Gartner dropped a number on July 20, 2026 that should make every marketer stop scrolling. Worldwide spending on AI platforms and models will hit $64 billion this year, up 63.4 percent from $39 billion in 2025. That's on top of the broader $2.52 trillion Gartner projects for total AI spending in 2026, a 44 percent jump year over year.
Nobody is asking where the money comes from. That's the problem.
Goldman Sachs estimates total AI capital expenditures at roughly $765 billion in 2026. Companies are not minting that cash from thin air. They are pulling it from existing budgets, existing teams, and existing programs. And marketing is one of the first places they reach.
The Money Has to Come from Somewhere
When a CFO sees a 63 percent increase in AI platform spending, they don't open a new budget line. They look at what they already have and start moving numbers around. The Gartner forecast is a spending number, not a creation number. That $64 billion has to be extracted from somewhere in the organization.
U.S. firms are now spending an average of $2,068 per employee on AI in 2026, up 50 percent from last year. Professional and business services firms are spending $3,470 per employee, a 74 percent increase. That is real money being redirected from salaries, tools, training, and marketing programs into AI infrastructure and platform contracts.
A CEO at one company reportedly told employees they wouldn't get raises in 2026 because the budget was going to AI services instead. That is the honest version of what is happening everywhere, just usually said more quietly.
This connects to something I wrote about in the AI budget illusion. Companies report AI spend as growth, but a significant chunk of it is displacement. The money moves from one column to another, and the column it leaves is usually the one that actually generates revenue.
Oracle Made It Obvious
Oracle cutting up to 30,000 employees is the clearest example of the pattern. Eighteen percent of its global workforce, gone, to free an estimated $8 to $10 billion in annual cash flow for AI infrastructure. The cuts hit Oracle Health, cloud infrastructure, and consulting hardest. The teams building Stargate data centers were spared.
The internal allocation tells you everything. Oracle is converting itself, department by department, into an AI infrastructure provider and financing the transformation with the salaries of the businesses it is deprioritizing. Consulting and customer-facing roles got cut. Data center construction got funded.
Oracle is not alone. Reuters reported in July that companies across industries are cutting jobs as investments shift toward AI. Big tech has cut roughly 142,000 jobs in 2026 to fund over $700 billion in AI capex. Meta, Amazon, Dell, and Salesforce all made similar moves, trimming headcount while pouring billions into compute.
The pattern is simple: cut people who do the work, buy the machines that might eventually do the work, and hope the gap between the two doesn't show up on the income statement before the AI delivers.
The Per-Employee Math That Should Scare CMOs
The $2,068 per-employee AI spend figure is the one marketers should sit with. If you run a 50-person marketing organization, that's $103,400 per year in AI platform costs that didn't exist in your budget two years ago. For a 200-person team, it's over $400,000.
That money is not additive. CFOs are not handing marketing teams an extra $400K and saying "go buy some AI tools." They are taking it from the marketing budget and handing it to IT, procurement, or the new "AI initiatives" line item that reports to the COO.
Gartner separately warned that at least 30 percent of generative AI projects would be abandoned after proof of concept. So a third of that redirected marketing budget is going to tools that will get shelved before they produce a single campaign, a single lead, or a single dollar of revenue.
This is the ROI proof problem at scale. Marketing budgets are being cannibalized to fund AI experiments that may never ship, while the programs that actually drive pipeline are being asked to do more with less.
What This Looks Like Inside a Marketing Team
Here is what the cannibalization looks like at the ground level, based on what companies are reporting.
A brand team loses its creative director because the budget went to an AI content platform. The platform produces content, but nobody with creative judgment is left to evaluate whether it's on-brand, differentiated, or even coherent. Output goes up. Quality goes sideways.
A performance marketing team loses two analysts because the spend went to an AI bidding optimization tool. The tool optimizes bids, but nobody is left to analyze why the strategy is working, whether the channel mix still makes sense, or whether the customer acquisition cost is actually improving or just looking better on a dashboard.
A content team gets cut in half because AI writing tools are "so much cheaper." The remaining team spends most of its time editing, fact-checking, and rewriting AI output, which is a different and often slower kind of work than writing from scratch. Productivity per person looks fine on paper. Total output quality drops.
The common thread: the AI tool gets purchased, the people who would have used it strategically get cut to pay for it, and the organization ends up with powerful tools and no one who knows how to deploy them with intent. That is not efficiency. That is the shadow AI economy playing out in budget line items.
The ROI Gap Nobody Talks About
The marketing AI pitch is usually some version of "do more with less." AI reduces marketing overhead costs by roughly 10.8 percent according to recent benchmarks. For a 20-person marketing team costing $120,000 per person, that's a potential $259,000 in savings.
But that math only works if the AI actually delivers equivalent output. And so far, the evidence is thin. Gartner's 30 percent abandonment rate for genAI projects is a polite way of saying a third of these investments produce nothing. The CFOs coming for the AI budget know this, which is why scrutiny on AI ROI is intensifying even as spending accelerates.
Meanwhile, Google's own AI efforts are faltering. Gemini 3.5 Pro missed its July 17 target for the third time. Alphabet shares dropped about 4 percent on the delay. If the company spending billions on AI can't ship its flagship model after three attempts, what are the odds that a mid-size company's $100K AI platform subscription is going to transform its marketing?
This matters for marketers because so much of the AI infrastructure spending boom is justified by projected productivity gains that haven't materialized. The cost escalation crisis is real. The returns are theoretical. And the budgets being cannibalized to fund it are the ones that actually drive revenue today.
The Question Every CMO Should Be Asking
The smart move is not to resist AI spending. It's to make sure the money being redirected from your budget is going to tools that actually help your team, not just tools that look good in a board presentation.
Here is the question: when the CFO comes for $200K of your budget to fund an AI platform, what evidence do you have that the platform will produce more pipeline than the programs you're being asked to cut?
If the answer is "the vendor said so," you have a problem. If the answer is "we ran a 90-day pilot and measured it against our actual funnel," you might be fine.
The companies that survive the AI infrastructure redirect will be the ones that demanded proof before they surrendered budget. The ones that don't will end up with expensive AI tools, smaller teams, and a marketing function that produces less than it did before the "transformation" started.
That's not a strategy. That's just a layoff with a software subscription.