Since our earlier reporting established that AI is simultaneously disrupting hiring and transforming healthcare, the financial case for the technology itself has taken a serious hit — with hard data now undermining the trillion-dollar spending narrative that Wall Street and Silicon Valley have been selling for two years.

The Savings Were a Projection. Not a Result.

Bain & Company surveyed 951 companies with more than $100 million in revenue across nine sectors in April 2026. The findings, shared exclusively with Bloomberg News, are damning.

Among companies actually measuring their AI cost savings, 40% saw reductions of 10% or less. That's it. A single-digit-to-low-double-digit cost trim, at best, on technology that cost far more to implement than that.

44% of large companies are funding their NEXT wave of AI investment based on the savings from the last round — savings that haven't actually shown up yet. Bain called it a "circular bet with a structural leak." That's consultant-speak for a house of cards.

"The prior wave underdelivered. The savings pool is smaller than assumed," Bain warned. "And the investment case for the current wave was sized against projections rather than actuals."

Microsoft Couldn't Even Afford Its Own AI Tools

The most concrete example of this cost crisis came from Microsoft itself, reported by Fortune on May 22, 2026.

Microsoft began canceling most of its direct Claude Code licenses — the AI coding tool made by Anthropic — after its own engineers adopted it so aggressively that the costs became unsustainable. This is six months after Microsoft actively encouraged thousands of developers, project managers, and designers to experiment with the tool.

It worked too well. The scale of usage made the economics impossible.

Microsoft is now steering engineers toward GitHub Copilot CLI instead. Amazon's investment relationship with Anthropic — Amazon has committed up to $4 billion in Anthropic — remains intact. But internally, Microsoft discovered that AI tools can cost MORE than the human labor they're supposed to augment.

Microsoft isn't alone. According to Fortune, Uber's CTO Praveen Neppalli Naga told The Information in April that Uber burned through its entire 2026 AI coding tools budget in just four months. Four months. Uber had even built internal leaderboards to incentivize AI tool adoption. It worked — and bankrupted the budget.

16,000 Jobs a Month Gone. Gen Z Paying the Price.

While companies struggle to justify the costs, the human toll is already accumulating. Goldman Sachs economist Elsie Peng, writing in a Goldman Sachs U.S. Daily note dated April 6, 2026, calculated that AI is erasing roughly 16,000 net U.S. jobs per month — the result of AI substitution eliminating about 25,000 jobs per month while augmentation creates back roughly 9,000.

Gen Z is absorbing the worst of it. In occupations most exposed to AI substitution — insurance claims clerks, bill collectors, entry-level data roles — the unemployment gap between workers under 30 and workers aged 31-50 has widened sharply compared to pre-pandemic baselines. The wage gap has deteriorated alongside it.

The Atlantic's reporting adds another layer: AI hasn't just replaced workers, it's broken the hiring process itself. Candidates are using AI to write resumes and ace interviews. Companies are receiving thousands of AI-polished applications that all look identical. Ken Schumacher, a tech executive who witnessed this firsthand, described it as "really, really hard for anyone to figure out who's real and who's fake." He now runs a startup using AI to detect AI cheating by job candidates. The ouroboros is complete.

Columbia Business School researchers described the phenomenon as "signal collapse" — CVs used to carry real information. Now everyone looks the same and no one can trust any of it.

The Uncomfortable Math Nobody Wants to Say Out Loud

Companies are spending real money, eliminating real jobs, and not getting real savings.

The Goldman Sachs data on job losses is real. The Bain data on missing savings is real. The Microsoft and Uber cost overruns are real. These aren't AI skeptics making theoretical arguments — these are the largest companies in the world reporting actual results.

The media framing from both sides is missing this. Left-leaning outlets like The Atlantic focus heavily on the job disruption and human cost — fair, but they largely skip the fiscal irresponsibility angle. Right-leaning outlets celebrating the AI productivity story are ignoring that the productivity gains, by Bain's own survey, haven't shown up yet.

Bain identified the root cause: most companies' data is so disorganized that their AI can't access it. That's not a technology problem. That's a management failure that no amount of additional AI spending will fix.

What This Means for Regular People

If you're under 30 and job hunting right now, you're competing against AI-generated applications while companies use AI filters to screen them — and Goldman Sachs says you're statistically more likely to be unemployed because of this technology than your older colleagues.

If you're a taxpayer or investor watching trillions flow into AI infrastructure, Bain's survey says nearly half of that next spending wave is being justified by savings that have not materialized.

If you're a company executive who approved this spending based on rosy projections, Bain says the data should be making you uncomfortable.