Real at Silicon. Fake at Salary.

The $725 Billion Cover Story

In February, the man with the most to gain from AI hype told a New Delhi conference that some of the layoffs being blamed on his industry weren’t really about AI at all. Sam Altman called it “AI washing” — companies pinning unrelated headcount cuts on the technology because the explanation goes down easier with shareholders, the press, and the people being let go. It was a strange admission from the CEO of OpenAI, and an inconvenient one. Big Tech has just announced it will spend $725 billion on AI infrastructure this year, and 81,747 tech workers lost their jobs in the first quarter alone. The two numbers are being treated as cause and effect. They aren’t.

The confession nobody repeated

Altman’s quote barely registered outside trade press. It should have. When the most prominent AI evangelist on Earth concedes that an unspecified slice of the layoff narrative is invented, the reasonable response is to look at the slice. Independent analysis of every layoff announced between 2020 and 2026 finds that only around 16% are directly attributable to AI. The rest trace to overhiring during the 2021–22 boom, post-pandemic discipline, M&A integration, and rate-cycle cost cuts that would have happened with or without GPT-5. AI is the cover story; the spreadsheet underneath is older than that.

Bloomberg has the receipts

Here is what the data actually shows. Bloomberg reports that roughly half of AI-attributed layoffs will result in the same role being rehired offshore or at a lower salary. Forrester finds 55% of employers now regret cuts made in the name of AI, and predicts most will quietly refill those positions — but not where they cut them, and not at the wage they paid. Gartner expects half of AI-driven customer-service layoffs to be reversed by 2027, with the work returning under different titles.

A Blind survey of 2,392 US and India professionals found 52% expect their employer to increase India hiring this year; 38% explicitly say it is replacing US-based roles, not adding to them. An Amazon engineer told the platform: “1/3 of my team was laid off recently and they opened the same number of positions in India.” None of that is automation. It is geography.

Real at the Silicon layer

The bullish case for AI capex deserves an honest hearing, because it is partly correct. Google Cloud grew 63% year-over-year last quarter; AWS posted its fastest growth in fifteen quarters at 28%. Alphabet’s enterprise cloud backlog now stands at $462 billion. Wall Street sees the sector heading past $1 trillion in capex by 2027. The compute is real, the demand is real, and someone is paying for it.

But notice what is converting to revenue: cloud contracts, GPU rentals, infrastructure. Not “AI replacing the engineer who used to handle ticket triage.” The capex is genuinely productive at the silicon layer. The labor story is a different story altogether — and stapling them together in the same earnings call makes the 10-K read better than “we found cheaper humans elsewhere.”

The older migration

What is actually happening is the same structural shift the BPO industry has been navigating for two decades, now playing out one tier up the org chart. Mid-level engineering, customer success, finance ops, parts of marketing — work that was always location-flexible is finally being treated that way. The Philippines’ IT-BPM sector is forecast to grow from $40 billion in 2025 to $59 billion by 2028, with call center and customer service roles the single most-applied-to job category in Q1 2026. India’s tech hubs are absorbing roles cut in Seattle and London, often within weeks.

This is the structural story underneath the AI story — visible in the data for several years, just less convenient to discuss than robots. The honest framing is not “AI took your job.” It is “your job moved” — to a different city, sometimes a different continent, often at a different price. That has costs. It also has beneficiaries. Both deserve naming, and naming the beneficiaries is what most coverage avoids.

The 2027 test

The interesting test of this thesis arrives next year, when Gartner expects the rehiring quietly to begin. If those roles come back at original salaries in original locations, the AI-replacement story holds. If they reappear with different titles, in different time zones, at lower wages — the cover story falls. Bloomberg, Forrester, and Gartner are all betting on the second outcome. So is every survey of the workers themselves. The capex is real, the productivity gains are coming, and somewhere a GPU is doing useful work. None of it is why the engineer at desk 14 was let go.

The question for your business

Are you cutting headcount because AI replaced the work, or because someone else can do it cheaper?

Read more thought leadership articles here:

• • Who Trains the Seniors of 2031?
  • From Sneakers to GPUs — AI’s New Player