Grid Board. Peak this week: n/a this week (October demand is seasonal and ERCOT's dashboard was not reachable at press time; the summer record is 91.1 GW on July 22, prelim) | Large-load queue: 474 GW (latest total I could source, mid-2026 reporting) | Houston Hub TB4: $35.83/MWh (7-day avg, Oct 16 to Oct 22, with 11 negative-price intervals) | Reserve margin: n/a this week (the November MORA published Sept 3 could not be opened to quote)

On Wednesday, Kinder Morgan told investors it had committed $4.2 billion to a new pipe across West Texas, built to feed a hyperscaler's power plants that do not exist yet. A day later, the press was describing the Texas Commission on Environmental Quality as the agency that had "paused" every data center permit in the state while it kept processing them.

Put those side by side and you get the week's theme: the state is deliberating about whether to let data centers connect, and the infrastructure that serves them is being contracted, permitted and poured regardless. Gas is the layer where commitments are hardest to reverse, so I will start there.

Kinder Morgan Commits $4.2 Billion to Gas for an Unnamed Hyperscaler

Kinder Morgan's third-quarter release, filed with the SEC on October 21, says its NGPL subsidiary reached a final investment decision on Permian Link: roughly 434 miles of new pipeline with about 2.8 Bcf/d of capacity, a total cost of about $4.2 billion (Kinder Morgan's share is about $1.6 billion) and an in-service date of the third quarter of 2030, subject to permits and approvals. The anchor is long-term take-or-pay agreements with what the filing calls "an A-rated hyperscaler" for new power generation. It does not name the customer, and neither will I.

The release also puts the company's expansion backlog at $14.4 billion, including $5.4 billion of newly added contracted pipelines for power and data centers. If you are wondering whether gas developers believe the load forecasts, look at the contract form: take-or-pay means the shipper owes the money whether or not the data center fills up.

A quick conversion, because I keep getting asked how a pipeline size maps to a campus. This is my arithmetic, not Kinder Morgan's. One Bcf/d of pipeline gas is about 43,000 MMBtu per hour. At a combined-cycle heat rate near 7 MMBtu/MWh, that fuels roughly 6 GW of generation, so 2.8 Bcf/d is about 17 GW on paper. At a simple-cycle or reciprocating-engine heat rate near 9, it is closer to 13 GW. Real projects will not hand one customer the whole pipe, and heat rate, cooling load and backup all shift the answer. The order of magnitude is the point: this is a pipeline sized for several gigawatts of new plants, delivering in 2030.

For anyone planning a gas-fed campus, the practical read is that firm pipeline capacity, not turbines alone, is becoming the scarce item, and the buyers with the best credit are locking it up four years ahead. A developer who arrives with a megawatt number and no molecule contract is now late.

Texas Paused Data Center Permits. The Paperwork Kept Moving.

Governor Abbott's September 21 letter told TCEQ to halt all permits sought by data centers until the ERCOT audit is complete. Last week's issue covered the early signs that the grid was not waiting. This week the question moved from whether the pause worked to what the agency does about it.

The baseline is an analysis from FracTracker and Public Citizen, reported by KUT on October 20, that counted 35 permitting actions on data center projects after the letter: 14 applications fast-tracked or assigned to engineers, 11 stormwater approvals, 7 emissions estimates certified and 3 public notices. The governor's press secretary, Andrew Mahaleris, answered that "No permits have been approved," and that TCEQ "has not taken and will not take any proactive actions related to the approval of data center permits." The two sides are describing different verbs. In the governor's office's reading, the pause covers approval, not processing.

The new development came October 22. TPR reported that TCEQ has shut down its automated electronic filing system. Applications now have to be handed to the agency physically so staff can determine whether they are data center related. Public Citizen, according to the same report, has documented at least 16 permits actually issued after September 21, including a construction stormwater permit near San Antonio on October 13. That figure sits in direct tension with the governor's office, and I would not call it settled until TCEQ reconciles its own records.

What this changes for a developer is process risk. A project whose paperwork was already in the system may be treated differently from one that has to be filed on paper, and nobody outside the agency can yet say how staff will decide what counts as data center related. Plan on slower, less predictable environmental permitting through at least ERCOT's audit filing, targeted for December 10.

Batteries Set a Quarterly Record While the Queue Cooled

The least dramatic news of the week may be the most useful. Modo Energy's third-quarter buildout report puts ERCOT's operating battery fleet at 18.7 GW and 33.4 GWh as of the end of September. The quarter added 2.2 GW, the most in any quarter on record, and 4.7 GW has come online year to date against 4.2 GW at this point in 2025.

The other half of the report is the slowdown. Only 2.4 GW of new storage entered the interconnection queue in the quarter, the lowest since 2020. Modo puts average battery revenue at $24 per kW-year through mid-August, against $28 in 2025 and $193 in 2023. Developers read that price signal accurately: once the fleet is big enough to flatten the evening spread, the next project has to clear a lower bar than its predecessors did, and fewer of them do.

Set that against the data center side. Storage is being built into a market where its own revenue is shrinking, while gas is being contracted for 2030 on take-or-pay terms. Those are two different views of what ERCOT will need, and only the batteries are already on the system. Houston's weekly Hub TB4 spread averaged $35.83/MWh with 11 negative-price intervals, which is the kind of week that makes the case for flexible load.

Three Clocks, One Grid

The mismatch is in timing. Permits and the audit move in weeks. The pipeline is a 2030 asset. Batteries arrive by the quarter. A developer siting load in ERCOT now has to make a decision that fits none of those clocks, and the 474 GW sitting in the large-load queue is mostly a record of people who made it on optimism. ERCOT's December filing is meant to show how many of roughly 300 Batch Zero projects survive verification.

My read is that the projects still standing in 2027 can show three things on paper: a firm fuel or power contract with a creditworthy counterparty, a permit file that survives a physical filing and a hearing, and a load ramp that does not assume the queue forecast. Everything else is still a place on a list.

What to Watch Next Week

TCEQ's records. Any reconciliation between the agency's count of approvals and the 16 permits Public Citizen says were issued. If the agency's number moves, the whole "pause" narrative moves with it.

The hyperscaler's name. Kinder Morgan will not say, but pipeline construction needs rights of way, county filings and eventually a regulatory docket. Those tend to name names well before 2030.

ERCOT's December 10 target. The Batch Zero eligibility report and the community impact review are the next hard milestones. The date is a target, not a commitment, and slippage would show up first in staff remarks at PUCT open meetings.

The winter MORA. The month-ahead assessment is the cleanest public read on winter tightness. I will carry the figure onto the board once I can quote it from the document.

Disclaimer: The Grid Report is Barrio Energy's market intelligence product. Nothing here is investment advice, and Andi is not your broker. Links go to primary sources wherever possible; form your own view.