Agents now generate 80% of all tokens on OpenRouter, about 130.9 trillion a week and roughly 7x human-driven usage, Menlo Ventures said in 2026: The State of Enterprise AI, published on 3 November. Menlo says agent traffic has grown 37x since it overtook human traffic in early February, against 5x for human usage. By our arithmetic, agent usage grew about 7.4 times as fast as human usage over that stretch.
Menlo says the report draws on a survey of nearly 400 U.S. enterprise decision-makers responsible for more than $4 billion in AI spending, fielded by Morning Consult from September 8 to 26, 2026. Menlo Ventures is an investor in Anthropic and OpenRouter, which the report discloses. The token figures come from OpenRouter, not the survey.
Business Pill · AI Agents Explained
An evergreen Business Engineer explainer on AI agents and the work they automate. It teaches the general idea only and does not discuss Menlo’s report or OpenRouter.
The key insight: By our arithmetic, agent usage on OpenRouter grew about 7.4 times as fast as human usage since the February crossover (37x against 5x, Menlo’s figures), and agents now make 80% of tokens there.
The report opens its agent section with: “The biggest story of the past year has been the rise of AI agents”. It says the launch of Anthropic’s Opus 4.5 in November 2025 “marked an inflection point”, because the model allowed AI to move to “executing complex, multi-step tasks with limited human oversight”.
According to Menlo, agent traffic “tripled the week after Opus 4.5โs launch, then overtook human traffic in early February following the launches of Opus 4.6 and GPT-5.3-Codex”. Menlo’s chart dates the crossover to 6 February 2026.
Menlo’s chart puts the latest week at 130.9 trillion agentic tokens, 18.3 trillion human and 14.9 trillion mixed. The report calls OpenRouter “just one channel” but says it “shows how quickly AI consumption is shifting from humans asking questions to machines doing work”.
Chart: Menlo Ventures, “2026: The State of Enterprise AI” (3 Nov 2026). Survey of ~400 U.S. enterprise decision-makers. Menlo is an investor in Anthropic. Token data: OpenRouter (not the survey), a Menlo portfolio company.
Why Prices Fall as Agents Grow
Menlo links agent usage to a lower average price per token. The report says cache reads “can make up 95% or more of an agentโs tokens”, and that Opus 5.5 charges $0.20 per million for them, against $4 for fresh input and $20 for output.
At a mix of 3% fresh input and 2% output, Menlo says, “an agent session costs about $0.70 per million tokens”. The report adds: “As agents account for more of the usage, the average price per token falls”.
Spend Tied to Work, Not Users
Menlo’s key findings say agents “changed the math”, “tying spend to work performed rather than users”. In its budget section, the report says agents “work independently, run in parallel, and keep going even after people step away, so spending scales with work rather than headcount”.
The report names apps that rebuilt around the change: Harvey and Legora re-architected contract review into multi-agent systems, Sierra launched Horizon, and Factory, a Menlo portfolio company, was “doubling revenue month over month for six straight months through April 2026”.
As we read the report, the agent shift and the pricestory are one mechanism: Menlo says cache reads can be 95% or more of an agent’s tokens and cost $0.20 per million on Opus 5.5, so more agent work means more tokens at a lower average price.
As we read Menlo’s framing, the unit of AI demand is moving from people to tasks: the report says spending scales with work rather than headcount, which is why its OpenRouter series and its budget survey point the same way.
Menlo Ventures, 2026: The State of Enterprise AI
“OpenRouter is just one channel, but it shows how quickly AI consumption is shifting from humans asking questions to machines doing work”
Three Implications
FOR MODEL LABS Menlo ties the tripling of agent traffic to Anthropic’s Opus 4.5 and the February crossover to the launches of Opus 4.6 and OpenAI’s GPT-5.3-Codex.
FOR PRICING Menlo says an agent session at its stated mix costs about $0.70 per million tokens on Opus 5.5, against $4 for fresh input.
FOR APP MAKERS Menlo cites Harvey, Legora, Sierra and Factory as apps that rebuilt around agents; Factory doubled revenue month over month for six months, it says.
The framework argues: “The surface, not the model, governs autonomy.”
As we read it, Menlo’s OpenRouter series measures that shift in tokens: the agent share rose after model launches, and the report credits apps that rebuilt around agents. The framework is the owner’s argument and does not discuss Menlo’s report.
The 80% share and token counts are OpenRouter data as charted by Menlo, not survey results. Menlo’s methodology says OpenRouter’s users “skew more toward individual developers and startups than the enterprise buyers in Menloโs survey”, and that the data split weekly tokens by API-key type (agentic, human, mixed).
Menlo is an investor in OpenRouter and in Anthropic, whose model launches the report ties to the shift. The 7.4 times figure is our arithmetic on Menlo’s 37x and 5x growth figures, which run from the February crossover; we did not check the underlying weekly series.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Menlo Ventures says agents now make 80% of tokens on OpenRouter, about 130.9 trillion a week. By our arithmetic, agent usage grew about 7.4 times as fast as human usage since the February crossover, and Menlo says the shift is pulling the average price per token down.
A note on sourcing. The figures, survey details and quotes come from Menlo Ventures’ 2026 State of Enterprise AI report, published on 3 November 2026, and where cited its December 2025 report. The charts in this piece are Menlo’s own, reproduced with Menlo’s attribution. Menlo Ventures is an investor in Anthropic and OpenRouter, among other companies the report names. The one figure labelled our arithmetic is ours, computed from Menlo’s numbers. We did not contact Menlo or the companies named.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
