Jensen Huang has a check for whether or not an engineer is price retaining, and it comes with a token funds hooked up. On the All-In Podcast on the shut of GTC 2026, the Nvidia chief government mentioned that if a US$500,000 engineer’s annual AI token consumption got here in beneath US$250,000, half their wage, “I’m going to be deeply alarmed.” Nvidia, he confirmed, is working towards a US$2 billion yearly token invoice for its engineering drive.
It’s a memorable provocation from the person who sells the compute. It’s additionally a tidy description of the trade-off now being made in company budgets in all places, often with much less candour: cash that after paid folks is more and more being paid for tokens. The query the trade has been slower to ask is whether or not that commerce is definitely working, and the sincere solutions arriving from the businesses that moved first counsel it typically isn’t.
The place the cash went
The reallocation itself isn’t in dispute. The 4 largest hyperscalers have guided roughly US$700 billion in mixed 2026 capital expenditure, almost double final 12 months, whereas Gartner projects AI agent software program spending will attain US$207 billion, up 139%. On the opposite facet of the ledger, Challenger, Grey & Christmas data exhibits AI because the most-cited cause for US job cuts for a report fourth straight month, with tech accounting for 31% of first-half layoffs.
An inside Meta memo obtained by Reuters described Might’s cuts of 8,000 roles as offsetting the firm’s substantial investments, at the same time as income grew 33% that quarter. Oracle’s filings present headcount down 21,000 as financial savings feed its information centre buildout. These are extremely worthwhile corporations. The layoffs aren’t survival measures. They’re financing.
Andy Challenger’s abstract of his agency’s information is the plainest out there: “Corporations are shifting budgets towards AI investments on the expense of jobs.” The duty a employee carried out might not have been automated in any respect. The funds that paid for it has merely moved.
What the cash purchased
Right here, the report turns awkward. Gartner surveyed 350 executives at corporations with over US$1 billion in income, all deploying AI brokers or automation, and located roughly 80% had lower headcount, with no correlation between the cuts and improved returns. Analyst Helen Poitevin’s verdict: “Workforce reductions might create funds room, however they don’t create return.”
Her analysis discovered the organisations that did enhance ROI had been these utilizing AI to amplify their folks fairly than take away them. The token facet of the ledger has its personal reckoning underway.
Uber gave 5,000 engineers AI coding instruments in December and exhausted its complete 2026 AI funds by April. Chief working officer Andrew Macdonald conceded that regardless of 70 per cent of dedicated code being AI-generated, the connection to something prospects expertise is lacking: “That hyperlink isn’t there but.” Uber’s engineers at the moment are capped at US$1,500 a month in AI spend.
Walmart imposed comparable token rationing on its inside assistant after demand blew previous projections, Bloomberg reported. One thing is evident in that element. When tokens exceed the funds, they get capped. When folks exceed funds, they get severance.
The admission
No firm has travelled the complete circle extra publicly than Klarna. The fintech big changed roughly 700 customer support roles with an OpenAI-powered assistant, froze human hiring for greater than a 12 months, and made its AI-first mannequin a part of its pitch to public market buyers.
Then buyer satisfaction fell, complaints rose, and chief government Sebastian Siemiatkowski went on Bloomberg to say what few executives have mentioned aloud: “We centered an excessive amount of on effectivity and value. The outcome was decrease high quality, and that’s not sustainable.” Klarna is hiring people once more, and its CEO now argues that investing within the high quality of human assist is the firm’s future.
Gartner expects the Klarna sample to generalise, predicting that by 2027, half of the businesses that lower customer support employees for AI will rehire, typically beneath new job titles. Its separate survey of 321 customer support leaders discovered solely 20% had genuinely diminished staffing due to AI within the first place, which suggests a lot of the reducing was strange value self-discipline sporting an AI costume.
OpenAI’s Sam Altman has acknowledged as a lot, conceding some “AI washing” in company layoff bulletins, and enterprise capitalist Marc Andreessen, co-founder of Andreessen Horowitz, calls AI the “silver bullet excuse”. The displacement narrative, in different phrases, is partly theatre. The funds shift beneath it’s actual, and so is the injury.
Who absorbs the experiment?
The verified hurt lands on the folks least capable of take up it. Stanford HAI’s 2026 AI Index discovered that employment for software program builders aged 22 to 25 fell almost 20% from 2024, at the same time as older cohorts saved rising. Corporations are successfully eradicating the underside rung of the ladder whereas nonetheless anticipating senior engineers, those directing all these tokens, to exist in 5 years.
The worldwide arithmetic is harsher nonetheless. Huang’s thought experiment assumes a US$500,000 engineer, a compensation bracket that covers maybe 2 to five% of American software program engineers and vanishingly few anyplace else. Apply his half-salary token ratio to a typical engineer in Kuala Lumpur, Manila or Jakarta, and the token funds prices greater than the individual.
Within the markets the place a lot of the world’s software program work and buyer assist really occurs, the trade-off he describes doesn’t amplify staff a lot as worth them in opposition to a machine, utilizing ratios set in Santa Clara.
What Klarna discovered at the price of 700 jobs and a dented model is roughly what Gartner’s information exhibits in combination: the returns comply with corporations that spend on individuals who use AI, not on AI that replaces folks. The CFOs now capping token budgets after burning via them in 1 / 4 have began to rediscover one thing the trade spent two years unlearning. Expertise was by no means the road merchandise holding the enterprise again.
(Picture by kate.sade)
See additionally: Per-token AI expenses come to GitHub Copilot
Wish to be taught extra about AI and large information from trade leaders? Take a look at AI & Big Data Expo going down in Amsterdam, California, and London. The excellent occasion is a part of TechEx and is co-located with different main expertise occasions together with the Cyber Security & Cloud Expo. Click on here for extra info.
AI Information is powered by TechForge Media. Discover different upcoming enterprise expertise occasions and webinars here.
