Available now
Report No.01
The UK AI Infrastructure Map
Where Britain's compute actually comes from — the fifteen-layer stack from chip-design software to electricity, the chokepoints where a single supplier can throttle the whole chain, the datacentres and AI Growth Zones now on British ground, and the committed private capital behind them.
- 65 entities
- 60 supply links
- 9 chokepoints
- 9 sites & zones
- £26.5bn+ capital tracked
Report No.02
The UK AI Exposure Map
Every occupation in Britain, how many people do it, and how far AI can reach into it — scored on two clocks: the office wave here now, and a later robotics era. With "The Quiet Subtraction" essay and an interactive assumptions engine: move the dials, watch your own scenario resolve against the mainstream projection.
- c.16m gross exposure today
- c.19m in the robot era
- 1–3m mainstream net by 2050
- 6 dials to move
Report No.03
The UK AI Money Map
Who has pledged what to UK AI, to which site, and how firmly — the commitment ledger from The Quernal's live dataset. Every pledge in its announced currency, dated, with its provenance and firming status on show. Deliberately not summed across currencies.
- £26.5bn+ UK commitments (GBP)
- 6 anchor pledges
- $30bn largest (Microsoft)
- 2 of 6 firmed to a primary
View the report ↗
Compiled 12 July 2026 · the commitment ledger as of the 5 July dataset snapshot
Report No.04
The UK AI Chokepoints Report
The handful of suppliers the whole stack depends on, and what has no substitute. Every datacentre in the Infrastructure Map sits downstream of four global single-source nodes — the lithography, the fabrication and packaging, the memory, and the accelerators — none with a domestic alternative.
- 9 critical dependencies
- 4 single-source nodes
- 0 domestic substitutes
- drill to source
View the report ↗
Compiled 12 July 2026 · a substitutability read of the sourced supply chain
Report No.05
The UK AI Power Report
Where the electricity actually comes from — the grid and energy constraint that now sets the ceiling on UK AI. The operational and permitted power draw on British ground, the generators and networks behind it, and the connection queue that has become the real gate.
- 50GW queuing vs 45GW peak
- £900m a year to run one (4× US)
- £10bn approved, under £1bn built
- 6GW needed by 2030
View the report ↗
Compiled 12 July 2026 · demand, the connection queue and the cost — sourced to Ofgem and National Grid
Report No.06
The UK AI Growth Zones Report
Culham, the North East, Slough, Elsham Wolds and beyond — the UK's AI Growth Zones as ground. What is designated, what power each has, its status and timeline, and the distance between what was announced and what is actually being built.
- 2 designated zones
- 4 sites mapped
- 2GW+ permitted + operational
- gap announced vs built
View the report ↗
Compiled 12 July 2026 · sourced to planning records and firm announcements
Report No.07
Sovereign Compute
The AI Research Resource, the national supercomputing spine, and the state's real position on public compute — owned, rented, or merely claimed. The thinnest report of the set, and honest about it.
- 1 public machine fully sourced (Dawn)
- 5 layers Britain doesn't control
- 85% of Europe's AI GPUs are Nvidia
- hours for one US letter to switch it off
View the report ↗
Compiled 12 July 2026 · lightest on the owned side, sharpest on the claim
How the reports are built
Each report is assembled from The Quernal's live AI-ecosystem dataset and from official sources — company filings, government publications, primary APIs. Every figure is sourced and dated. Nothing is modelled or inferred; where a number is programme-level, or not yet checked against a primary source, the report says so on its face.
Everything is drillable: click any datum and it opens to the entity, the supply link, or the source it came from. Reports are living assets — when the underlying record moves, the report is refreshed to a new dated version, and the change is logged. The date is always the real one; we never backdate.
How The Quernal is made