UK's £400 billion AI forecast came from a Google-commissioned report

For almost two years, UK ministers and government agencies said AI could add £400 billion to the economy. Last week AI minister Kanishka Narayan said in a parliamentary answer that the figure came from a 2023 report commissioned by Google. The government's own central estimate is £120 billion, less than a third of that number.

The gap matters because the larger figure was used to argue for public spending and policy decisions under two governments. WIRED reports that ministers had the lower estimate since at least November 2025. That was almost a year before the government set the two numbers side by side.

Where the £400 billion came from

In July 2023 Google published its annual economic impact report for the UK, produced by the British consultancy Public First, which now operates as PF Global. In a blog post announcing it, Google said that "AI-powered innovation could create over £400 billion in economic value for the UK economy by 2030." The same report estimated that generative AI could save the average UK worker more than 100 hours a year. Google framed the projection as conditional on "the right conditions".

In his answer to Parliament, Narayan said the figure "was drawn from a publicly available 2023 report commissioned by Google and produced by Public First." According to WIRED, this is the first recorded instance of the government naming Google as the commissioner of the report. The answer did not withdraw the £400 billion figure and did not say it had been wrong.

Former technology minister Peter Kyle was among those who used it when promoting AI. WIRED found that Skills England, the agency responsible for apprenticeships, was still publishing the old figure four months after ministers had the lower estimate.

The government's own estimate

Narayan said AI adoption and data investment could add between £55 billion and £140 billion to the UK economy by 2030, with a central estimate of £120 billion. He described even this "more recent assessment" as "hugely uncertain".

PF Global's head of economics, Jonathan Dupont, told WIRED that the gap is partly explained by adoption. "Compared to our older model, the numbers the government uses assume a much slower rate of adoption than looked possible back in our model that was created in 2023," he said. PF Global says it has not changed its view of the technology's overall potential. But given the slower take-up seen in practice, its central estimate for 2030 is now "much closer to the Government's number," with growth rising in the years after.

The firm said its work for Google continues, but that its research is independent and its judgments and assumptions are its own. It pointed to more top-down adoption and integration of AI into civil service workflows as ways take-up could improve.

Questions about how the number was used

Lord Holmes of Richmond, a Conservative peer, asked ministers about the origins of the £400 billion claim in January 2025 and did not receive a response. He told WIRED it "would have been better for the government to clarify this much sooner and more directly," adding that "billions of pounds of our money and how the government chooses to deploy it" depend on robust data.

Dame Chi Onwurah, the Labour MP who chairs the Commons Science, Innovation and Technology Committee, said hype is one of the central barriers her committee has identified to digital transformation of public services. "Overstating the benefits of technology risks undermining the case for change," she said.

Sumedha Deshmukh of the Ada Lovelace Institute said the episode reflects a wider pattern: figures about AI that look precise often carry built-in assumptions and incentives, and industry often funds or shapes the research behind them. She said the government should test the numbers it receives, fund independent research and measure whether promised gains arrive. The Cabinet Office and Google were approached for comment by WIRED.

What it means for work

Both estimates rest on the same mechanism: gains arrive only when organisations actually change how tasks are done. Most of the difference between £400 billion and £120 billion is a difference in assumed adoption speed, not in what the technology can do. PF Global's own suggestions point to where that adoption is expected: routine workflows inside the civil service and other large employers. Google's 2023 claim of more than 100 hours saved per worker each year depends on people using these tools daily, which the slower take-up suggests is not yet happening at scale. What remains open is which tasks are being handed to AI in practice, and whether anyone is measuring it. Without that measurement, forecasts like these will keep shaping budgets and training plans before the evidence does.