Markets

Four AI spending reports in one week. The budget is outrunning the measurement.

Menlo puts enterprise AI spend at $176.8 billion. Only 10% of CFOs in an RGP survey see significant returns. What the four reports agree on, and what to do before FY2027 locks.

Edited by Emilio Santos 6 min read
An open-plan office with desks and screens
Most FY2027 AI lines will be set on totals. Photo: Circuit & Capital

Four surveys on enterprise AI spending landed between November 2 and November 5, 2026. Read together, they say the same uncomfortable thing: AI budgets are rising faster than finance teams can measure what the money does. For anyone setting FY2027 numbers this month, that gap is the part worth acting on.

Here is what each report found, what the sources have at stake, and how to use the numbers at your own budget table and at your next software renewal.

The four reports

Menlo Ventures: spending up 5.4 times

Menlo Ventures’ 2026 State of Enterprise AI report, released November 3, puts enterprise AI spending at $176.8 billion by September 2026, up from $33 billion in 2025. The share of enterprises paying for AI rose only from 55% to 65%, so most of the growth is existing buyers spending more, not new buyers arriving.

Two other findings matter for budget owners. New budgets funded 55% of incremental enterprise AI spending and 22% came from existing software budgets, mostly staying with incumbents as they convert licenses, such as Microsoft Office to Copilot. And Anthropic, OpenAI and Google together took 69% of deployment spending, up from 50% in 2025. Menlo says companies budgeted nearly twice what they spent in 2025 and still ended up spending 2.5 times what they budgeted.

Read it with care. The market-size figures are Menlo’s own estimates from a bottom-up model, and Menlo discloses that it is an investor in Anthropic and OpenRouter, both of which appear in its findings. The survey covered nearly 400 U.S. enterprise decision-makers.

West Monroe: 74% now put more than a tenth of IT into AI

West Monroe’s The New Rules of Enterprise Software & AI Spend, released November 5, found that the share of buyers putting more than 10% of their IT budget into AI rose from 23% in 2025 to 74% in 2026. The online survey ran in Q3 2026 among 300 U.S. executives; buyers came from organizations with at least $100 million in revenue.

The useful finding is about pricing. Buyers prefer predictable models: enterprise subscriptions (33%) and usage-based pricing (24%). Yet 36% of sellers think buyers prefer hybrid pricing that includes outcome-based components. Sellers also underestimate how often buyers simply buy: 42% of buyers say they mainly buy AI rather than build it, while only 22% of sellers think so.

RGP via CFO.com: 10% see significant returns today

CFO.com reported on November 3 that consulting firm RGP surveyed 400 finance chiefs in September at companies with at least $250 million in revenue across North America, Europe and Asia Pacific. Only 10% said they see “significant positive” ROI from AI today; 54% expect it within 24 months. At companies under $1 billion in revenue, just 1% reported significant AI ROI, against 17% at companies of $5 billion or more. RGP’s own framing: “The 24-month AI promise is now on the record, and boards will hold CFOs to it.”

CloudZero: raising budgets they cannot justify

CloudZero, which sells cloud cost software, published a survey of 252 senior finance leaders on November 2. It is a vendor survey, so treat it as one input. It found 87% are raising their AI budget for FY2027, 45% of those by more than 20%, while 61% already spend more on AI than they can justify with measurable outcomes. Among those who cannot justify current spending, 91% are increasing it anyway. Only 12% can break AI spend down finely enough to act on it within the quarter, and 54% cannot get below a rough category.

Where the numbers agree, and where they do not

The reports measure different things for different audiences, so do not add them up. Menlo sizes a market; West Monroe and RGP survey large buyers; CloudZero surveys finance leaders on its own subject. What they share is direction. Spending is up sharply, the buyers already paying are paying more, and few can show the return yet.

The disagreement is commercial. West Monroe’s sellers are moving toward outcome-linked pricing that buyers did not ask for. That is a lever for buyers, not a threat, as long as you know your own usage.

What it means for CFOs and CISOs

  • Allocation is the gap. If only about one in eight finance teams can attribute AI spend within a quarter, most FY2027 AI lines will be set on totals. Totals only ever go up.
  • Some of the increase is a renewal. Menlo’s 22% from existing software budgets is license conversion. Copilot-style upgrades should go through the same renewal review as any other price increase, not a separate AI budget.
  • The ROI clock is public. RGP’s 24-month figure gives boards a date. Write down what “paid off” means now, in cost per task or hours saved, so the checkpoint is not argued after the fact.
  • Concentration is a supplier risk. With 69% of deployment spend going to three labs, a price change at one of them reaches you through every tool built on it. CISOs should know which vendors route data to which model provider.
  • Smaller firms should be most careful. The 1% figure for companies under $1 billion is a reason to start small and measure, not to stop.

What to do this week

  1. List every AI line in the FY2027 draft: API usage, AI add-ons on existing suites, standalone tools. Note owner, team and contract end date.
  2. Require a unit for each line. Cost per task, per ticket or per team. Our free AI cost-per-task calculator and token cost and margin calculator give a first estimate from vendor prices.
  3. Treat license conversions as renewals. Run each one through the software renewal scorecard and ask for a predictable subscription cap in writing rather than an open usage or outcome term.
  4. Set a dated ROI checkpoint. Pick a date inside 24 months and the metric you will use. Put it in the budget memo.
  5. Ask vendors which model they use and whether a change of model provider changes your price or data terms.

If you want a second pair of eyes on the renewals behind these lines, our Renewal Audit reviews up to 25 software renewals for a fixed $199 and returns a written report in 5 business days. The Renewal & Risk Operator Pack ($29) has the renewal scorecard and tracker as spreadsheets.

What to watch next

  • The full reports. Menlo and West Monroe have published full reports with methodology; check sample definitions before quoting a figure internally.
  • Seller pricing changes at renewal. If a vendor proposes outcome-based or hybrid pricing, ask for the formula, a cap and a fallback to a fixed subscription.
  • Copilot and suite conversions. Our Microsoft 365 renewal briefing covers the July 1 list prices that are reaching customers at renewal now.
  • Your own quarter-end numbers. The best data point for your FY2027 AI line is what you spent this quarter, by team.

Sources

Related: Gartner revised the AI bill and all AI cost briefings.

ES
Emilio Santos edits Circuit & Capital. How we source and correct pieces: editorial standards.
circuits50capital@circuits.fit