AI Market Figures for 2029: How to Read Them Without Being Misled

In August 2024 IDC announced that global spending on AI would reach US$632 billion by 2028, at a compound annual growth rate of 29.0% between 2024 and 2028, about US$202 billion of it for generative AI. About a year later, in September 2025, the same company published an estimate for 2029: about US$1.26 trillion in 'total AI IT investment', at 31.9%, with generative AI accounting for about US$607 billion. The 2024 base in the newer edition is about US$316 billion. The two pages do not explain the difference between the editions, but it is enough to show that a figure follows its definition.

What goes into the figure?

Gartner estimates 2025 spending on generative AI at about US$644 billion, up 76.4% on 2024. Its table shows that about 62% of it is devices, about 28% servers, 5.8% software and 4.3% services. So hardware is about 90% by the table (the text of the announcement says 80%). The company's analyst notes that expectations of the technology are falling because of the high share of proofs of concept that do not succeed, and that much of the growth comes from features built into devices that consumers have not asked for. Spending measures what is paid, not the value that is realised.

Grand View Research expects the market to reach about US$1.8 trillion by 2030, at a growth rate of 37.3% between 2023 and 2030 from a base of about US$136.55 billion in 2022. The published article about its report does not define the market explicitly, and that is every reader's first question.

The compound growth rate and its limits

Most of these estimates are announced as a compound annual rate. It is a useful arithmetic tool, but it assumes the pace of growth lasts for years. So it is no surprise that the final figure changes greatly when the assumptions or the starting year change.

Direction and magnitude

The direction towards growing AI spending is visible in every estimate cited here, although the magnitudes differ by multiples. The magnitude is the fragile part: it follows the definition, the starting point and the assumption of continuity. Whoever bases a decision on the direction stands closer to the evidence than whoever bases it on one particular figure. As for whether the 2029 figure will come true as stated, that knowledge is with God, and all we have are estimates revised as new information arrives.

What has actually been observed?

The observed differs from the forecast. The Stanford AI Index 2025 records that US private investment in AI reached US$109 billion in 2024, against US$9.3 billion in China; that the share of organisations reporting AI use rose from 55% in 2023 to 78% in 2024; and that the price of running a model at GPT-3.5 level fell from US$20 to 7 cents per million tokens between November 2022 and October 2024, a drop of more than 280-fold. These are observed figures, but they do not answer the question of return: large investment does not prove that profits have materialised.

Four questions before you believe a figure

• What does the figure count: hardware, software, services, or every product touched by an algorithm?
• Is it spending, revenue or value added?
• What growth rate was it built on, and from which year?
• When was it issued? The same firm changed its estimate within a year.

This is an explanation to help understand figures, not a recommendation to buy or sell. Anyone who wishes to invest should consult a licensed specialist, and observe the Sharia guidelines in what they choose.