But the categories interact with phasing in ways worth understanding before the first meeting.

• Allowances follow projected revenue, and phasing changes the projection. A load that arrives over years produces a different revenue forecast, and potentially different allowance treatment, than the same load presented as day-one demand. Get the allowance treatment of each phase in writing rather than assuming the full-buildout allowance arrives with tranche one.
• Build-early and build-late both cost money, differently. Sizing trenches, ducts, and structures for the ultimate build in one mobilization is usually cheaper in lifecycle terms, but it pulls the applicant's contribution, and its tax component, forward onto capacity that may sit idle for years. Building per phase defers cash and preserves optionality, at the price of remobilization, repeated permitting, and repeated outage windows. Neither answer is generally right; the choice should be priced both ways against your actual confidence in the later phases.
• Cancellation mechanics run in both directions. If a later phase dies, the arrangement will say what happens to contributions already paid and to utility capital already spent; if it does not say, that silence favors whoever drafted it. Underutilization and termination provisions are not boilerplate. They are the allocation of the project's largest single risk, and they deserve counsel's time before signature.
• At transmission level, refund policy is still being written. The interim framework trades upfront applicant funding for acceleration, with the refund methodology deferred to a future decision.⁴ Funds advanced under a policy in motion should be priced as a risk position, not booked as a receivable with a date on it.

A useful discipline for this section of any deal: treat every dollar figure you hear in conversation as unverified until it appears in writing under the applicable rule, and treat every category above as a question to be answered in the agreement, not after it.

Section 05Bridging the tranche you did not get

Once tranche one is fixed, the unserved remainder becomes a defined problem: a known capacity, needed for an estimated number of years. That is a materially better-shaped problem than "no power," and it is the problem the bridge must solve. Two questions do most of the sorting. First, is the gap energy-shaped or peak-shaped: does the site lack kilowatt-hours around the clock, or only capacity during certain hours. Second, what is the honest range of the phase-two date, because the bridge must be economic across that whole range, not at its optimistic end.

• Shaping the load itself is the cheapest bridge and the most overlooked. Sequencing fit-out, staging occupancy, and scheduling flexible processes off-peak can shrink the gap before any equipment is bought. The case against is candid: deferral has a business cost, and that cost belongs in the model, not the footnotes.
• Storage fits peak-shaped gaps well: it permits quickly, runs silently, and pairs naturally with interim-flexibility terms, charging in the hours the connection allows and discharging when the site is constrained. The case against: duration is finite, recharge depends entirely on the tranche you do have, and a bridge lasting years is an augmentation program rather than a purchase. Storage shifts energy; it does not create it.
• Combustion generation, reciprocating engines and gas turbines or microturbines, deploys fast, costs comparatively little per unit of capacity, and draws on deep service and rental ecosystems; turbines add usable exhaust heat where the site wants it. The case against is heaviest in California: strict air districts impose demanding control requirements, operating-hour limits, and in some cases offset obligations on non-emergency units, and a "temporary" combustion fleet that runs for years invites both regulator and community attention. Noise, fuel logistics, and part-load efficiency belong in the comparison.
• Non-combustion generation reverses those trades. Fuel cells run quietly at high electrical efficiency with near-zero criteria-pollutant emissions, which in several districts means a materially lighter permitting path for a multi-year bridge; against that stand higher capital cost, exposure to gas price, stack-lifecycle costs, and procurement lead times that must be shorter than the gap to be relevant. Linear generators are fuel-flexible, modular, and fast-starting, with the diligence burden shifted to fleet history and service depth, examined rather than assumed. Solar with storage burns nothing and buys no fuel, and for daytime-weighted gaps it can carry real weight; its capacity factor and footprint mean it rarely carries a continuous gap alone.
• Rental fleets deliver the two things a bridge exists for, speed and reversibility: the equipment leaves when the grid arrives, and capital stays off the balance sheet. The case against: the highest operating cost per unit of energy on this list, strict air-district treatment of non-emergency rental engines, and fuel contracts that become your operations problem for as long as the bridge runs.

The design rule that ties the family together: choose bridge assets by their exit plan. Every candidate should have a written answer to the question of what happens to it when phase two energizes: returned, redeployed to another site, resold, or rolled into a permanent resilience role. And permit the bridge for the pessimistic duration, not the hoped-for one, because the bridge's end date is the least certain date in the entire project. Which brings us to the queue.

The first tranche is engineering. The later tranches are promises, and promises age inside a queue.

Section 06Later phases inherit the queue

Load service is not always a formal queue in the generator-interconnection sense; depending on the process it may be a pipeline, a study cluster, or a capacity ledger. The mechanics rhyme regardless: finite headroom, ordered claims, and restudy when the system changes. A phased arrangement places your later tranches inside that machinery for years, and three erosion mechanisms deserve explicit attention.

Shared headroom erodes. The capacity finding behind a future tranche describes a system that other applicants, and organic load growth, keep drawing from. Unless the arrangement states how tranche-two capacity is held, and what holding it costs, the finding is a snapshot, not a promise. Ask the uncomfortable question directly: if a competing request arrives next quarter, what, precisely, protects my phase two.

Projects slip. Later tranches typically wait on physical works: circuits, banks, substations, transmission reinforcements, each with procurement lead times on long-cycle equipment, its own permits, and its own place among the utility's other commitments. The state's timeline framework caps and reports the steps utilities control, and its biannual reporting and delay channels give owners real visibility.¹ Visibility is worth having; it does not pour concrete.

The rules themselves are moving. The transmission-level framework approved in July 2025 is explicitly interim, with cost-refund methodology still to be decided.⁴ Nationally, in June 2026 the Federal Energy Regulatory Commission directed the six jurisdictional regional grid operators to justify or reform their tariff provisions for integrating large loads.⁶ Reform can shorten dates; it can also reshuffle study assumptions mid-stream. A multi-year phase schedule signed today will live through rule changes nobody can currently draft.

Protection, honestly stated, is partial. Written milestones tied to named projects, a standing monitoring cadence with a named utility counterpart, contractual review triggers if dates move beyond a stated tolerance, and the regulator's reporting machinery together create early warning, leverage, and a record. None of them creates an entitlement. The professional posture is to treat the phase-two date as a distribution rather than a fact: run the project's economics at an early, an expected, and a late date, and size the bridge, and the bridge's permits, to survive the late one. A phased project that only works if phase two arrives on time is not a phased project. It is a bet dressed as a plan.

Section 07Six questions to settle before you sign

A phased arrangement is a package: an engineering finding, a schedule, a financing structure, and a risk allocation, all wearing one signature block. Before that signature, six questions should have written answers.

1. What exactly is tranche one? Which facilities serve it, what the capacity finding says, when it was dated, and what events invalidate it.
2. What does each later tranche depend on, by name? Which circuit, bank, substation, or transmission project; whose scope it sits in; and what stage that project has actually reached today.
3. Who pays for what, and when? Allowance treatment per phase, contribution and tax-component categories, special-facilities scope, refund terms, and the fate of money already paid if a phase cancels, in both directions.^2,3
4. What binds you? Contract demand, ramp commitments, deposits, underutilization and termination provisions, and the cure paths when the business ramps slower than the paper says.
5. What does interim flexibility really cost? Curtailment windows, notice, event frequency, compensation if any, and a sunset that actually extinguishes the obligation when the upgrades land.
6. What does the bridge cost across the honest range of phase-two dates? Priced at early, expected, and late, with a written exit plan for every bridge asset.

None of these questions requires an adversarial tone; utilities generally prefer counterparties who ask them, because answered questions become durable projects instead of disputes. What the questions do require is an analysis run from the owner's side of the table, by a party with nothing to gain from which technology bridges the gap or how large the first tranche turns out to be. Phasing is where time-to-power stops being a grievance and becomes a design problem. Designed well, partial power is not a consolation prize. It is the fastest honest path to a running site.