Your AI Economics Pulse for October 2026

Last month we called the median the least informative number in this report. September proved us wrong.

The median company’s AI share posted its biggest move on record. The top of the market kept climbing, just slower. New adopters barely registered.

The middle caught up.

Same window, different lens: in a new CloudZero survey, 252 finance leaders told us their FY2027 plans while these bills were running. They plan to spend more, but most can’t say what the last round bought.

We’ll get into the plan at the end, but first, a look at our three major metrics for the month: share, intensity, and breadth.

Download Now

Get your own copy of the AI Economics Pulse now

The October Pulse digs into the median’s biggest AI spend jump on record, who’s driving it, and what it means for FY2027 budgets.

Share: the median crossed 3%

Here, we track AI spend as a share of each organization’s total cloud bill, reported as the panel median with the 25th and 75th percentiles.

The median company spent 3.02% of its cloud bill on AI in September, up sharply from 2.66%. That’s the 14th straight monthly increase and the largest of the 14, one month after August posted the smallest. It’s also now 4.5 times the level of a year ago.

The edges moved too. The 75th percentile rose nearly 1.5 percentage points to 12.55%. The 25th percentile reached 0.27%, from 0.17%. Still a small number, but it’s a 65% jump.

Also, last month, the 75th percentile sat 4.2 times above the median, the widest gap since we began publishing. In September it narrowed to 4.1. The middle closed some distance.

Now, how big do finance leaders think that number is? In our survey cited above, the median respondent put AI at 21% to 35% of infrastructure spend. More than half (54%) said above 20%. The billed median is 3.02%.

Part of that gap has an explanation. Survey respondents count AI bundled into SaaS subscriptions, which a cloud bill doesn’t show, and they’re a separate sample from our panel. 

But the estimate doesn’t track visibility. Leaders who see only a total bill or a rough category split claimed AI above 20% as often (54%) as those with team-level allocation (53%).

So, treat 3.02% as a floor. Treat 21% to 35% as a guess. The real number sits between those two figures, and without the needed allocation and visibility capabilities, most companies can’t say where that number lives.

Intensity: the 25% tier came back

Here, we count the share of organizations where AI is at least 1%, 5%, 10% or 25% of total cloud spend. Each threshold is relative to the company’s own bill.

In last month’s Pulse, the 25% tier posted its first decline since January. We offered two explanations, rationalization or arithmetic, and said we’d report which way it resolved.

It rebounded, to 8.6% from 8.2%.

That points to arithmetic: total cloud spend outran AI for a month, then AI caught up. One month isn’t proof, but it does argue against a retreat.

Elsewhere, the climb slowed. The 10% tier reached 29.2%, up 1 point after August’s record 4.3. The 5% tier hit 42.5%. The 1% tier, 64.1%.

The survey explains why nobody’s retreating. Respondents who put AI above 35% of infrastructure spend are the most uneasy: 72% say they spend more than they can justify, against 50% of those at 20% or less.

They’re also the most committed. 59% plan FY2027 increases above 20%, against 25% of lighter spenders.

The pressure is landing on process. The number keeps going up. Among teams that can’t justify current spend, 93% are restructuring FY2027 budgeting. Teams adding approval gates plan 20%-plus increases at the same rate as teams tying funding to ROI: 39% each.

Breadth: the line stopped moving

Here, we track the share of organizations spending at least $1,000 on AI in a month. It’s an absolute floor, not a percentage. It marks the line between running a pilot and running nothing.

Nearly seven out of 10 (68.0%) cleared that floor in September, up from 67.7%. That’s 11 straight monthly gains, but this one was 0.3 points after August’s 2.0, the smallest jump in the streak.

Last month we said three quarters of the panel would clear the line before spring. That assumed a continuation of the pace we saw in summer. At September’s pace, it looks closer to two years out.

Breadth is interesting on its own, but read it against the share data above. The median jumped while breadth stalled. That means the growth came from companies already spending on AI, not from new ones.

The CloudZero survey agrees, and also shows where the money comes from. 87% of finance leaders are raising FY2027 AI budgets. Only 10% call it net-new money. 56% expect AI to displace SaaS licenses in FY2027, up from 45% this year.

AI is being paid for out of other lines, starting with SaaS licenses.

FY2027: the plan is bigger than the bill

A year ago, Forrester predicted enterprises would defer 25% of planned AI spend into 2027. 

What happened in the end? If anyone deferred, it doesn’t show in the bills. This September, the median posted its largest monthly gain on record.

The token got cheaper. The bill didn’t.

The Silicon Data LLM Token Expenditure Index, a usage-weighted price per million inference tokens, fell to 97 cents on August 31, its lowest since launch and less than half its summer high, per CNBC. It still sat at $0.97 on October 11. Over the same stretch, AI kept taking a bigger share of the cloud bill.

Tokens got cheaper, but companies used more of them which drives the bill up anyway. 

Budgets aren’t holding

Futurum Research found 46.9% of enterprises over their AI budgets in its second-half survey, and 5.6% under. In CloudZero’s FY2027 survey, 51% ran over their FY2026 plan.

Finance’s lesson from the overrun: spend more. 57% of our over-budget respondents say they should have set a larger budget. In Futurum’s data, 47.6% of over-budget organizations went back for more money.

The numbers get approved blind

According to CloudZero’s survey report, 44% of finance leaders approved AI spend without reliable projections. Only 12% can break it down finely enough to act within the quarter. And 61% already say they spend more than they can justify.

CJ Gustafson’s advice for this planning season in Mostly Metrics: budget each role toward what its heaviest users already spend, not a company-wide average. His line: “AI doesn’t get grandfathered in.”

What this means for October

Budget season is here. Four things belong in the FY2027 plan.

In short: the median finally moved. For FY2027, the open question is whether anyone can see what the spending bought.

See where you land

CloudZero’s AI Benchmark Tool shows how your AI spend governance compares with 260 finance leaders. For the full finance view, download Why Finance Caps What It Can’t See.

Frequently asked questions

What percentage of cloud spend goes to AI in 2026?

Across CloudZero’s same-store panel of 422 customer organizations, AI reached 3.02% of the median company’s total cloud bill in September 2026, up from 2.66% in August, the largest monthly gain since tracking began. The 75th percentile was 12.55%. The figure counts only AI cleanly attributed in cloud and infrastructure bills, so treat it as a floor.

Did the share of companies spending 25% or more on AI recover?

Yes. The tier rose to 8.6% in September from 8.2% in August, after its first decline since January 2026. The rebound suggests August’s dip reflected total cloud spend briefly outpacing AI spend, not companies cutting back. One month isn’t conclusive.

Is AI adoption still spreading to new companies?

Slowly. 68.0% of the panel spent at least $1,000 on AI in September, up 0.3 points. That’s an 11th straight gain and the smallest of the streak. Growth is coming from companies already spending: the median AI share rose 37 basis points in the same month.

How much are companies planning to spend on AI in FY2027?

87% of finance leaders plan to raise AI budgets in FY2027, according to CloudZero’s survey of 252 finance leaders, Why Finance Caps What It Can’t See. 61% say they already spend more on AI than they can justify with measurable outcomes.

Author Spotlight

Keith MacKenzie

Keith is CloudZero's Content Marketing Manager with a specialty in topics around AI & cloud spend. He also brings more than a decade's worth of background as an editor and writer in the mainstream media and content marketing industries. He's also been to Chernobyl twice – and (sometimes) still has all his marbles.