Enterprise AI Spending Hits $177B, Up 5.4x in One Year

Enterprise AI spending reached $176.8 billion by September 2026, a 5.4x jump from $33 billion a year earlier, according to Menlo Ventures' fourth annual State of Enterprise AI report, released November 3, 2026. The growth accelerated from 3.9x the previous year, and the report attributes most of it to autonomous agents rather than new human users.

"Wall Street is questioning whether the massive investments in AI will pay off, but our latest data shows just how quickly enterprise demand is growing," said Tim Tully, Partner at Menlo Ventures. "Enterprise AI spending has jumped more than fivefold, from $33 billion last year to $177 billion this year. Based on what we've heard from enterprise buyers, companies aren't pulling back. They're finding more ways to put AI to work and agents are driving a lot of that demand. If anything, we may still be underestimating just how big this market can become."

The finding cuts against the claim that AI budgets are contracting. It also reframes who — or what — actually consumes AI inside companies.

What is driving the $177 billion?

Agents, not people. On OpenRouter, a routing platform for AI models, agents now generate 80% of all tokens — roughly seven times human-driven usage. Agent token consumption has grown 37x since overtaking human traffic in February 2026, compared with 5x growth in human-driven usage. OpenRouter is one channel, but it shows how quickly AI consumption is shifting from humans asking questions to machines doing work.

The money also outran the plans. Companies entered 2026 expecting to spend more, and budgeted nearly twice what they had spent in 2025. They then spent 2.5 times what they had budgeted, according to the report's survey.

"Agents are changing the economics of software. When AI can do the work itself, enterprise spending is no longer constrained by the number of employees using it, but by how much work it can get done," said Derek Xiao, Principal at Menlo Ventures. "That demand is growing faster than even the investment required to build these models. For every dollar spent on AI model training this year, enterprises are spending $1.70 putting AI to work."

Adoption, by contrast, grew slowly. The share of enterprises paying for AI rose only from 55% to 65%, meaning growth came primarily from heavier spending among existing adopters rather than a wave of new buyers.

Is AI cannibalizing the software market?

No, according to the data. New budgets funded 55% of incremental enterprise AI spending. Another 22% came from existing software budgets, most of it staying with incumbents as they convert licenses to AI offerings — Microsoft Office to Copilot being the report's example. The research found little evidence of enterprises ripping out core systems such as Salesforce's CRM or Workday's HRIS.

Applications became the growth engine. Enterprise AI application spending hit roughly $77 billion, up from $19 billion last year. Two categories account for 76% of that spending:

• Coding: $34 billion
• AI assistants and productivity tools: $25 billion

Healthcare remains the largest vertical market at $3.7 billion.

Where do startups beat incumbents — and where do they not?

The report draws a sharp line between new workflows and established ones. AI-native providers capture 96% of coding spending, because they are fundamentally changing how software gets built. Incumbents lead the mature categories:

• IT: 91%
• Data science: 88%
• HR: 68%

"AI isn't disrupting every software category in the same way," said Joff Redfern, Partner at Menlo Ventures. "Our data shows when AI changes how work gets done, new companies have an opening to reshape a market. When AI improves the way people already work, established players are harder to displace."

Who captures the model spending?

The frontier labs, and increasingly so. Anthropic, OpenAI and Google together account for 69% of deployment spending, up from 50% in 2025. Within enterprise LLM API spending, the shifts were sharper:

• Anthropic rose from 40% to 55%
• OpenAI rose from 27% to 31%
• Google fell from 21% to 9%

Open-source models handle approximately one in five tokens but account for only about one in 25 dollars of model spending — a gap that favors the closed, frontier labs.

"Right now, the best models are capturing the lion's share of enterprise," said Deedy Das, Partner at Menlo Ventures. "Anthropic grew its share of enterprise API spending from 40% to 55%, while Google fell from 21% to 9%. That's how quickly the competitive landscape can change. The labs have a tremendous advantage because building frontier models is incredibly hard, but no one's position is guaranteed. Every major model release is an opportunity to shift market share."

Menlo Ventures discloses that it is an investor in Anthropic and OpenRouter.

How Menlo built the numbers

The 2026 edition combines a survey of nearly 400 U.S. enterprise decision-makers overseeing more than $4 billion in AI spending with a bottom-up market analysis. Menlo built its market-size estimates from survey responses, public filings and third-party reports; token-share figures draw on OpenRouter data and Menlo's estimates of model mix. The report is authored by Tim Tully, Joff Redfern, Deedy Das and Derek Xiao.

Menlo Ventures, which marked its 50th anniversary in June 2026 and announced $3 billion in fresh capital to back AI companies, is betting its own balance sheet on the trend it is measuring — its portfolio spans Anthropic and OpenRouter on the infrastructure side and applications including Lovable, Legora, OpenEvidence and Wispr Flow.

The report's central signal for buyers and investors: enterprise AI demand is outrunning both budgets and the capital needed to build the models, and the next major model release could again reorder a market where 69% of spending already sits with three labs.