How AI adoption could reshape UK sovereign risk and gilts

Moody's Ratings said on 27 February 2026 that generative artificial intelligence could lift productivity by about 1.5% a year on average across a broad set of the sovereigns it rates. That figure is a scenario estimate spanning many countries with different workforces, technology readiness and demographics. It is not a forecast for the UK, and it is not a prediction of what will happen to Britain's credit rating or its cost of borrowing.

The UK is nonetheless a useful case to test the idea against. Productivity growth has been weak for two decades, public debt is high, and debt-interest spending has more than doubled as a share of the economy since 2019-20. Small differences in how AI plays out for jobs and tax receipts could therefore matter for the UK's public finances.

That distinction matters because sovereign financing conditions reach further than the Treasury's own accounts. They affect the tax and spending choices open to the UK government, the rates households and businesses pay to borrow, the balance sheets and market-making role of UK banks, and the pricing of other sterling assets that use gilt yields as a benchmark.

Moody's thesis: a balance, not a bonus

Moody's frames the sovereign-credit effect of AI as a balance. On one side sit productivity gains: if workers and firms produce more per hour worked, that can widen the tax base and support growth. On the other side sit the social and fiscal costs of labour disruption, including retraining, unemployment support and the transition costs of reorganising work. Which side wins, Moody's says, depends on a country's workforce structure, technological readiness, demographics, unemployment and labour costs.

That framing matters because it rules out a simple story in which AI adoption is automatically good, or automatically bad, for a sovereign's credit profile. The outcome is conditional on policy choices and labour-market structure, not a fixed property of the technology itself.

From productivity to the public purse

Higher output per worker does not become fiscal strength by itself. It has to pass through wages, company profits, employment and the tax system before it shows up in the government's accounts. If gains flow mainly to workers who keep their jobs, income tax and National Insurance receipts can rise. If they flow to company profits instead, corporation tax rises but personal taxes and VAT may not. If AI displaces workers faster than the economy can redeploy them, the government can face higher welfare spending and lower personal-tax receipts even while measured productivity improves.

The Office for Budget Responsibility's March 2026 Economic and fiscal outlook illustrates exactly this fork with two scenarios built around its central forecast.

The UK's fiscal fork

The OBR describes the 1.5% case as potentially consistent with a more optimistic AI outcome, though it is not an AI-only model; other technological or structural changes could produce the same productivity path. The displacement scenario is the more striking result for this analysis: aggregate GDP growth does not fall, yet the government still borrows more, because lower income tax, National Insurance and VAT receipts outweigh the gain in corporation tax. In other words, a technology that raises output per worker can still weaken the public finances if it reduces the number of people in work.

For context, the Bank of England's Decision Maker Panel found in its February 2026 Monetary Policy Report that firms themselves expected AI to raise their own productivity by about 0.6% a year on average over the following three years. That is a firm-level expectation, not an economy-wide forecast, and it should not be read as a competing UK number against Moody's 1.5% cross-sovereign estimate; the two measure different things over different scopes. The Bank also noted that measured UK productivity growth had averaged about 0.5% a year over the previous two decades, and that potential productivity growth had averaged about 0.5% a year since 2024, below its assumed long-run trend of 1%. Weak recent history is part of why the AI question carries so much weight for the UK specifically.

Why gilt yields would not move mechanically

It is tempting to read a favourable productivity scenario straight through into lower government borrowing costs. That link exists in principle: stronger growth and a wider tax base can support investor confidence in a sovereign's ability to service its debt. But nothing in the sources reviewed here quantifies how many basis points a given UK productivity outcome would add to or subtract from gilt yields, and a rating or thematic credit view is only one input into market pricing.

The OBR and the Bank of England both point to a longer list of drivers: inflation, expectations for Bank Rate, global government-bond yields, the volume and maturity of new issuance, and the depth and liquidity of the gilt market itself. The OBR noted that UK 10-year government bond yields were the highest in the G7 and fourth-highest among advanced economies when its March 2026 forecast was prepared. That is a statement about relative market pricing at a point in time, not a claim that AI, or any single factor, explains the gap.

A country already exposed to rates

The UK's existing debt position amplifies whatever happens next. Public debt-interest spending rose from £39bn, or 1.7% of GDP, in 2019-20 to £106bn, or 3.6% of GDP, in 2024-25, according to the OBR. In its March 2026 central forecast, public sector net debt rises from 94.3% of GDP in 2025-26 to 96.3% in 2028-29 before easing, with net interest costs projected at 3.2% of GDP in 2029-30.

The way that debt is structured also matters for how quickly market-rate changes reach the Exchequer. The Debt Management Office's data for March 2026 put the average maturity of the UK debt portfolio at 13.38 years on a net uplifted nominal-value basis, with index-linked debt at 24.4% of the combined gilt and Treasury-bill portfolio. The OBR observed that shorter-maturity issuance lowers near-term interest costs but makes future payments more sensitive to refinancing rates. A shift towards shorter maturities, which the Bank of England's July 2026 Financial Stability Report also noted as a feature of the gilt market's changing structure, is a trade-off rather than a straightforward saving.

Why this reaches beyond the Treasury

Gilts are not only a financing tool for government. The Bank of England has said gilt and gilt-repo markets also benchmark other sterling financial instruments and support the transmission of monetary policy, which means changes in sovereign financing conditions matter for the wider economy, not just the government's own accounts. The Bank's July 2026 report noted that historically high issuance, a shift towards shorter maturities and greater participation by price-sensitive investors had changed the gilt market's structure, and warned that leveraged positions could amplify disorderly selling, though it said markets had remained functional through the shocks it examined.

What it means for banks

The exposure for UK banks runs in both directions. If AI adoption improves borrower income and credit quality across the economy, banks could see stronger loan performance and credit demand. Banks are also central to how gilt-market conditions reach the rest of the financial system, since they help intermediate trading and secured funding in gilts and gilt repo.

Set against that, the Bank of England's April 2025 assessment identified risks specific to AI use inside financial firms, including model risk, cyber risk, concentration in a small number of technology suppliers, and the possibility of correlated behaviour across institutions using similar systems. The Bank's 2025 capital stress test found the UK banking system could continue supporting the economy under materially worse economic and financial conditions, but that exercise was not designed as an AI-specific or UK-sovereign downgrade test, so it does not settle how banks would cope with an AI-driven shock combined with sovereign-market stress.

How much AI adoption is actually happening

Official statistics suggest change is real but shallower than headline adoption rates imply. ONS analysis published on 20 July 2026 found self-reported AI use among UK businesses with at least 10 employees rose from about 12% in late 2023 to about 35% in 2026, with the average adopting business using about 1.6 AI technologies, up from about 1.4. Adoption varied sharply by sector: 58% of information and communication businesses reported AI use in 2026, against 13% of construction businesses. A separate ONS study using 2023 survey data found adoption at 9% across its sample, with large service-sector firms most likely to have adopted it. The two figures come from different surveys, populations and questions, so they should be read as evidence of a rising trend rather than a precise, comparable jump.

What would make the upside credible

Moody's own framing, and the OBR's contrasting scenarios, point to the conditions under which AI could plausibly strengthen the UK's fiscal position rather than strain it: gains that diffuse beyond a narrow set of large service firms, workers who move into new roles rather than out of employment, and a tax system that captures value wherever it lands, whether in wages, domestic profits or consumption. None of this is guaranteed, and none of it is quantified as a single UK fiscal number in the sources reviewed here.

What to watch next

Readers who want to track whether the scenario is turning into reality can follow the primary sources this analysis draws on: Moody's future sovereign research, the OBR's subsequent economic and fiscal outlooks and any updated productivity scenarios, the Bank of England's financial-stability and monetary-policy reports, the ONS's ongoing AI-adoption and productivity statistics, and the Debt Management Office's quarterly reporting on gilt issuance and portfolio structure. Any of these would help readers assess whether output per hour, employment, tax receipts or borrowing were actually shifting in either direction.