02 · Transformation
Is AI driving explosive growth yet
Verdict: 1 strong · 1 mild · 7 neutral.
No clear sign of takeoff, which matches what Stanford finds for the US. The one strong signal is the money the UK spends importing cloud and compute. Core output and productivity look ordinary.
The grid reads left to right by year, from 2022 onward. Hover any cell for that year's value and its evidence level; the charts below show each indicator against its pre-AI average.
Data & methodology
This track asks whether AI is yet showing up in the wider economy as fast, capital-led growth, the kind of break from the past that Nordhaus (2021) called an economic singularity.
How each indicator is scored
Every series is compared with its own 1997 to 2019 average, before AI, and scored by how far above that norm it now sits, measured in standard deviations and in the direction that would signal takeoff. Growth series are shown as year-on-year change; share series are transformed so that rising always means more transformation. The four levels are strong, mild, neutral, and contradictory. This is a test against history, not a forecast.
The series and their sources
Output, productivity, the labour and capital shares, trade and investment come from the ONS; the real interest rate from the Bank of England; electricity from DESNZ. Three of the nine are annual official statistics whose latest complete year is 2024 (computer-services imports, multifactor productivity, and electricity); the rest run to late 2025 or early 2026. The 2025 annual figures are not published yet.
Why it is tuned for the UK
The UK buys AI more than it builds it, so the clearest signal is computer-services imports, the cloud and compute bought from abroad. Domestic-capital measures are weaker here than in the US, because the hardware is imported, so we read them with that in mind.