UK minister says £400bn AI windfall claim came from Google-linked report

The UK’s AI minister has said the government’s long-circulating estimate that artificial intelligence could add £400 billion to the economy came from a report commissioned by Google, and that ministers now work from a far lower official range of £55 billion to £140 billion by 2030.

The disclosure matters because the £400 billion figure has been used for almost two years to sell the economic case for AI, even though internal government estimates were materially smaller and the original number was rooted in a 2023 industry-commissioned study.

What the minister said

According to WIRED, AI minister Kanishka Narayan told Parliament that the government’s current estimate of AI’s economic impact is a central case of £120 billion by 2030, with a broader range stretching from £55 billion to £140 billion. He also said the larger £400 billion figure came from a 2023 report commissioned by Google and produced by Public First.

That is the first recorded case WIRED found of a government minister setting the two estimates side by side. But Narayan did not withdraw the £400 billion claim or say it was wrong, leaving the older talking point in an ambiguous position.

“hugely uncertain”

Narayan used that phrase in response to the economic forecasting behind AI adoption, according to WIRED.

Why the £400 billion figure mattered

The number was not a minor footnote. It became one of the headline claims used by ministers and government departments to argue that AI could deliver a large productivity and growth dividend for the UK.

WIRED reports that departments and agencies under two governments continued using the £400 billion figure for nearly two years. Peter Kyle also used it while promoting AI benefits, reinforcing the sense that it had become part of the government’s standard pitch on the technology.

The problem is not simply that the number was large. It is that the figure appears to have been treated as if it were an official public-sector estimate, when in fact it came from a study commissioned by a private company with a strong interest in AI adoption.

How did the government end up with two numbers?

The answer, based on the material reported by WIRED, is a combination of timing, institutional inertia and different assumptions about how quickly AI will be adopted.

The older £400 billion estimate came from a 2023 Public First report commissioned by Google. Public First later rebranded as PF Global. The newer government estimate, according to Narayan, puts the economic impact much lower: £55 billion to £140 billion by 2030, with a central estimate of £120 billion.

Ministers reportedly had the lower estimate by at least November 2025, yet Skills England still published the old £400 billion figure four months later. That lag is important because it suggests the government’s public-facing messaging was not updated in step with its own internal view.

Lord Holmes asked ministers about the source of the £400 billion claim in January 2025. Narayan’s answer last week was the first time WIRED found a minister explicitly linking the number back to the Google-commissioned report.

By the numbers

Why it matters

This is about more than a disputed forecast. Economic claims shape policy, spending priorities and public consent for major technology programmes. If the government overstates the likely gains from AI, it risks making its case on inflated expectations rather than evidence.

That concern is especially sharp because the UK government has been actively promoting AI as a route to productivity growth. A headline number as large as £400 billion can influence how seriously voters, civil servants and businesses treat the technology. Once such a figure enters circulation, it can take on a life of its own, even when newer internal modelling points to a smaller effect.

The episode also exposes a transparency problem. When a government repeatedly cites a number derived from an industry-commissioned report, it should be clear about where the figure came from, how it was produced and whether it reflects an independent policy judgment. WIRED’s reporting suggests that did not happen quickly enough.

For readers, the practical lesson is simple: AI policy debates often rest on assumptions about adoption, investment and labour-market change that are highly sensitive to the inputs chosen. A small shift in those assumptions can dramatically alter the headline economic upside.

Background: how the AI growth pitch took hold

UK governments have been eager to frame AI as an engine of national competitiveness. That is not unusual. Governments of different political stripes have used big technology forecasts to justify reform, investment and deregulation. What stands out here is how long the same AI boost figure kept reappearing.

According to WIRED, the £400 billion claim circulated under two governments and across multiple departments and agencies for nearly two years. That endurance matters because it shows how a single attractive statistic can become embedded in official communication, even after more cautious internal estimates are available.

The source of the older number also matters. Public First, later PF Global, produced the report in 2023 for Google. That does not automatically make the estimate invalid, but it does mean the figure originated in a private-sector commissioned analysis rather than a neutral public statistical exercise.

PF Global’s head of economics, Jonathan Dupont, told WIRED that AI adoption has moved more slowly than expected. He said that helps explain the gap between the older model and the government’s newer estimate, and that PF Global’s own central estimate is now closer to the government’s number, with gains ramping up later.

“AI is a very fast moving technology”

That observation helps explain why any long-range forecast for AI will be uncertain. Adoption rates, capital spending, regulation and workforce changes can all shift quickly. But uncertainty is not the same as a free pass to keep citing an outdated figure without context.

What the critics are saying

Opposition and civil society figures are warning that the episode could weaken trust in the government’s digital strategy.

Dame Chi Onwurah, the Labour MP who chairs the Commons Science, Innovation and Technology Committee, said overstating the benefits of technology risks undermining the case for change. Her point is straightforward: public support for innovation can erode if official claims appear overstretched.

Lord Holmes said ministers should have clarified the figure sooner and more directly. That criticism goes to process as much as substance. Even if a forecast is uncertain, the origin of the number should be easy to trace, especially when it is being used in public-facing policy arguments.

Sumedha Deshmukh of the Ada Lovelace Institute argued that government should independently test the numbers it is given before building policy on them. Her view captures the central policy lesson in this case: when a figure is politically useful, it should be scrutinised more, not less.

“The government should test the numbers it is given before it builds policy on them, fund independent research, and measure whether promised gains are delivered.”

Comparison: the old claim and the new estimate

What changed?

The headline change is that the government’s own estimate is far below the £400 billion figure it had been using publicly.

• Old talking point: £400 billion AI boost.
• Current official estimate: £55 billion to £140 billion by 2030.
• Central case: £120 billion by 2030.
• Origin of old figure: 2023 report commissioned by Google and produced by Public First.
• Reason for gap: Slower-than-expected AI adoption, according to PF Global.

The spread between the two numbers is wide enough to matter politically. It changes the tone of the government’s argument from transformational certainty to more modest, conditional optimism.

What happens next

The key question now is whether the government formally revises or drops the £400 billion talking point. Narayan’s answer appears to have opened the door to greater transparency, but it did not fully close the issue.

Watch for three things in the weeks ahead:

1. Whether other departments or agencies stop using the older figure.
2. Whether the Cabinet Office explains how the number continued circulating after the lower estimate was known.
3. Whether Google, Public First or PF Global responds publicly to the renewed scrutiny.

There is also a broader policy question. Will ministers now insist on independent testing of major AI economic claims before using them in speeches, strategy documents or agency materials? If they do not, this episode could repeat itself with the next headline number.

For now, the clearest lesson is that the politics of AI growth remain as much about narrative control as economic modelling. A persuasive number can travel fast. A corrected one usually moves more slowly.

Timeline of the claim

Here is how the story unfolded, based on WIRED’s reporting:

Bottom line

The UK government’s AI pitch has been leaning on a £400 billion boost figure that now appears to have come from a Google-commissioned report, while the government’s own current estimate is much smaller. The disclosure raises questions about transparency, the reliability of tech-economic claims and how quickly official messaging is updated when the numbers change.