14AUG

Texas grid queue hits 410 GW

4 min read

10:48UTC

ERCOT told Texas legislators its large-load queue has reached 410 GW, roughly 87 per cent data centres, even as transformer lead times stretch to 128 weeks.

← Back to GPUs now collateral as Nebius borrows $775m Jump to analysis ↓

IndustryDeveloping

Key takeaway

ERCOT's queue hit 410 GW, but a 128-week transformer wait decides what can actually connect.

ERCOT told Texas legislators its large-load interconnection queue has reached 410 GW, with roughly 87 per cent of it data centres ¹. ERCOT, the Electric Reliability Council of Texas, runs about 90 per cent of the state grid. The figure comes from its own presentations to the Senate Committee on Business and Commerce on 1 April and the House Committee on State Affairs on 9 April. It is up from the 225 GW ERCOT logged weeks earlier , and it sits at nearly five times the state's 85 GW peak demand.

The requests keep coming; the iron to serve them will not arrive in time. PwC analysts, via pv-magazine, put average large power transformer lead times at 128 weeks, with some orders quoted at four years ². That tracks GE Vernova's backlog, which grew to 100 GW with deliveries booked into 2029 before the firm added $5bn of transformer capacity through its Prolec deal . An order placed now arrives in 2028 at the earliest.

Industry trackers put close to half of US data-centre builds already delayed or cancelled on supply-chain grounds. ERCOT's connection-friendly rules let developers queue cheaply, so the 410 GW figure measures appetite, not committed demand. The transformer book measures what can actually be fed. When requests run at 410 GW and deliverable hardware runs at a fraction of that, the queue stops measuring intent and starts measuring who secured a slot.

Deep Analysis

In plain English

ERCOT is the company that runs the electricity grid for most of Texas. It told state lawmakers that companies have put in requests to connect 410 gigawatts of new load, almost all of it data centres. To put that in perspective, the entire state only uses about 85 gigawatts at its peak. The problem is that the equipment needed to handle all those connections, called large power transformers, now takes up to four years to build and deliver. So even if ERCOT approved every request tomorrow, most of them could not actually receive power until 2028 or later. The queue measures how many companies want to build; the transformer wait decides how many actually can.

Deep Analysis

Root Causes

ERCOT's connect-first interconnection design was built for predictable industrial load growth in a state whose regulatory philosophy favours market entry over prescreening. Applications require a study deposit of roughly $10,000 to $100,000 per MW, trivial against the $1bn-plus capital cost of a hyperscale campus. That entry price creates a queue that measures ambition and option-holding rather than committed demand, pushing the 410 GW figure far beyond any plausible build rate.

Separately, the large power transformer supply chain runs on a fundamentally different timescale from data-centre planning cycles. A 500 kV autotransformer requires roughly 14 months of lamination steel processing, nine months of core winding, and six months of factory testing before shipment.

The total 24-to-30-month manufacturing window cannot be compressed because the steel supplier and the transformer manufacturer are different firms. GE Vernova's Prolec acquisition adds $5bn of production capacity, but Prolec's Mexican plants must also tool up for the US specification transformers that ERCOT's grid requires. The supply constraint is structural over a five-year horizon, not a temporary backlog.

What could happen next?

Risk
Transformer lead times of 128 weeks mean that a large share of the 410 GW queue cannot receive power before mid-2028 regardless of regulatory approval, creating capital-at-risk for developers who ordered hardware in advance of grid connection.
Short term · Assessed
Consequence
ERCOT's queue overhang will drive a secondary market in interconnection queue positions and transformer delivery slots as speculative entrants face deposit deadlines, rewarding operators with early-mover hardware contracts.
Medium term · Suggested
Opportunity
Jurisdictions with shorter transformer queues and cleaner grid-connection pathways, such as Finland's Fingrid or Arizona's APS, gain a supply-chain arbitrage advantage over ERCOT for near-term hyperscale builds.
Short term · Suggested

Source Landscape

This story draws on neutral-leaning sources

Primary parallel: During the US natural gas boom of 2000-2001, FERC's wholesale electricity markets attracted a wave of merchant gas-plant interconnection requests that collectively exceeded US capacity by roughly 200 per cent. The queue congestion triggered FERC Order 2003 in 2003, which overhauled interconnection procedures.

The physical analogue was not transformers but gas turbines, where GE Power's lead times stretched to 24 months and speculative orders inflated manufacturer books before cancellations followed. The ERCOT data-centre queue reproduces that pattern with a harder physical ceiling: gas turbines could be cancelled and reallocated, but large power transformer slots, once booked, are harder to redirect due to bespoke specification requirements.

Counter-parallel: The UK's National Grid ESO (now NESO) ran a comparable demand-queue surge in 2024-25, with its connection register reaching 125 GW by mid-2025 against a 45 GW peak demand. NESO responded with its Gate 2 queue reform , requiring firm evidence of planning consent before a connection offer, which immediately reduced the speculative element.

ERCOT adopted analogous reforms in 2025, but the 2025 vintage applications already in study carry the old light-touch rules, so the 410 GW figure reflects a mix of old-rule and new-rule applications.

Nvidia's Q1 2027 (May 2026) data-centre networking revenue of $14.8bn, up 199 per cent year-on-year , reflects the accelerating capex cycle that feeds ERCOT's queue. But the connection between chip orders and grid connection is not automatic: a campus that orders H200 racks in Q2 2026 cannot power them if the transformer it needs is a 2028 delivery. Capital tied to hardware without grid connection generates zero revenue and accumulates financing costs.

Wood Mackenzie estimates that each month of commissioning slippage on a 100 MW hyperscale build costs the operator roughly $12m to $15m in delayed revenue. On a 410 GW queue that is almost entirely at risk of slippage, the aggregate revenue deferral figure runs to tens of billions of dollars. That pressure creates a secondary market for transformer slots and interconnection queue positions, which ERCOT's current rules do not prohibit.

Consensus view: SemiAnalysis and Wood Mackenzie both assess that ERCOT's connect-first queue design, which charges applicants a small study deposit rather than requiring a firm capacity commitment, incentivises speculative filing. SemiAnalysis's grid team calculates that roughly 60 per cent of projects in mature queues like PJM never reach commercial operation; ERCOT's recent design reforms target the same dynamic but have not yet cleared the 2025 and 2026 vintage applications.

Construction Physics author Brian Potter, tracking transformer physics, argues that the binding constraint has already shifted from grid consent to switchgear and transformer manufacturing: a steel-core large power transformer requires 12 to 14 months of core lamination alone, and global cold-rolled grain-oriented steel capacity cannot be ramped in under five years.

Counter-view: Hitachi Energy's grid integration team and GE Vernova's commercial division argue that transformer scarcity is real but manageable for committed buyers willing to secure slots two to three years ahead. Hitachi Energy published order data in Q1 2026 showing its own backlog at approximately $8bn with 18-month average lead times for distribution-class units, shorter than the power transformer figure PwC cited.

Uptime Institute's global infrastructure research adds that many large-queue requests overstate actual MW demand by 30-40 per cent due to peak-over-provisioning in early planning phases, so the deliverable gap is narrower than the 410 GW headline suggests.

Key tension: Whether the queue's speculative inflation will correct naturally as deposit deadlines bite, or whether transformer supply bottlenecks persist long enough to constrain even genuine projects with firm financing by 2028.

Sources:ERCOT·KWTX·KXXV

Mentions:PwC →ERCOT →GE Vernova →Prolec →Texas →National Grid →Uptime Institute →SemiAnalysis →NESO →Hitachi Energy →FERC →Wood Mackenzie →PJM Interconnection →

First Reported In

Update #5 · Who pays when the grid bends for AI

ERCOT· 2 Jun 2026

Read original →

Causes and effects

Caused by

ERCOT large-load queue passes 225 GW

ERCOT's 225 GW queue baseline from April 2026 (ID:3055) was the prior data point; the 410 GW figure represents a 76% step-change disclosed at the same legislative hearings.

Occurred 1 Apr 2026

Read story →

GE Vernova gas backlog hits 80 GW into 2029

GE Vernova's 80 GW backlog (ID:2790) established the transformer supply crunch; the 128-week lead-time figure quantifies the same constraint against the 410 GW queue.

Occurred 1 Dec 2025

Read story →

This Event

Texas grid queue hits 410 GW

The queue now runs at nearly five times Texas peak demand, but transformer scarcity caps how much can physically connect before 2028.

Led to

Texas queue swells to 438 GW

ERCOT's queue grew from 410 GW in April to 438 GW by June, the 28 GW increase that triggered the Batch Zero approval

Occurred 18 Jun 2026

Read story →

Transformer waits stretch to four years

Transformer lead times have extended from the 128-week ERCOT baseline in April to 36-48 months reported in May, worsening the supply-chain constraint

Occurred 11 May 2026

Read story →

Different Perspectives

Antel, Uruguay's state telecom

Antel will build a $70m AI data centre in Canelones, announced after an 11-12 August Council of Ministers meeting, with construction from Q1 2027 and roughly 250 direct jobs. Uruguay is putting a sovereign AI facility on the state balance sheet rather than courting a hyperscaler to build one.

Comtel

Comtel is building a Figline Valdarno data centre inside the H2-Era Green Valley complex, powered by an 84 MW solar plant and selling waste heat to on-site farming. Co-location with generation and a heat offtaker is the model regulators elsewhere are now writing into connection terms rather than negotiating case by case.

Greg Abbott, Texas Governor

Abbott directed PUCT and ERCOT on 3 August to audit every queued data centre's ownership, incentives and water use, saying projects that fail the state's survey should lose grid access. QTS's co-chief executives welcomed the review; Agriculture Commissioner Sid Miller called it too vague and wants a special legislative session instead.

Government of Gujarat

Gujarat notified a Data Centre Policy on 7 August offering capital, interest, tax and duty subsidies for projects sourcing at least 51% green power, targeting 7.5 GW against a 74 GW installed base. Where other jurisdictions priced or restricted grid access this month, Gujarat is paying developers to come instead.

Taiwan's Legislative Yuan Economics Committee

The committee gave preliminary approval on 5 August to a self-generation mandate for large power users, including data centres, with fines up to NT$750,000. It agreed the obligation in principle while leaving the megawatt threshold that defines its reach to later implementing regulation.

Australian Energy Market Commission

AEMC advised ministers on 5 August that data centres should surrender renewable certificates against actual consumption, prove new demand with new firm capacity, and register as market participants. The advice would turn passive customers into obligated participants inside the market's dispatch and settlement machinery, ahead of an access-standards determination due in late October.