Updated: July 2026 | 9 min read

AI SaaS statistics are often inflated by mixing infrastructure, models, services, and application software. This article keeps those categories separate and combines Gartner’s 2026 spending forecast with observed enterprise-adoption data from McKinsey and Eurostat.

Worldwide AI Spending Forecast

Gartner forecasts worldwide AI spending of $2.595667 trillion in 2026, up 47% from $1.764947 trillion in 2025. The forecast covers multiple layers of the AI economy and is much broader than SaaS.

Gartner AI category 2025 2026 forecast
AI services $436.4B $585.5B
AI software $282.9B $453.2B
AI models $15.5B $32.6B
AI infrastructure $975.6B $1.432T
Total AI spending $1.765T $2.596T

AI infrastructure is the largest forecast category. Consequently, the $2.596 trillion total must not be described as AI SaaS revenue. The closest listed category is AI software, forecast at $453.209 billion, but even that includes software beyond subscription applications.

Enterprise AI Adoption

McKinsey’s 2025 State of AI survey gathered 1,993 responses across 105 countries. Nearly nine in ten respondents said their organizations regularly used AI. At the same time, nearly two-thirds said their organizations had not begun scaling AI across the enterprise.

This combination shows why access and scale are different measures. A company may use an AI feature in one SaaS product while still lacking enterprise governance, integrated data, or a repeatable deployment process.

AI Agent Experimentation

McKinsey found that 62% of respondents said their organizations were at least experimenting with AI agents. However, only 39% reported enterprise-level EBIT impact from AI. The survey also found that 64% said AI enabled innovation.

These are respondent-reported outcomes, not audited returns. For SaaS buyers, they support a measured approach: document the task, baseline current performance, run a controlled pilot, and track human review and failure rates before scaling.

Observed EU Enterprise Use

Eurostat reported that 20.0% of EU enterprises with at least 10 employees used AI technologies in 2025, up 6.5 percentage points from 13.5% in 2024. Large-enterprise adoption was 55.03%.

The most common measured use was analysis of written language at 11.8% of enterprises. Generation of pictures, video, or audio was reported by 9.5%, generation of written or spoken language by 8.8%, and conversion of speech into machine-readable form by 7.2%.

Eurostat’s result is lower than McKinsey’s because the studies measure different populations and definitions. Eurostat covers enterprises in specified economic sectors and size bands; McKinsey surveyed business respondents globally. The numbers should not be treated as contradictory or averaged together.

What Counts as AI SaaS?

  • AI features embedded in an existing subscription application.
  • Standalone AI applications delivered by subscription or usage pricing.
  • Platforms used to build and manage AI applications.
  • Model APIs consumed by a SaaS product.

These layers can overlap. A SaaS vendor may pay for cloud infrastructure and model APIs, then sell an AI-enabled application. Adding revenue estimates for every layer would double-count the same economic activity.

Evaluating AI SaaS Adoption

Feature availability is not the same as active use. A vendor may enable an assistant for every account, while only a subset of users invoke it repeatedly. A useful adoption measure should specify the eligible population, active-user window, task completed, and whether use was optional or automatic.

Value measurement also needs a baseline. Teams can compare handling time, error rates, completion rates, or revenue outcomes before and after deployment, ideally with a control group. They should include review time, model and infrastructure costs, integration work, and failures when calculating the result.

Governance is part of the product evaluation. Buyers should document data retention, model-training policies, access controls, audit logs, human escalation, and geographic processing. These controls are particularly important when AI features operate across customer records, financial information, source code, or regulated data.

Forecast Versus Observed Data

The Gartner values describe expected 2026 spending, while McKinsey and Eurostat report survey observations collected in 2025. Forecasts help with planning but can be revised. Observed adoption data is more concrete, yet it remains sensitive to survey population and definition. Keeping the date and measurement type beside each number prevents a projection from being mistaken for a completed result.

Vendor telemetry can add another perspective, but it often measures enabled accounts, prompts, tokens, or feature events rather than organizations receiving value. Those measures should be labeled precisely and should not replace independent adoption or financial evidence.

Renewal, expansion, and sustained task completion are stronger adoption signals than a one-time feature event.

Methodology and Limitations

Spending values are Gartner forecasts published in May 2026. Adoption figures are observed survey or statistical results from 2025. This page does not publish a single AI SaaS market size because none of the three sources defines and audits that exact category consistently.

Key Takeaways

  • Gartner forecasts $2.596 trillion in total AI spending for 2026.
  • AI software accounts for $453.2 billion of that forecast.
  • Infrastructure, not SaaS, is the largest forecast component.
  • McKinsey found broad AI use but limited enterprise-wide scaling.
  • Eurostat measured 20.0% adoption among covered EU enterprises in 2025.
  • AI spending forecasts and observed adoption rates answer different questions.

Sources

  1. Gartner: Worldwide AI spending forecast.
  2. McKinsey: The State of AI in 2025.
  3. Eurostat: Enterprise AI use in 2025.