Four instruments decide what a stalled phase 2 costs you, and you have signed all of them.

Current as of September 14, 2026. The PUCT Project 58481 rule discussed here is a Staff Recommended Adoption Order. Confirm the final text before acting on it. Revision log at the foot of this article.

Part of a five-article series on phased Texas data center campus development and the commitments sized to a buildout that may not arrive.

• Part one. The exposure: why a compliant tenant is the expensive one
• Part two. Where the exposure bites: security, tariffs, hedges and covenants
• Part three. Sizing the shared works against an uncertain phase 2
• Part four. Public commitments sized to full buildout
• Part five. Recovery once the phase has already stalled

The short version, for the capital committee.

---

Part one established why a compliant tenant that declines phase 2 is worse for you than one that defaults, and what the present value loss comes to. This article is about the specific instruments through which that loss arrives, because each one has a clock, a trigger and a counterparty, and most campus teams have signed all four without reading two of them.

Posted security, and a number most analyses report backwards

Under ERCOT's Batch Zero process for large load interconnection, established through PGRR145 and NPRR1325 for loads of 75 MW and above and partially implemented on 11 July 2026, an interconnecting large load entity posts financial security tied to contracted peak demand. Planning Guide Section 9.2.1.1(1)(e)(vi) states that requirement as $50,000 per MW of peak demand, revised down from an earlier $100,000 per MW to track the PUCT's parallel work in Project 58481. On a 300 MW contracted position that is roughly $15 million at full contracted peak.

That is refundable security, not a fee. Separately, SB6 provides for a flat transmission screening study fee of at least $100,000, which is a different and much smaller instrument. Do not model them as two per-MW charges.

Establish early whether your security posts in full at commitment or phases against energization milestones, because it changes the exposure materially during the period that matters most, and phase 1 of a phased posting is a very different number from the headline.

Two things about that money matter more than the headline, and both moved recently enough that anything written earlier in 2026 is out of date.

Cash CIAC for direct interconnection costs is not recoverable through regulated retail rates, so it does not come back. That part has been stable throughout.

The treatment of posted security on a missed phased energization milestone is the part that changed, and it changed in the developer's favor on both axes. The March 2026 proposed rule used a six-month miss as the trigger and was widely read as forfeiting the large majority of remaining security. Industry objected that transformer and long-lead equipment cycles alone run twelve to twenty-four months, so a six-month trigger penalized supply chains rather than speculation. Staff's Recommended Adoption Order in PUCT Project 58481 sets the trigger at missing a phased milestone by twenty-four months.

Read the retention carefully, because it inverts the number that has been circulating. Widely published analyses describe an 80 percent forfeiture. That was the proposed rule. The adopted order reverses it: 80 percent of the security associated with allocated capacity is returned, and 20 percent is subtracted.

But the retention is tiered, and the tier that applies to a Batch Zero developer is the one to check. In the Commission's own words, for batch zero loads that meet the criteria in ERCOT Planning Guide Section 9.2.1.2(1), or loads in a study other than batch zero, the balance is returned "less outstanding amounts owed and less 20% of the security associated with the allocated capacity." For batch zero loads that do not meet those maturity criteria, the subtraction is 50 percent. A customer allocated 0 MW, or withdrawing before the study begins, gets a full return less amounts owed.

So the headline 20 percent is the favorable tier, not the universal one. Establish which side of the Section 9.2.1.2(1) maturity test your project sits on before you model the number, because the difference on a $15 million posting is $4.5 million.

It is a recommended adoption order rather than an adopted rule, so confirm the final text and its application to your specific request with interconnection counsel before the next capital release.

The direction of travel matters more than either number. A twenty-four month runway and a 20 percent retention is a materially softer regulatory backstop than the six-month, majority-forfeiture reading, which means the regulator is no longer doing much of your risk management for you. If the rule had held at six months, a tenant slowing its ramp would have hit a hard external wall that forced the conversation. At twenty-four months, a tenant whose credit facility tightened can slow the ramp well inside the grace period, and nothing external compels anything. You are still posting eight figures of security and non-recoverable CIAC against a ramp the tenant controls. The regulatory penalty that might have surfaced the problem has just been pushed out two years, which puts the burden back in your lease file, where it is not.

Your own utility agreement may already contain the clawback

This is the provision most campus teams have signed and not read.

Oncor's Tariff for Retail Delivery Service, in the standard facilities extension agreement, carries a non-utilization clause: the CIAC you paid was calculated on estimated contract kW, and if within four years after Oncor completes the extension the load measured by actual maximum kW billing demand has not materialized, Oncor may recalculate the CIAC on the demand actually realized. The difference becomes a non-utilization charge, invoiced and payable within fifteen days.

Read that against the scenario in this series. A tenant that takes phase 1, performs perfectly and declines phase 2 produces exactly the fact pattern the clause is written for: an extension sized to a contract kW number the campus never reaches. The charge lands on you, not on the tenant, on a four-year clock you do not control, and it is separate from anything in the ERCOT security framework. Check your own TSP's tariff for the equivalent term and the term length before the next capital release.

Your hedge may be sized to a campus you are not building

This one is two-sided, and the direction matters enough that getting it backwards would be worse than not raising it.

Construction lenders routinely require an interest rate hedge as a condition of closing, sized to projected drawdown. Covenants of the form "maintain Secured Hedge Agreements with coverage in a notional amount of not less than 50% of the outstanding principal amount of the Term Loans" are standard, and the notional schedule is set against the full facility. If phase 2 never funds, the hedge notional exceeds the drawn balance and you are paying fixed on money you never borrowed. Loan documents treat this as a defect to cure, typically capping notional at about 105 percent of outstanding principal and requiring you to unwind the excess. Hedge counterparties negotiate the mirror right, as Clifford Chance describes it, "a right to partially terminate the swap if the amount of the hedging exceeds a commercially agreed percentage of the outstanding debt."

Which way the unwind cuts depends entirely on rates, and right now it cuts in your favor. A pay-fixed swap gains value as rates rise. Medium-term Treasury yields are up roughly 100 basis points over the past year even with the Fed holding at 3.50 to 3.75 percent, so a developer terminating excess notional today is more likely receiving a payment than making one. The real cost of a stalled phase 2 here is the periodic drag of paying fixed on undrawn notional plus commitment fees on undrawn capacity, not a headline breakage loss. In a falling-rate environment the same mechanic reverses and becomes expensive.

The actionable detail is the instrument, not the forecast. Standard over-hedging covenants carve out caps expressly, on the logic that a cap cannot go against you the way a swap can. If your draw schedule depends on a phase you do not control, that carve-out is the reason caps are conventional for construction loans with uncertain draws. Hedging a lesser share of each anticipated draw does the same work. This is established project finance mechanics applied to a campus context rather than a documented campus phenomenon, so treat it as a question for your treasury function rather than a finding.

What your lender asks when phase 2 slips

At the first covenant test after the slip, the lender asks where reimbursement comes from, whether the phase 2 commitment is contractual or an option, and whether the borrower can service the oversized asset on phase 1 revenue alone.

If the honest answer to the third is no, you have a refinancing event. Where the shared works sit in a dedicated entity, that event lands in a vehicle whose assets the operating neighbors depend on for service, and the campus has a single financial failure point even though the buildings are physically separate. Where common assets sit at the developer parent or inside the phase 1 entity, the exposure is real but differently placed. Know which structure you are in before the test, not during it.

Two smaller mechanics belong in the same register. A queue position carries cost, is use-it-or-lose-it, and took longest to obtain, so right-sizing the substation after phase 2 slips does not get that position back on the same terms. And a long-term lease is enforceable against a tenant whose expansion capital has become expensive, protecting rent on the leased premises, but it does nothing to compel an expansion option to be exercised. An unexercised option on phases 2 through 4 is what the shared assets were sized for.

Do not ask for the structure. Diagnose it from where they fight.

Requesting financing-structure disclosure at LOI does not work. You are one of several sites in a competitive process, the counterparty's real estate team cannot authorize disclosure of group capital structure, and the question reads as a signal that you will be difficult. Refusal is also not information, because everyone refuses.

Ask instead for terms whose cost depends on the structure: a parent guarantee on the expansion option, a requirement that the expansion commitment sit at the same entity as the phase 1 lease, and a capacity reservation deposit against reserved megawatts. Each reaches past the signing entity into whatever holds the expansion capital.

Then read the counter-offer, and nothing else. A letter of credit proposed in place of a parent guarantee, or a commitment offered at a named affiliate rather than the signing entity, is the counterparty showing you the relevant part of its entity chart in order to close. A counter-offer costs something to make and its content is a balance-sheet fact.

Timing tells you nothing. Escalation is the base rate for any non-standard term with a large counterparty and is routinely produced by approval calendars and tax counsel review. A clean balance sheet behind a slow treasury function looks identical to a complicated structure. Do not read it.

Attach the decision before you have the answer, so you are not negotiating with yourself afterward. Rated parent that signed: size the growth increment closer to the stated plan. Sponsor vehicle, joint venture or an entity you cannot see through: build phase 1 plus the smallest sensible increment, hold the rest at design and permit stage, price the reservation so the option carries itself. On a 300 MW campus the difference between those two postures is tens of millions of dollars of timing.

One note on counterparty type, because it inverts the whole approach. The indirect method above is a workaround for a top-tier hyperscaler where your leverage is thin. Against a neocloud or sponsor-backed operator, which is where this risk actually concentrates, you are frequently one of few sites that can deliver the power and a parent or sponsor guarantee is a live commercial ask. Ask directly there.

The capacity release provision, in detail

If you win one negotiated term, win this one. It costs the counterparty nothing on their balance sheet, which is why it is gettable, and it is the only instrument that converts a silent stall into a dated decision.

A provision that actually works has six moving parts.

The trigger. Not a calendar date. Tie it to the event that creates your exposure: a missed funded-trigger date on the growth increment, or the utility milestone that puts your interconnection position at risk. A calendar date drifts with the project and gets amended. An external milestone does not.

The notice and cure. On trigger, written notice to the tenant with a defined window, commonly measured in weeks rather than months, to either exercise the expansion option, post the deposit that keeps it alive, or let it lapse. The cure has to have a price. A cure right that costs nothing is a free extension and the tenant will take it every time.

What lapses, and how much. Specify whether lapse releases the whole expansion block or a defined increment. Increment-based release is easier to negotiate and usually better for you, because it lets the tenant keep a realistic option while returning the capacity they were never going to take.

Your marketing rights on release. The provision must expressly permit you to market, reserve and lease the released capacity to a third party, including a competitor, without further consent. Tenants will try to make release consensual. A consent gate makes the whole clause decorative.

Tenant priority after release. This is what you concede to get the rest. A right of first offer on the released capacity for a defined period, or a most-favored pricing right if they re-take it within a window. It is genuinely valuable to them and costs you very little, because you keep the timing control that matters.

The utility and regulatory tie. Say explicitly what happens to the associated interconnection position, reserved water capacity and any public commitment when the block is released, including who bears any reallocation or reservation charge in the interim. This is the part most drafts omit, and it is the part that bites, because the capacity sitting on your balance sheet is frequently reserved with a third party under a separate contract that knows nothing about your lease.

On the clock: the twenty-four month ERCOT milestone makes this term more important, not less. A regulatory trigger that long will not surface a stalled phase on any timeline useful to you. Your own release mechanic is the only thing that will, and it is the reason to set your internal trigger well inside the regulatory one.

---

The next action

Establish your security tier. Whether your project meets the Planning Guide Section 9.2.1.2(1) maturity criteria decides a 20 or 50 percent retention. On a $15 million posting that is $4.5 million, and it is a documentary question you can answer this week.

Pull your TSP's facilities extension agreement and find the non-utilization clause. Note the clock length and whether CIAC recalculates on realized load. Most teams have signed this and not read it.

Ask treasury whether the construction hedge is a swap or a cap, and what notional it is sized to. If it is a swap sized to full drawdown and your draw schedule depends on a phase you do not control, that is a conversation to have before the next draw rather than after.

Draft the capacity release provision now, not at the next renewal. Six parts, and the trigger has to be an external milestone rather than a calendar date. It is the most valuable term in this series and it costs the counterparty nothing.

---

Revision log

September 14, 2026, revision 1. Initial publication. This article separates the instrument mechanics from the tenant-credit article, where they first appeared, so the security, tariff, hedge and covenant material can be read as a checklist rather than as a digression inside a longer argument.

---