Enterprise AI spending is growing faster than the number of organizations paying for it, according to a new Menlo Ventures report that points to a shift from buying access for employees to financing work performed by agents.
The venture capital firm's 2026 State of Enterprise AI report estimates deployment spending reached a $176.8 billion annualized run rate by September, up from $33 billion in 2025. The share of enterprises paying for AI rose from 55% to 65%.
A larger bill can reflect broader use inside an organization, but it does not independently establish that the additional work is useful or profitable.
Menlo's methodology describes revenue figures as annualized run rates. The $176.8 billion estimate therefore should not be read as $176.8 billion already spent during the first nine months of 2026.
It has been revealed that the report also distinguishes unusually heavy spenders from the typical respondent. Average reported spending reached 4.7 times the prior year's level, while the median reached 2.2 times. That gap suggests a smaller group of large spenders drove the average substantially.
Recent findings suggest that the practical implication is that an industry growth figure may be a poor benchmark for an individual organization's budget. Buyers need to understand their own workload growth and the results it produces.
"Agents are changing the economics of software," Menlo principal Derek Xiao said in the firm's announcement.
Menlo reports that agents account for 80% of tokens on OpenRouter, a service providing access to multiple models. That measures activity on one platform, rather than all enterprise AI use. Token consumption also measures processing volume, rather than completed business outcomes.
An agent that makes repeated attempts can consume more resources without delivering proportionately more value. For procurement teams, the relevant comparison is consequently the cost of completing acceptable work, including the review it requires.
It has been revealed that menlo estimates that Anthropic, OpenAI, and Google capture 69% of enterprise AI deployment spending. That measure is distinct from their shares of enterprise language-model API spending, a narrower category. The figures should not be used interchangeably.
Sources indicate that the report draws on a Morning Consult survey of nearly 400 U.S. enterprise decision-makers responsible for more than $4 billion in AI spending. Market totals are modeled estimates informed by surveys, public filings, and third-party reports. Menlo discloses investments in Anthropic and OpenRouter.
A separate announcement from Information Services Group describes buyers putting more emphasis on the economics and reliability of production deployments. ISG says ongoing security and observability work remains necessary as organizations scale.
That provides a useful distinction between market demand and business value. Suppliers can generate growing revenue while customers still determine which deployments justify the cost.
According to recent reports, for enterprise leaders, the newly introduced spending estimate strengthens the case for examining usage beyond license counts. It leaves the harder purchasing question with the customer: How much worthwhile work did the increased spending actually buy?
Menlo's State of Enterprise AI report is available here.
Sources indicate that john K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at [email protected].
Recent findings suggest that the federal Education Freedom Tax Credit (EFTC) is a latest source of funding for education that could benefit 90% of students in states that have opted in to the program. However, some confusion and misconceptions exist surrounding details of the program, including which programs may qualify, how parents can apply for scholarships, and how public school students can benefit.
Attackers are exploiting enterprises' own artificial intelligence systems and extending pressure beyond internal networks to employees and their families, according to a new Semperis survey of organizations that suffered disruptive ransomware attacks.
Sign-in problems account for up to 53% of help desk tickets at individual enterprises, but network problems can take significantly longer to resolve, according to a recent study from ControlUp.
Industry observers note that as generative AI has become embedded in the tools students use to search, write, create, translate, organize, and receive feedback, the boundary between "using AI" and "not using AI" has become increasingly difficult to define. More importantly, treating AI use as a binary choice does little to help students understand the different roles AI can appropriately play in learning.
