AI Software Market Worth $143.6 Billion, Bottom-Up Model Finds

End customers spent on AI software at an annualized run rate of $143.6 billion in the first half of 2026, according to AI Insider's AI Market Model, within a range of $121.0 billion to $219.0 billion. That puts AI at roughly 10% of the $1.44 trillion the world spends on software this year, based on Gartner figures used in the model. The estimate rests on 179 company positions, with revenue vendors earn from each other removed so each customer dollar is counted once.

The number stands out because public estimates of the AI software market vary wildly. A Research and Markets report sizes AI software platforms at about $107 billion in 2026, counting services alongside tools. PwC puts the potential addressable market for AI software at $6 trillion. Much of the spread comes down to what each estimate counts: chips, consulting, or the same dollar counted twice as it passes between AI companies.

AI Insider's model, H1 2026 edition, takes a deliberately narrow view: what do end customers pay for AI models, applications and tooling? The figure is a run rate. Each company's revenue is taken at its latest reporting date in the first half of 2026 and annualized. Figures were updated on Nov. 3, 2026 to reflect Anthropic's S-1 filing.

What does the estimate count?

The model counts AI software bought by consumers, businesses and the public sector:

• Access to foundation models
• AI products sold by cloud and software platforms
• AI applications across 16 segments, from software engineering to financial services
• The surrounding developer tools

Consulting, chips, servers, data centers, power and rented cloud computing are excluded, because none is software a customer buys to use AI. Advertising revenue is excluded too, since advertisers rather than users pay it.

How is double counting handled?

When a coding application pays a model developer for the AI inside its product, both companies book revenue, but the customer paid once. The model strips out these flows. Its 179 company positions add up to about $153.8 billion; removing what vendors earn from each other brings the total to $143.6 billion.

How much of software spending is AI?

Global software spending is about $1.44 trillion in 2026, based on the Gartner figures used in the model, which puts AI at roughly 10% of the total. Other benchmarks land nearby: PwC puts global software expenditure at about $1.3 trillion, and Golden Section, citing Gartner, at $1.47 trillion for 2026. The ratio matters because the model projects AI's share of software spending rather than extrapolating AI revenue growth directly.

How certain is the estimate?

Less certain than a single number suggests. Every company position carries a central, low and high value and an evidence grade. Some come from filings, others are triangulated from several reports, and many smaller ones are modeled from funding, headcount and comparable companies. Summed, they give a range of $121.0 billion to $219.0 billion.

The range is wider on the upside because several large private companies disclose little. The estimate also moves with new information: Anthropic's S-1 filing in October and November 2026 changed how the model allocates some revenue between vendor types, though not the total. The model is refreshed quarterly.

Who sells and who buys?

On current estimates, model developers keep the largest share, about 71%, and businesses pay about four of every five dollars. Those splits rest on more assumptions than the total, such as how each developer's revenue divides between products and sales channels.

Buyer behavior adds context. Simon-Kucher's Global Software Study 2026 found that 98% of buyers say AI capabilities have met or exceeded expectations, yet only 4% measure AI's business impact across the enterprise and just 17% describe AI as broadly embedded. Spending is running ahead of proof.

Where could it go?

The model's base case has AI's share of software spending rising to about 25% by 2030, which would put end-customer spending at about $564 billion, and to about 37% by 2035, about $1.23 trillion. Lower and higher cases bracket that path; a follow-up article will explain all three. The full report and Excel dataset are available from AI Insider Intelligence.

By Daniel Okafor