Enterprise AI spending did not just grow in 2026 — it exploded. Menlo Ventures' fourth annual State of Enterprise AI report, published on November 3, puts enterprise AI deployment spending at $176.8 billion by September 2026, up 5.4 times from $33 billion a year earlier. The growth is accelerating, not slowing: last year's increase was 3.9 times, and the three-year climb runs from $1.7 billion in 2023 to $8.5 billion in 2024 to $33 billion in 2025 to $176.8 billion now. The driver, according to Menlo, is agents: autonomous AI systems have become the biggest consumers of AI models, and spending is now tied to work performed rather than to people using a tool. The report is based on a survey of nearly 400 US enterprise decision-makers responsible for more than $4 billion in AI spending, fielded by Morning Consult from September 8 to 26, 2026.
The report's most striking finding is who — or what — is using the models. On OpenRouter, a service that routes requests to many model providers through one API, agents now generate 80% of all tokens, about 130.9 trillion per week, roughly seven times human-driven usage. Agent traffic overtook human traffic there in early February 2026, following the launches of new coding models, and has since grown 37 times, compared with 5 times for human-driven usage. The economics follow the usage: for every $1 the industry spends building AI models, enterprises now spend about $1.70 putting them to work, with model-building spend at $102.5 billion, up 5.3 times.
The spending is concentrated where agents work best. Coding accounts for $34.0 billion and chatbots for $25.0 billion, together making up 76% of the $77.3 billion application market, because those are the environments where agents can act, follow workflows and verify results. Coding is 96% AI-native, meaning the category is dominated by companies built around AI rather than incumbents adding it on. Deployment spending splits between $99.5 billion of infrastructure and $77.3 billion of applications, and the frontier labs — Anthropic, OpenAI and Google — capture 69% of deployment spending between them.
Perhaps the most important number in the report is the one that did not explode. The share of companies using AI rose only from 55% to 65% over the period, while average spend per company rose 4.7 times. The $176.8 billion surge came not from new adopters but from existing ones spending far more heavily — buying more tokens, running more agents, and putting AI into more workflows. That is the pattern of a technology crossing from experiment to infrastructure: the question is no longer whether to use AI, but how much work to hand it.
Menlo is upfront that the dollar figures are its own estimates, built from a bottom-up model informed by survey responses, public filings and third-party reports, not audited revenue. The firm also discloses that it is an investor in Anthropic and OpenRouter, two of the companies whose numbers feature prominently in the report. Those caveats do not invalidate the findings — the direction and the order of magnitude are consistent across the survey, the OpenRouter data and the public filings — but they are a reminder to treat the $176.8 billion as a modelled estimate, not a counted total.
For the search industry, the Menlo report is a map of where the money is going — and it is going to agents that do work. The same shift is visible in search: AI agents that research, compare and buy on behalf of users are the new audience, and they consume content the way Menlo's agents consume tokens, at machine scale and without patience for thin pages. The brands winning agent attention are the ones with structured, verifiable, citable content, because that is what an agent can act on.
There is also a direct business lesson. The report shows that spending concentrates where agents can complete verifiable tasks — coding and customer-facing chat, above all. For agencies and SaaS companies, the equivalent is to build offerings an agent can use: clear product data, documented capabilities, pricing a machine can read. The enterprises spending $176.8 billion are not buying demos. They are buying work done, and the content that gets cited, linked and recommended by their agents is the content that describes real work in terms a machine can verify.
Sources: Menlo Ventures, November 3, 2026.
Related reading on this site: Google Organic Click Share Hits a 15-Month High as Zero-Click Searches Fall, Reddit Begins Testing Ads in Its Own Search Results, 83% of ChatGPT Shopping Picks Also Rank in Google's Top 40, Semrush Finds, Google Rewrites Its 'Crawled – Currently Not Indexed' Guidance, Search Console's Device Filter Is Now Multi-Select, and Email and Paid Search Are Growing as Organic Search Shrinks.
Linkbaits helps you create content that earns citations and backlinks. Start with the toolkit and build your first linkable asset this week.
Start free trial