25 standards walk into a facility.

Three teams. Three tags on the same pump. All correct. None matching.

The German design contractor labelled it under KKS. The European operations team tagged it under RDS-PP following ISO/TS 81346-10. The plant owner's maintenance system used an internal code. The piping and instrumentation diagram used one schema. The electrical single-line diagram used another. The fire safety documentation used a third.

Nobody made a mistake. They followed the rules. The rules just do not agree.

A single data centre must comply with 15 to 25 standards simultaneously. A nuclear plant exceeds 40. An LNG terminal sits somewhere in between.

Each standard was developed by experienced engineers solving a real problem. Each assumes it is the primary classification system. Each conflicts with the others at the boundaries.

Here is a sample. One facility. Five disciplines. Five different answers to the question: what do we call this piece of equipment?

Now multiply across 30,000 documents, three subcontractors, four geographies, and a commissioning deadline.

The shift supervisor from Edition 3 does not need to know which committee wrote which standard. She needs to find the right breaker. The commissioning engineer from Edition 2 does not need a 200-page compliance matrix. He needs the wiring diagram that matches the equipment in front of him.

The standards exist to make facilities safer and more interoperable. In practice, at the point of handover, they make them harder to operate.

Fire safety says the server room must be sealed for gas suppression. Cooling says it must have airflow for thermal management. Electrical says access panels must be reachable for maintenance. Security says access must be restricted. Each is correct. Combined, they create a coordination problem that no single engineering team can resolve from memory.

This is why handover takes 12 to 18 months. Not because the documents are missing. Because the documents disagree about what to call the same equipment, how to classify it, and which standard takes precedence when two conflict.

Across the 467 facilities Vistergy tracks spanning nuclear, data centres, and LNG, every one of them operates under multiple overlapping standards regimes.

In Japan, a nuclear plant follows JEAC for electrical systems, JSME for mechanical, and IAEA safety standards for regulatory compliance. In France, EDF's fleet operates under ASN regulation, RCC-E and RCC-M design codes, and ISO 55000 for asset management. In Brazil, Petrobras LNG terminals reference ABNT standards alongside API and ISO. In Singapore, data centres comply with BCA codes, SS 564, and Uptime Institute certification simultaneously.

Different countries. Different sectors. Same structural problem. The standards were written independently. They are applied simultaneously. And the conflicts surface during commissioning, when the cost of delay is highest.

Edition 1 put a number on that cost: £79m ($100m, €96m) per day for a 1 GW AI factory.

It is not because nobody tried. The international standards community has been working on interoperability for decades. CFIHOS developed a comprehensive handover specification for process plants. ISO 15926 created a lifecycle data model for oil and gas. ISO 19650 standardised information management for BIM. ISO 55000 established asset management principles. ISO 18136, currently under development, aims to unify industrial data across the full lifecycle.

Each of these was developed by brilliant engineers. Each solves a real problem within its domain. But each was developed by a different committee, in a different decade, for a different primary use case. They were never designed to work together at the point of handover.

The result is not a technology gap. It is an architecture gap. The standards exist. The harmonisation layer between them does not.

Or rather, it did not.

Edition 4 described the team that compressed 18 months of handover into 4 weeks. Here is what made it possible: they built an extraction engine that does not require the standards to agree first.

The engine reads every format, every naming convention, every classification schema. It ingests the conflict rather than resolving it in advance. KKS, RDS-PP, internal codes: all normalised automatically. 1,000 documents per hour. No universal schema required. No committee approval needed.

That is the difference between the approach that was shelved for years and the approach that cleared a three-year backlog in one year. The standards do not need to agree. The system needs to understand all of them.

I have spent five editions diagnosing the problem. The documents that disagree. The billions lost while facilities sit dark. The shift supervisor trusting a drawing that was already wrong. The standards that conflict with each other.

Diagnosis is necessary. But it is not sufficient.

Starting next week, I am launching a podcast called Permit to Operate. It lives in the same gap as this newsletter: between digital delivery complete and operational readiness. But where Still Dark diagnoses the problem, Permit to Operate explores the solutions. The people who close the gap. The systems that work. The decisions that get facilities from still dark to fully operational.

Episode 1 of “permit2operate” will be live soon…

Two publications. One thesis. Same gap.

Edition 6: The energy facility does not know the AI factory exists.

Edition 7: The grid was not built for what is coming.

Edition 8: The gap is not closing. It is getting wider.

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.

Permit to Operate.

David Ayeni

The world needs millions of new homes. Traditional construction, built on manual labor and decades-old methods, was never going to get us there.

Azure Printed Homes is the alternative. Our LA factory runs three robots that print modular structures, automate steel truss manufacturing, and set ground screw foundations. One home every 24 hours. Move-in ready in about 20 days. Starting at $40,000.

In 2024, we hit $5M in revenue. We currently hold $62M in pipeline orders, all backed by deposits.

That's a proven model. Now we're raising capital to open a second factory in Denver in 2026, which doubles production capacity.

Housing will be built differently. We're already doing it.

See How It Works on Wefunder