Quick Answer
Across a same-store panel of 428 CloudZero customer organizations, AI reached 2.61% of the median company's cloud bill in July 2026, up roughly fourfold in a year. Nearly two in five organizations (39.6%) now run AI at 5%+ of their cloud bill, and the median company's AI spend still grew 13.3% month over month. The market keeps asking whether AI is slowing down. The spend data says it isn't. What's missing is proof that the spend is paying off.
There’s a running argument this summer about whether the AI boom is finally cooling. The people writing the biggest checks answered it on their earnings calls a few days ago.
The hyperscalers are spending more. In their latest quarter, Amazon, Google, and Microsoft committed roughly $595 billion in combined 2026 capital spending. That money buys AI data centers and the chips that fill them. They still can’t build fast enough.
Amazon CEO Andy Jassy raised the company’s number to about $220 billion: “We will still not have enough capacity to meet all the demand we have in 2026, and I believe this dynamic will also be true in 2027 too.”
Google’s CFO Anat Ashkenazi said much the same: “The demand still outpaces that investment. And we are, just like the rest of the industry, working in a supply-constrained environment.”
That’s the supply side. As a CFO, the question you care about is the demand side: is your AI spend slowing? Our panel says no.
Here’s what July looked like across three readings.
Share: AI keeps taking a bigger slice of the bill
Let’s start with share. The median company in our panel spent 2.61% of its total cloud bill on AI in July, up from 0.66% a year earlier. That’s a near-fourfold increase in 12 months, and it rose another 24 basis points in July.
AI’s share of the cloud bill has set a new high every month for the past year.
AI/ML share of total cloud spend, 12 months ending July 2026
Panel median with 25th–75th percentile band, trailing 12 months.
Median25th–75th percentile
Source: CloudZero data (anonymized and aggregated)
View the data: AI/ML share of total cloud spend by month, August 2025 – July 2026
| Usage month | Median | 25th percentile | 75th percentile |
|---|---|---|---|
| August 2025 | 0.66% | 0.09% | 2.65% |
| September 2025 | 0.67% | 0.07% | 2.66% |
| October 2025 | 0.69% | 0.04% | 2.69% |
| November 2025 | 0.72% | 0.02% | 2.60% |
| December 2025 | 0.74% | 0.01% | 2.55% |
| January 2026 | 1.01% | 0.04% | 3.19% |
| February 2026 | 1.29% | 0.05% | 4.33% |
| March 2026 | 1.42% | 0.07% | 5.61% |
| April 2026 | 1.74% | 0.06% | 6.93% |
| May 2026 | 2.08% | 0.10% | 7.68% |
| June 2026 | 2.38% | 0.10% | 8.39% |
| July 2026 | 2.61% | 0.11% | 9.63% |
The median is the calmest part of this story. The top quartile of companies is already putting close to a tenth of its cloud bill into AI. For those organizations, AI has evolved from a line item to a top-of-P&L cost that requires the same financial rigor as headcount or infrastructure.
Worth flagging: 2.61% is the AI spend we can cleanly attribute inside cloud and infrastructure bills. It doesn’t capture the AI now bundled into software you already pay for, like Gemini folded into Google Workspace or Copilot into Microsoft 365. That spend is real; it’s just invoiced as “Workspace” or “Office,” not “AI.” So, treat 2.61% as a floor and the MoM increase as the real indicator. The true AI footprint is higher.
Report
Finance needs to prove AI’s return: CloudZero report
260 senior finance leaders (more than half CFOs) told us why the speed of seeing AI spend, not the size of it, separates who pulls ahead on AI from who gets burned.
Intensity: more companies are going deep on AI
The second reading is intensity: not whether a company spends on AI, but how far AI has pushed into its cloud bill. We track the share of organizations where AI is at least 1%, 5%, 10%, or 25% of total cloud spend. Because each threshold is relative to a company’s own bill, it reads the same whether you do $50 million in revenue or $500 million.
AI adoption intensity, 12 months ending July 2026
Share of panel organizations where AI is at least 1%, 5%, 10%, or 25% of total cloud spend, trailing 12 months.
≥ 1%≥ 5%≥ 10%≥ 25%of each company’s cloud spend
Source: CloudZero data (anonymized and aggregated)
View the data: AI adoption intensity by month, August 2025 – July 2026
| Usage month | ≥ 1% | ≥ 5% | ≥ 10% | ≥ 25% |
|---|---|---|---|---|
| August 2025 | 44.1% | 12.1% | 7.1% | 3.6% |
| September 2025 | 43.1% | 15.8% | 7.4% | 2.7% |
| October 2025 | 45.1% | 16.3% | 6.9% | 2.2% |
| November 2025 | 46.0% | 16.7% | 7.8% | 2.4% |
| December 2025 | 45.9% | 16.2% | 9.7% | 2.3% |
| January 2026 | 50.5% | 17.6% | 10.4% | 2.2% |
| February 2026 | 53.6% | 21.6% | 10.8% | 2.7% |
| March 2026 | 54.9% | 28.2% | 13.0% | 3.4% |
| April 2026 | 57.5% | 31.4% | 16.7% | 4.8% |
| May 2026 | 61.9% | 33.8% | 19.7% | 5.6% |
| June 2026 | 61.3% | 36.3% | 21.6% | 6.7% |
| July 2026 | 62.5% | 39.6% | 23.9% | 9.1% |
The entry tier is broadening: 62.5% of organizations now put at least 1% of their cloud bill toward AI, up from 44.1% a year ago. The faster movement is in the deeper tiers. The share spending at least 5% of their cloud bill on AI more than tripled in a year, from 12.1% to 39.6%. The 10% tier followed the same path, from 7.1% to 23.9%, nearly one in four companies. The middle of the market is turning dabbling into real commitment.
The heavy end is the one worth watching. The share of organizations spending a quarter or more of their cloud bill on AI sat flat, even drifting down, through the first half of the year, then took off. Since January it has climbed every month, from 2.2% to 9.1%. Nearly one in ten companies now runs AI at 25% or more of total cloud spend, a cohort that was just emerging six months ago.
For finance, the takeaway is depth. More companies are pushing AI to a material share of the bill, which is exactly where cost-to-serve and margin start to bend. That brings us to how fast that spend is still moving.
Velocity: the typical company's AI spend is still compounding
Velocity is the third reading. The median organization grew its AI spend 13.3% month over month in July. Compounding at that rate roughly doubles a company's AI spend every six months.
AI spend growth, month over month, 12 months ending July 2026
Median month-over-month AI-spend growth per organization, with 25th–75th percentile band, trailing 12 months.
Median25th–75th percentile
Source: CloudZero data (anonymized and aggregated)
View the data: median month-over-month AI-spend growth per organization, August 2025 – July 2026
| Usage month | Median MoM growth | 25th percentile | 75th percentile |
|---|---|---|---|
| August 2025 | 4.1% | −8.9% | 22.3% |
| September 2025 | 4.9% | −7.2% | 28.1% |
| October 2025 | 5.0% | −6.2% | 26.2% |
| November 2025 | −2.3% | −14.4% | 18.3% |
| December 2025 | 3.4% | −9.0% | 21.8% |
| January 2026 | 3.8% | −6.0% | 28.8% |
| February 2026 | 4.1% | −10.2% | 36.4% |
| March 2026 | 21.2% | 2.1% | 69.2% |
| April 2026 | 10.3% | −7.6% | 39.5% |
| May 2026 | 10.8% | −7.8% | 38.5% |
| June 2026 | 13.3% | −4.2% | 42.2% |
| July 2026 | 13.3% | −3.9% | 41.1% |
The spread underneath that median is the part that matters. The middle of the panel ran from about 4% down at the 25th percentile to more than 40% growth at the 75th. Some companies are flooring it; some are pulling back hard in the same month. That dispersion, not the average, is the real signal, and increasingly the difference comes down to whether they can see what the spend is doing.
The bill has no gate
Put the three readings together:
- AI spend is large.
- It's everywhere.
- It's still accelerating.
So the pressing question for finance has moved from "how much are we spending?" to "what are we getting for it?" And that's exactly where most teams are stuck.
Gustafson names the structural problem: token spend, unlike every other major cost a CFO controls, has no checkpoint.
Headcount goes through approvals. Software has renewals. "Token spend doesn't really have a gate (yet)," he wrote, and "the last cost that behaved this way was cloud, which took the industry about a decade to get a handle on." This time, a decade is a luxury finance doesn't have. In CloudZero's June 2026 survey of 260 finance leaders, 87% said they need to tie AI spend to business outcomes within the year.
Gustafson's prescription is the one we could write ourselves: drive the spend down to a unit. "Every variable cost eventually gets driven down to a per unit number," he argued, and token cost per employee per month "is useless on its own because you don't know what they're using those tokens for."
What matters is "cost per ticket resolved, cost per document reviewed, cost per deal cycle, cost per claim completed." His conclusion: "When everyone has access to the same models and tools, the company that does the best work at the cheapest trackable cost wins."
That's the gap between spend and proof, and it's wide: only 22% can tie AI spend to outcomes today. Spend isn't the risk; most orgs can already track the money. Blind spend is. The companies that can read AI spend as a per-outcome signal, cost per customer, per feature, per dollar of revenue, are the ones funding what works while everyone else freezes.
We opened with the hyperscalers pouring roughly $595 billion into AI capacity this year and still insisting they can't build it fast enough. Our panel is the other end of that same pipe: the demand those data centers exist to serve, accelerating up and down the customer spectrum. Microsoft, Google, and Amazon keep raising their capex guidance; our customers keep raising their AI bills. Both sides are running hot, and neither number is bending.
That's what finance can't wish away. The spend curve won't flatten and hand teams time to catch up on measurement; it compounds every month. The leaders who build the per-outcome view now, while the spend is still climbing, keep investing with conviction. Everyone else keeps reading the story off the invoice, after the money is already gone.
See where you land
Curious how your own AI economics compare to the panel? CloudZero's AI Benchmark Tool shows where your AI spend sits against your peers, so you can walk into the next board meeting with a number instead of a shrug.
FAQ
