Britain’s computer-programming industry made a measurable contribution to growth in July. The amount attributable to artificial intelligence did not.
The Office for National Statistics estimated that monthly gross domestic product rose by 0.4% in July 2026. Within services, computer programming, consultancy and related activities grew by 3.5%, adding 0.14 percentage points to services output and 0.12 percentage points to real GDP.
The ONS also found that many of the businesses reporting the largest turnover in computer programming and information services were involved in AI and cloud computing. Its next sentence supplies the essential limit: the nature of the data collection makes the exact impact of those activities difficult to quantify.
The 0.12-point figure should therefore be read as a contribution from a broad computer-programming industry classification. It is not an estimate of AI output, AI adoption or AI-driven productivity.
What the contribution measures
Monthly GDP is an early estimate of changes in real economic output. The 0.12-point contribution describes how much the movement in computer programming, consultancy and related activities added to the change in economy-wide output in July, after weighting the industry within GDP.
That is useful evidence. It shows the industry’s rise was large enough to matter to the monthly total, rather than merely reflecting enthusiasm around a small group of technology companies. The ONS also reports that information and communication output as a whole rose by 2.4% on the month.
But the precision has limits. The headline GDP and industry growth rates are rounded to one decimal place, while contributions are rounded to two decimal places. The ONS warns that components may not sum to the total because of rounding. Treating 0.12 as a precise share of the rounded 0.4% headline would imply more certainty than the release provides.
The monthly and three-month measures must also remain separate. Over the three months to July, GDP grew by 0.4% and services by 0.6%, while production and construction each fell by 0.5%. Computer programming grew by 4.4% over that period. Those figures describe a services-led economy with weakness elsewhere. They do not show that AI lifted productivity across the whole economy.
Why AI remains inside the industry total
The current collection is designed to measure industries, not to produce a product-level account for AI. A business associated with AI or cloud computing can also earn turnover from conventional software development, consultancy, infrastructure migration, support and other digital services. The bulletin identifies the activities of large reporters, but it does not divide their July turnover between those lines.
There is a second measurement issue. The Monthly Business Survey covered 43.3% of the services sector by industry weight, with an 84.6% turnover response rate for July at this stage of reporting. More responses are expected and can feed into later estimates.
For most industries, the ONS says monthly estimates use deflated turnover or volume as a proxy for gross value added. Gross value added is output minus the goods and services consumed in producing it, but intermediate consumption is collected annually rather than monthly. A rise in turnover-based output can therefore establish neither how much was produced by AI nor whether AI allowed firms to produce it with fewer inputs.
That distinction is the gap between an output claim and a productivity claim. Productivity requires relating output to an input such as hours worked or capital used. The July GDP bulletin does not perform that calculation for AI.
Two releases will test the number in different ways
The first test is methodological. On 21 September, the ONS is due to publish its approach to an AI thematic account. The confirmed release is intended to explain the account, the approach to compiling it and how it could help measure AI’s impact on the UK economy.
A thematic account could place an economic boundary around AI activity that crosses established industry classifications. The questions to watch are whether it distinguishes AI production from adoption, separates AI from cloud and other software activity, and connects output to the inputs needed to produce it. The announcement does not promise a revised attribution for July or say that the 0.12-point contribution will be decomposed.
The second test is revision. The July time-series release identifies the data as an early monthly estimate, while the bulletin says the entire series will reopen in the 15 October monthly release. That update will include Blue Book 2026 revisions. An indicative monthly path is due with the quarterly national accounts on 30 September.
Blue Book will incorporate newer survey and administrative information and methodological improvements. It will also estimate 2024 for the first time using annual supply and use tables, where output and intermediate consumption are measured together. The July industry contribution may survive unchanged, rise or fall. Until the revised series arrives, 0.12 remains a first estimate.
A signal, not a verdict on AI
Technology operators and investors can use the July figure as evidence of strong activity in a broad UK programming market. They cannot use it to infer the sales, output or productivity of an individual cloud or AI company. Nor can policymakers claim that AI generated 0.12 percentage points of UK growth.
The stronger conclusion is narrower. Computer programming was a material contributor to July’s expansion, and some of the largest turnover reporters work in AI and cloud. The official statistics cannot yet say which activity delivered how much. The 21 September methods release will show whether the ONS has a credible route to closing that gap; the autumn GDP revisions will show how stable the starting number is.
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.