Enterprise AI Is Now
a $177B MarketEnterprise AI
Is Now a $177B Market
Based on Menlo Ventures “2026: The State of Enterprise AI”
Enterprise AI has gone from early pilots in 2023 to mass adoption in 2026, and it is now time to take stock. The market has grown to $176.8B a year. Clear leaders have emerged, the most profitable categories have taken shape and AI agents have changed how models are consumed.
Enterprise AI Spending Has Grown 100× in Three Years
Global Enterprise AI Spending
$B, growth vs. the prior year under each year
| Year | $B | YoY growth |
|---|---|---|
| 2023 | 1.7 | — |
| 2024 | 8.5 | 5.0× |
| 2025 | 33 | 3.9× |
| 2026* | 176.8 | 5.4× |
The share of companies paying for AI rose just 10 percentage points, from 55% to 65%. The growth is coming from companies that are already scaling AI
Agents vs. Humans on OpenRouter
Weekly tokens, trillions
| Week | Agents | Humans | Mixed |
|---|---|---|---|
| Sep ’25 | 0.4 | 2.5 | 1.1 |
| Sep 12 ’25 | 0.4 | 2.5 | 1.1 |
| Sep 19 ’25 | 0.4 | 2.5 | 1.2 |
| Sep 26 ’25 | 0.4 | 2.6 | 1.2 |
| Oct 3 ’25 | 0.5 | 2.6 | 1.3 |
| Oct 10 ’25 | 0.5 | 2.6 | 1.4 |
| Oct 17 ’25 | 0.5 | 2.7 | 1.4 |
| Oct 24 ’25 | 0.5 | 2.7 | 1.5 |
| Oct 31 ’25 | 0.5 | 2.7 | 1.6 |
| Nov 7 ’25 | 0.5 | 2.7 | 1.6 |
| Nov 14 ’25 | 0.6 | 2.8 | 1.7 |
| Nov 21 ’25 | 0.6 | 2.8 | 1.8 |
| Nov 28 ’25 | 0.7 | 2.7 | 1.8 |
| Dec 5 ’25 | 0.8 | 2.7 | 1.9 |
| Dec 12 ’25 | 0.9 | 2.6 | 2 |
| Dec 19 ’25 | 1 | 2.5 | 2 |
| Dec 26 ’25 | 1.1 | 2.5 | 2.1 |
| Jan ’26 | 1.3 | 2.7 | 2.2 |
| Jan 9 ’26 | 1.4 | 2.9 | 2.2 |
| Jan 16 ’26 | 1.5 | 3.1 | 2.3 |
| Jan 23 ’26 | 2 | 3.4 | 2.3 |
| Jan 30 ’26 | 2.4 | 3.6 | 2.4 |
| Feb 6 ’26 | 4.1 | 3.8 | 2.5 |
| Feb 13 ’26 | 5.3 | 3.9 | 2.6 |
| Feb 20 ’26 | 6.1 | 4.1 | 2.8 |
| Feb 27 ’26 | 6.4 | 4.2 | 3 |
| Mar 6 ’26 | 7.2 | 4.3 | 3.1 |
| Mar 13 ’26 | 8.5 | 4.4 | 3.3 |
| Mar 20 ’26 | 11.1 | 4.5 | 3.5 |
| Mar 27 ’26 | 14.3 | 4.6 | 3.6 |
| Apr 3 ’26 | 13.5 | 4.8 | 3.8 |
| Apr 10 ’26 | 10.8 | 4.9 | 4 |
| Apr 17 ’26 | 11 | 5 | 4.1 |
| Apr 24 ’26 | 12.4 | 5.3 | 4.3 |
| May ’26 | 13.4 | 5.6 | 4.5 |
| May 8 ’26 | 14.2 | 5.9 | 4.6 |
| May 15 ’26 | 15 | 6.2 | 4.8 |
| May 22 ’26 | 16.7 | 6.5 | 4.9 |
| May 29 ’26 | 19.8 | 6.8 | 5.3 |
| Jun 5 ’26 | 23.8 | 7.2 | 5.8 |
| Jun 12 ’26 | 27.3 | 7.5 | 6.3 |
| Jun 19 ’26 | 29 | 7.8 | 6.8 |
| Jun 26 ’26 | 29 | 8.2 | 7.3 |
| Jul 3 ’26 | 33.1 | 8.6 | 7.8 |
| Jul 10 ’26 | 40 | 9 | 8.3 |
| Jul 17 ’26 | 41.2 | 9.4 | 8.8 |
| Jul 24 ’26 | 38.8 | 9.8 | 9.3 |
| Jul 31 ’26 | 43.2 | 10.1 | 9.8 |
| Aug 7 ’26 | 50.6 | 10.4 | 10.1 |
| Aug 14 ’26 | 62.6 | 10.6 | 10.3 |
| Aug 21 ’26 | 79.8 | 10.9 | 10.5 |
| Aug 28 ’26 | 89.2 | 11.2 | 10.6 |
| Sep ’26 | 95.8 | 11.7 | 10.8 |
| Sep 11 ’26 | 102 | 12.3 | 11 |
| Sep 18 ’26 | 110.2 | 14 | 12.2 |
| Sep 25 ’26 | 130.9 | 18.3 | 14.9 |
Since February, agent workloads have grown 37× and human workloads 5×
Key Companies
Two-Thirds of All Spending Goes to Three Labs
The market has two layers: infrastructure and applications. Anthropic, OpenAI and Google lead in both. Together they capture 69% of all spending, up from 50% a year earlier.

$99.5B
InfrastructureThe models and compute that power every AI service

$77.3B
ApplicationsTools for employees, from AI coding agents to chat assistants
Anthropic and OpenAI Hold 86% of the Model Market
Share of enterprise LLM API spending
| Year | Anthropic | OpenAI | Meta Llama | Other | |
|---|---|---|---|---|---|
| 2023 | 12 | 50 | 7 | 16 | 15 |
| 2024 | 24 | 34 | 12 | 16 | 14 |
| 2025 | 40 | 27 | 21 | 8 | 4 |
| 2026 | 55 | 31 | 9 | — | 5 |
Businesses are willing to pay more for the best model
Where the AI Budget Goes
Spending by category in 2026, $B; growth over the year on the right
| Name | Value | Note |
|---|---|---|
| Models and compute | 94.2 | Infrastructure · 2025: 12.5 |
| Coding | 34.0 | Applications · 2025: 4.4 |
| Assistants | 25.0 | Applications · 2025: 8.4 |
| Other apps | 18.3 | Applications · 2025: 6.1 |
| AI tooling | 5.3 | Infrastructure · 2025: 1.5 |
Infrastructure: $99.5B, applications: $77.3B. Compute and coding are growing fastest

$25.0B
3.0× YoYClaude
Gemini
Microsoft
Chat assistants are now as standard at work as email
New budgets account for 55% of new AI spending. AI has become a line item of its own and is expanding the software market
Source: Menlo Ventures, “2026: The State of Enterprise AI.”
AI’s Biggest User Is Now AI Itself
Growth is being driven by AI agents — software that completes tasks on its own, start to finish. An agent calls a model thousands of times a day, which is why companies spent 2.5 times more on AI than planned.

$34.0B
7.7× YoYClaude Code
Codex
Cursor
The most mature category. Claude Code and Codex account for 70% of spending

$3.7B
Healthcare · 2.5× YoYAI takes notes during patient visits and frees up doctors’ time

$1.8B
Legal · 2.8× YoYHarvey and Legora handle deal due diligence end to end
Source: Menlo Ventures, “2026: The State of Enterprise AI.”
Token Prices Are Falling, While Volumes Keep Rising
Open models can run on a company’s own servers, which makes them several times cheaper. As AI gets cheaper, demand for it only grows.
Kimi
Moonshot AI, ChinaThe open-source leader by enterprise spending
Open Models Win on Tokens, but Not on Spending
Share of open models, %
| Month | Token share, % | Spending share, % |
|---|---|---|
| Sep ’25 | 11 | 6 |
| Sep ’26 | 19 | 4 |
Share of Open-Source Spending
Source: Menlo Ventures, “2026: The State of Enterprise AI.”
Five Key Points for Investors
The Market Is Accelerating
Enterprise spending has reached $177B a year, 5.4 times the level of a year ago. Growth has accelerated for three years running, and the market is still far from saturated.
The Leaders Are Pulling Ahead
Anthropic, OpenAI and Google capture 69% of all AI spending. Businesses are willing to pay for the best model, so the gap between the leaders and everyone else keeps widening.
Agents Drive Demand
Software agents already generate 80% of model workloads. An agent works around the clock and calls models thousands of times a day, which is why companies are spending 2.5 times more than planned.
Cheaper Means More
Open models such as Kimi cost a fifth as much as closed ones. As token prices fall, companies hand more and more tasks to AI, and usage keeps climbing.
The Next Verticals
Coding grew to $34B in a year and showed how AI can create a new market. Healthcare, legal and other industries where agents take over entire workflows are next.
This material is based on the Menlo Ventures report “2026: The State of Enterprise AI” and OpenRouter data.
Axevil Capital, LLC is a company registered in the U.S. (State of Delaware, #6832739) with Exempt Reporting Adviser (ERA) status and is regulated by the U.S. Securities and Exchange Commission (SEC #802-126907). We work only with qualified investors. This material is for informational purposes only and does not constitute an offer, an investment recommendation or an invitation to enter into a transaction. Investments in private markets are illiquid, long-term and involve high risks, including the total loss of capital; past performance does not guarantee future results.
© Axevil Capital, 2026





