In partnership with
In a control room at the National Energy System Operator in London, an engineer opens this morning's connection queue. 738 gigawatts of unbuilt capacity sit ahead of her. While she works, three more applications arrive. By the time she signs off this evening, the queue is longer than when she logged in. The country's installed generation, today, is around 88 GW. The queue is more than eight times the country.
Most will never be built. Which of them are real, no one this morning can tell her.
Last week, the grid was the constraint. This week, the trajectory is.
I have spent seven editions on the gap between when infrastructure is built and when it works. The documents that disagree. The drawings that are already wrong. The standards that conflict. The facilities that do not see each other. The grid that was sized for something else.
Eight weeks in, the gap is not closing. It is getting wider.
Two curves pulling apart
Since Edition 1, the announcements have not slowed.
NVIDIA's 10 GW deployment with OpenAI, with up to $100bn invested progressively. The Lenovo Gigawatt AI Factories Program at GTC 2026. Stargate Abilene, on schedule, scaling from 1.2 GW past 2 GW. New 1 GW class commitments stack quarter on quarter.
The transmission corridors that would deliver this capacity, on the same timeline, do not exist. Grid build at the matching scale is not under construction anywhere.
The workaround is becoming explicit: when the public grid cannot deliver, hyperscalers buy their own. In September 2024, Microsoft and Constellation signed a 20-year PPA to restart Three Mile Island Unit 1 as the Crane Clean Energy Center, dedicated to Microsoft's data centres. In June 2025, AWS and Talen Energy signed a 17-year, $18bn PPA for up to 1,920 MW from the Susquehanna nuclear plant. Amazon has invested in X-energy and partnered with Energy Northwest on the Cascade small modular reactor facility, scaling to 960 MW. The largest energy customers in the world have stopped waiting for the public grid.
The build commitments are exponential. The coordinating capacity behind them is not.
Two curves on the same chart. Pulling apart.
Coordination runs on human time
A new nuclear plant takes six to ten years from commitment to commissioning. Korea Hydro and Nuclear Power delivers reactors on a six-year cadence at best. In France, EDF is commissioning Flamanville 3 after more than a decade in construction. The transmission corridor from a single substation requires years of routing, permitting, and build.
In the UK, the 738 GW queue is the public scale of the problem. In the US, Berkeley Lab's Queued Up reports document an interconnection backlog past 2,000 GW. The backlog spans every major regional operator.
Standards review committees meet in cycles measured in years. ISO updates a major lifecycle standard every five to seven. Commissioning engineers train across years. The handover specialists who can ingest 30,000 documents and structure them against 25 conflicting standards exist. They number in the dozens, not the thousands.
The US construction sector entered 2026 short by around 439,000 workers, most of them skilled trades. The 2026 hiring need is another 349,000 on top. Commissioning is the thinnest and most fragile talent pocket in the data centre stack.
The output of every coordinating layer is human capacity. The input is exponential demand.
The gap widens at every altitude
Inside the facility, documents grow faster than ingestion can keep up. The extraction engines that compress 18 months of handover into four weeks at scale exist. None has been deployed at every facility under construction.
Between facilities, the grid was wired for an older economy. The 1 GW campus arriving now will run flat out for decades. Transmission corridors built for a different load profile will not catch up before commissioning.
Across sectors, three different clocks tick at different rates. Nuclear plants take six to ten years to commission, and that is at the disciplined end. LNG terminals run three to five. Hyperscale data centres run twelve to twenty-four months from groundbreaking. No two schedules converge. Three sectors, three languages, one grid.
The gap is the same structural object at every altitude. It just gets wider.
The questions arrive faster than the answers
Across Atlas, the pattern is the one Edition 6 introduced. Nuclear, data centres, and LNG in one frame. Three sectors that share grid connections and decades of operational dependency. None of them, anywhere else in public, share a view of each other.
Edition 6 widened the rule. Every reactor, every data centre under construction, every LNG terminal: open, free, permanent. The questions that have arrived since are more specific.
Which 1 GW campuses sit within baseload range of an operating reactor with spare capacity?
Which transmission corridors are sized for the load now planned to flow through them?
Which LNG terminals feed grid nodes that will draw new compute load over the next 24 months?
Which sectors have commissioning timelines that need to coordinate, and where are they not coordinating today?
Those questions arrive faster than any single answer can be written. That is not a publication problem. It is the gap.
The gap is not closing
The number Edition 1 set down was £79m ($100m, €95m) per day for a 1 GW AI factory sitting dark. It reads now as the floor. Every new public commitment since has pushed the ceiling up, not down. The IEA projects global data centre electricity demand to more than double by 2030. Every measure on the demand side is rising faster than the matching coordinating function on the supply side.
Eight weeks of evidence are now on the chart. The two curves are still pulling apart. The build commitments since Edition 1 have not slowed. The coordinating capacity behind them has not accelerated. Every layer of coordination is bounded by human time. Every layer of demand has slipped that boundary.
The gap is not closing. What is closing is the time to act on it.
What comes next
Edition 9: The handover specialist does not scale.
Edition 10: The standards committee meets next year.
Edition 11: What the gap looks like five years out.
If you build, operate, invest in, or regulate infrastructure anywhere in the world, this is written for you. Subscribe to Still Dark.
This newsletter lives in the gap between digital delivery complete and permit to operate. That gap is where value dies, and where it can be recovered.
I also co-author The Vistergy Brief at vistergy.com/archive. Satellite and geospatial monitoring, facility lifecycle intelligence, and standards architecture across LNG, nuclear, data centres, utilities and construction. Subscribe to both for the full picture.
The podcast is live. Permit to Operate is available at vistergy.com/pod, Apple Podcasts, and Spotify.
Permit to Operate.
Investors see ANOTHER return from Masterworks (!!!!)
That’s 6 sales in 7 months. 29 all time. And the performance?
16.5%, 17.6%, and 17.8%, net annualized returns on sold works held longer than one year (See all 29 at Masterworks.com)
It’s not from stocks, private equity, or real estate… it’s from contemporary and post war art. Crazy, right?
With Masterworks, you don’t need to be a BILLIONAIRE to invest in multi-million dollar art anymore.
Historically, the segment overall has had attractive appreciation and low correlation to stocks.*
Masterworks targets works featuring legends like Banksy, Basquiat, and Picasso, identifying what they believe to have significant long-term appreciation potential, not just at the artist level but at the level of individual artworks.
As one of the largest players in the art market, with $1.3 billion invested over 500 artworks, they pass critical advantages through to their 70,000+ members to add art to their portfolios strategically.
Looking to diversify your investments in 2026?
*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.