Enterprise AI deployment spending reached $176.8 billion by September 2026, up from $33 billion last year, a 5.4-fold increase, Menlo Ventures reported on November 3 in its fourth annual State of Enterprise AI report. The report, 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, plus a bottom-up market model, found that agents now generate 80 percent of tokens on OpenRouter, about seven times human-driven usage. Adoption rose only from 55 to 65 percent of companies, so the surge came from existing users spending far more, not from new adopters.
The headline figure, $176.8 billion, is Menlo's estimate, built from survey responses, public filings, and third-party reports, not an audited total. That caveat matters, because the figure is a model, and Menlo discloses that it invests in Anthropic and OpenRouter, two companies whose data underpin parts of the report. Even with that caveat, the direction and scale are striking: enterprise AI spending grew 5.4 times in a year, accelerating from 3.9 times the year before, and now equals more than 30 percent of public cloud revenue, up from just 6 percent a year ago.
The composition of that spending is equally telling. Enterprise AI application spending reached about $77 billion, up from $19 billion, with coding at $34 billion and AI assistants and productivity tools at $25 billion together accounting for about three-quarters of application spend. Anthropic, OpenAI, and Google together capture 69 percent of deployment spending, up from 50 percent last year. Inside enterprise LLM API spending, Anthropic rose from 40 to 55 percent, OpenAI from 27 to 31 percent, and Google fell from 21 to 9 percent. Concentration is increasing, even as total spending explodes.
Agents, AI systems that carry out multi-step tasks on a user's behalf, now generate 80 percent of all tokens on OpenRouter, a platform that routes developers to many models through a single interface. Agent token consumption has grown 37 times since it overtook human traffic in February 2026, compared with five times for humans. That is not a gradual shift. It is a change in who, or what, is using AI: the customer is increasingly another piece of software, acting for a human, rather than the human directly.
That shift explains the spending surge better than adoption alone. The share of enterprises paying for AI rose only from 55 to 65 percent, but average spending reached 4.7 times last year's level against a planned 1.9 times. Companies did not just adopt AI; they deployed it more deeply, gave it more tasks, and let agents run more work autonomously. The report also notes that surveyed companies spent 2.5 times what they had budgeted for the year, which suggests that agent-driven usage is harder to forecast than human usage, because agents do not stop when a human would.
For search, the agent surge connects directly to the agentic layer that Google's CEO described in the same week. If agents generate most AI tokens, then agents are increasingly the ones asking questions, comparing options, and making choices that used to be made by humans typing queries. A business that is invisible to agents, because its data is unstructured, its booking path is broken, or its information is inconsistent, will lose even if it ranks well for humans. Agent readiness, structured data, accurate facts, and working transaction paths, is becoming a search discipline in its own right.
For link builders and content marketers, the agent economy changes what visibility means. An agent that chooses a product will lean on trusted sources, consistent information, and verifiable reputation, because it cannot judge a brand the way a human does. Links, mentions, and reviews from reputable sources become the evidence base for agent decisions. Content that is clear, factual, and machine-readable will be more useful to agents than content that is merely persuasive to humans, and the two audiences will increasingly need to be served by the same pages.
The report measures spending, not returns. It does not show whether the $176.8 billion is paying off in productivity or profit, and Menlo is candid about that limit. A spending surge driven by agents could reflect genuine value, or it could reflect experimentation, over-provisioning, and budget overruns that will correct. SEOs and marketers should not assume that because enterprises spend more on AI, every AI initiative is working. They should measure their own AI spending against outcomes, just as they measure search spending against conversions.
The concentration risk also deserves attention. If three companies capture 69 percent of deployment spending, then the AI ecosystem is more concentrated than the cloud market it is compared with, and a change in one provider's pricing, policy, or model could ripple across many enterprises. For businesses building on AI search, diversification, across models, surfaces, and channels, remains prudent. The agent era rewards those who are visible everywhere agents look, not just on one provider's platform.
Sources: Menlo Ventures, 2026 State of Enterprise AI, November 3, 2026.
Linkbaits helps you create content that earns citations and backlinks. Start with the free tools and build your first linkable asset this week.
Start free trial