Free AI diagnosis: 30 minutes, and you leave with a plan

Statistics

AI agent market statistics with sources

The numbers behind the AI agent market: adoption, analyst forecasts and failure predictions, each tied to its publisher and year, with a plain reading of what they mean.

Updated 6 min read

The numbers

Key AI agent numbers, with sources

What the AI agent market numbers say

  • Software

    A third of enterprise apps with agents by 2028.

  • Decisions

    15% of daily work decisions made autonomously.

  • Spend

    $234 billion of application spending exposed.

The AI agent market is growing fast from a very small base. Gartner expects a third of enterprise software to include agentic AI by 2028, against under 1% in 2024.

Spending exposure is large too. Gartner sees up to $234 billion of application spending open to agentic arbitrage through 2030.

That is an exposure estimate, not a market size. Gartner also expects 15% of day-to-day work decisions to be made autonomously by agentic AI by 2028.

Adoption is uneven by company size

Large firms

  • 40% scaling agents
  • Up from 27% last year

Smaller firms

  • 22% scaling agents
  • Essentially flat

McKinsey finds 40% of large organizations scaling agents, up from 27%. Among smaller organizations the share stayed flat at 22%.

So the gap is widening. Big companies move faster, while many smaller ones are still testing or waiting.

  • Large organizations: 40% scaling agents.
  • Smaller organizations: 22%, about the same as before.

The failure and cancellation forecasts

Causes named

  • Escalating costs
  • Unclear business value
  • Inadequate risk controls

Not claimed

  • Agents do not work
  • Every project fails

Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027. The reasons are rising costs, unclear business value and weak risk controls.

A second forecast says 70% of enterprises will abandon agentic AI built by vendor engineers by 2028, trapped by costs and unable to evolve it on their own.

These are forecasts, not measured outcomes. They point to the same risk: building without a clear process, a measured return and a way to run it afterward.

How to read statistics on this topic

  1. 01

    Who was asked

    Size and role of respondents.

  2. 02

    What was measured

    Forecast, survey or platform data.

  3. 03

    When

    The year of the study.

  4. 04

    Who paid

    Vendor data is not market data.

Check who was asked. A survey of executives at large firms says little about a 30-person company, and McKinsey itself shows the split.

Check the year and the verb. A forecast is not a result, and an exposure estimate is not a market size. Figures from different publishers measure different things, so we keep them apart.

Check who ran the survey. Salesforce reports its own platform data, which shows activity among its customers, not the whole market.

What it means for a mid-size company

  • Pick one process

    Start where manual work costs most.

  • Measure first

    Know the hours before you build.

  • Control risk

    Set limits and review points early.

The forecasts do not say to avoid agents. They say to pick one costly process, measure it and build where you can own and change the result.

If you want background first, read the guide to AI agents, or see what they cost on the AI agent cost page. For real projects, browse AI agents examples.

Our data

What AI answers when buyers ask

We asked ChatGPT, Gemini and Perplexity 10 buyer questions about AI agent development. They named 103 companies, and 79% appeared in only one of the three.

CriterionChatGPTGeminiPerplexity
Companies named443648
Sites cited mostmasterofcode.com (4 of 10)devcom.com and neurons-lab.com (3 of 10)litslink.com and brocoders.com (4 of 10)

Nexum measurement, October 20, 2026, via DataForSEO: 10 buyer questions to each AI, US English, 30 answers in total. 3 companies appeared in all three. Google showed an AI Overview for 8 of the 10 questions.

Methodology

How we chose these numbers

  1. 01Every figure comes from a published study, press release or report, with its publisher and year shown.
  2. 02Figures stay as the source states them: we do not round, merge or extrapolate.
  3. 03Market-size estimates from different publishers would be shown as separate items, never combined.
  4. 04Forecasts are labeled as forecasts, and survey results as survey results.
  5. 05The page carries its last update date and changes when a newer study replaces a figure.

FAQ

AI agent market questions

  • Publishers measure it differently, so no single number is reliable. Gartner gives an exposure figure of up to $234 billion of application spending through 2030. Check each publisher's definition before quoting a market size.

  • Definitions depend on the publisher. Some count software revenue, others count spending that shifts from existing applications. That is why this page lists each estimate with its source and does not merge them.

  • Adoption data says yes, mostly at large firms. McKinsey reports 40% of organizations with over $1 billion in revenue scaling agents, while smaller ones sit at 22%. See the latest AI statistics.

  • An AI agent is software that takes a goal, decides which steps to run and acts on it using your tools, rather than only answering questions. The AI agent definition goes deeper.

  • A chatbot mainly answers messages. An agent also acts: it can update records, send emails or start tasks. See AI agent vs chatbot for the full comparison.

  • Gartner forecasts over 40% will be canceled by the end of 2027. That is a prediction tied to cost, unclear value and weak risk controls, not a measured failure rate.

Related

Keep reading

Final step

Pick a time. Let's talk.

You leave knowing which processes to automate first and what each one costs you today, whether we work together or not.

You can keep paying for manual work, or find out this week exactly what it costs you.

What happens on the call

  1. Before

    You answer 3 quick questions when you book, so we come prepared.

  2. On the call

    We find the manual work that costs you most and put a number on it, live.

  3. Decide

    An honest read on whether we're the right fit. If we're not, you'll hear it on the call.

Once you book, we look at your site and tools before the call, so the 30 minutes go to your numbers, not introductions.

Prefer to write first? Email us directly: contact@nexumautomations.com

30 minutes · Video call · Free