Summary
- Gartner's 23 October 2024 release put 2025 worldwide IT spending at $5.74 trillion, up 9.3% from 2024; the 21 January 2025 release, quoting the same analyst, put the same year at $5.61 trillion, up 9.8% — a smaller year and a larger percentage, four months apart.
- The published tables carry the mechanism in plain column arithmetic: the 2025 total fell $129,522 million while the 2024 base that both growth rates divide by fell $145,013 million, so the percentage rose because the denominator dropped faster than the numerator.
The same name on two different years
John-David Lovelock is a Distinguished VP Analyst at Gartner, and both releases say so in the attribution line — the October one on the release announcing 9.3% growth, the January one on the release announcing 9.8%. That is the whole of his formal standing in this story. What makes him the subject rather than a letterhead is that he supplies the explanation, in his own quoted words, for the thing the arithmetic shows.
The October 2024 release states worldwide IT spending of $5.74 trillion for 2025, up 9.3% from 2024. Its Table 1, denominated in millions of U.S. dollars, puts 2025 Overall IT at $5,747,317 million and the 2024 column at $5,259,784 million. The January 2025 release states $5.61 trillion for the same year, up 9.8%, with a table total of $5,617,795 million over a 2024 base of $5,114,771 million. The two bodies of text and the two tables are internally consistent: the round figure and the granular one are two precisions of one forecast, not a contradiction.
Read across, the forecast year fell $129,522 million. That subtraction is mine, not Gartner's, and so is the next one: the base year fell $145,013 million. Divide each release's own two columns and the stated rates come back — roughly 9.27% in October, roughly 9.83% in January. The rate climbed because the thing being divided by was cut harder than the thing on top.
The perimeter did not move
A falling total and a rising rate would be a much more interesting finding if the definition had changed, and it did not. Both tables list the same five segments ahead of the total: Data Center Systems, Devices, Software, IT Services, and a communications services line that reads "Communication Services" in October and "Communications Services" in January. One letter, no stated change of scope.
Neither table carries AI or GenAI as its own top-level row, so a reader comparing the two documents cannot isolate an AI line at all; the January release mentions $202 billion of AI-optimized server spending in its narrative text, and that number sits outside the table rather than inside a row.
That matters for anyone planning against the figure. When the perimeter holds and the numbers move, what moved is an estimate inside a model no subscriber-free reader can open. The published table lets you verify that the revision happened and how large it was in each column. It cannot tell you which input changed.
Prices, not appetite
Lovelock's January wording is unusually blunt about what the higher total means. As the release carries it: "While budgets for CIOs are increasing, a significant portion will merely offset price increases within their recurrent spending." He continues that "nominal spending versus real IT spending will be skewed, with price hikes absorbing some or all of budget growth," and that all major categories reflect higher-than-expected prices, prompting CIOs to defer and scale back their true budget expectations. Those sentences are the release's quotations, not a transcript of an interview.
They also make the restatement less mysterious without excusing anyone from explaining it. A worldwide spending total is nominal: it counts dollars spent, not work bought. If prices rise faster than volume, the same budget buys less and the aggregate can still grow. A forecaster who tells you this in the same release that raises the growth rate is not hiding the price effect — he is naming the mechanism that lets a bigger percentage describe a weaker year for buyers.
The prediction that aged in his own house
Lovelock's October words were a forecast about expectations, not just spending: "CIOs will begin to spend on GenAI, beyond proof-of-concept work, starting in 2025. More money will be spent, but the expectations that CIOs have for the capabilities of GenAI will drop." Four months later his January words report that outcome as fact: "GenAI is sliding toward the trough of disillusionment which reflects CIOs declining expectations for GenAI, but not their spending on this technology."
Read as a record, that is a self-report. The same institution that published the prediction published the verdict, and neither release points to an independent survey that measured the change in CIO expectations. The prediction may well be right. What the two documents establish is that Gartner says so, twice, in the same voice.
The October release also carries his longer horizon: a projection that $500 billion will be added in spending every year in terms of growth rates, and that IT spending should cross the $7 trillion mark in 2028. That is a claim about a year nobody has reached, published alongside a 2025 figure that his own institution revised lower within a quarter. Both facts belong in the same paragraph, because they are the same kind of object.
What a reader can check, and what only Gartner can
With the two releases open side by side, anyone can confirm that both documents exist and are dated, that both quote Lovelock by name and title, that the five segment rows match, that neither table has an AI row, and that the column arithmetic produces the stated percentages. That is a real audit, and it needs no subscription.
No reader can confirm which input inside the model moved the 2024 base by $145,013 million, because neither release states a reason for it in the passages reviewed, and no Gartner methodology note or restatement notice was among the documents read for this piece. The gap is not a scandal. It is the ordinary boundary of a published forecast: the publisher owns the model, the reader owns the comparison, and the comparison is enough to show that the headline rate and the headline level moved in opposite directions.
What this adds to what we had
This archive has already written about Lovelock as the author of the language that boards, vendors and buyers use to decide whether an IT spending wave is real demand or early enthusiasm — a Reading of his influence on how AI spending gets talked about. That piece did not test one forecast year's published number against a later restatement of the same year, and it did not check a growth rate against the base it divides by. This one does, and the answer is a specific, dated divergence rather than a general observation that forecasts change.
The date is the artefact
A planning committee that took 9.8% as good news and a committee that took $5.61 trillion as bad news were reading the same table and both were partly right. The number that actually governs their decision is neither the rate nor the level. It is the date on which the rate was last restated — because the next restatement will move the base again, and the percentage will move with it, and the person whose name sits on the release will explain the result in the same confident voice.
Member Briefing
Read the full trend briefing
Members read the full briefing with the source notes behind every finding.
Strategic Circle
A daily briefing in your language, built from changes we have verified.
Join for USD 20/monthCancel anytime.
Members also get the full research library, a daily briefing on the networks they follow and the private forum.






