The Displacement Nobody's Counting

There's a number floating around that says 55,000 to 70,000 people lost their jobs to AI in 2025.

That number is wrong. Not slightly wrong. Off by maybe 50x.

I'm going to show you why. And I'm going to show you where the real displacement is happening — in places that no government report, no BLS survey, and no cable news segment is looking.

The Canary

If you want to know what AI is actually doing to the labor market, don't look at corporate layoff announcements.

Don't look at unemployment filings. Don't look at any of the numbers that policymakers and journalists use to tell you everything is fine.

Look at Fiverr.

Fiverr and Upwork are real-time marketplaces. No hiring committees. No quarterly budget reviews. No corporate inertia. When AI gets good enough, the orders just stop.

Fiverr's active buyers: 4.2 million in 2022. 3.3 million by Q3 2025.

That's a 21% decline since ChatGPT launched.

Writing job demand on the platform: down 33%. Translation: down 19%, with hourly rates collapsed 20% or more. Software and web development posts: down 21%. Graphic design: down 17%.

Fiverr's stock is down 90% from its peak.

And here's the part that broke my mental model. I would have assumed the mediocre freelancers got hit first. The ones who were already borderline. The "you get what you pay for" tier.

Wrong.

The research — peer-reviewed, published in Organization Science — found the opposite. For every 1% increase in a freelancer's past earnings, they experienced a 1.7% decrease in monthly income after AI arrived.

The best performers got hit the hardest.

Think about what that means. The "I'm too good to be replaced" thesis didn't just fail. It failed first.

The premium that top freelancers charged was the first thing AI erased, because the gap between "good enough AI output" and "expensive human output" wasn't worth the price difference anymore.

The People Who Don't Exist

None of these freelancers filed an unemployment claim.

They don't appear in any Bureau of Labor Statistics report. They're independent contractors who simply stopped getting hired.

Their dashboards went quiet. Their invoices stopped. Their income dropped. And statistically, as far as the government is concerned, nothing happened.

This is the invisible displacement. And it's massive.

It's not just freelancers.

Graduates Walking Into a Wall

42.5% of recent college graduates are underemployed as of Q4 2025. That's the highest since 2020.

33% of 2025 graduates are straight-up unemployed. Compared to 20% in 2024.

Only 30% found jobs in their field.

56% say the AI and digital skills gap is the barrier.

This is Dario Amodei's Davos prediction — that 50% of entry-level roles would be affected — already materializing. Companies aren't hiring juniors because AI handles junior-level work.

Why pay a 22-year-old $65K to do research and write first drafts when Claude does it for $20 a month?

The 2025 and 2026 graduating classes are the first to face this structurally. Not as a tough job market. Not as "it takes a few months to find something." As a fundamental repricing of what entry-level knowledge work is worth.

And nobody's counting them as "displaced by AI." They're just... underemployed. A statistic in a different column.

The Hiring That Never Happens

This is the one that's hardest to see and probably the biggest number of all.

Planned hires for 2025: 507,647. Down 34% from 2024. The lowest since 2010.

These are jobs that were supposed to exist and don't. Positions that got budgeted and then frozen. Roles that would have been posted and never were.

Nobody got laid off. Nobody filed a claim. No announcement was made. The job just... didn't happen.

How many people are affected by jobs that were never created? You can't count them. They don't know they were supposed to be hired. There's no category for "the role that would have existed if AI hadn't made it unnecessary."

But the number is almost certainly larger than the layoff number.

The Workers Nobody Talks About

OK. Now the part that makes me angry.

Because everything I just described is happening in the US and getting zero attention. What's happening internationally is worse, and getting even less.

India

63,000 jobs cut at TCS, Infosys, and Wipro alone in 2024. IT services growth — the engine that powered India's middle class for two decades — is stalling.

India's IT sector is $200 billion in exports. 5.4 million workers. This isn't a niche industry. This is the economic backbone of an entire class of people.

And the threat isn't that Indian developers are bad. It's that AI is cheaper than any human at any price point in any country. The cost arbitrage that made offshoring work — "we'll do it for a quarter of the US price" — breaks when AI does it for a hundredth.

The Philippines

1.82 million BPO workers. Eight percent of the entire country's GDP.

The IMF says 5 million Filipino workers are at risk from AI. 36% of all jobs in the country are "highly exposed."

Filipino freelancer rates on global platforms have collapsed from $200 a week to near-zero per project as AI competition intensified.

Near-zero.

These are real people. In Manila. In Cebu. Watching their order dashboards go dark. They exist in no developed-country unemployment statistic. No American news outlet is covering them. No policymaker in Washington is thinking about them.

Eastern Europe

Poland, Romania, Ukraine — the nearshore tech hubs that grew on being cheaper than the US but closer than India. Their advantage was time zones and cultural proximity.

AI doesn't have a time zone. Contract rates are compressing. The nearshore model is eroding.

The Common Thread

Every one of these countries — India, Philippines, Poland, Romania — built their economic development on being the affordable alternative for knowledge work. That was the escalator. Study hard. Learn English. Get a job doing what Americans or Europeans do, for less money. Build a middle-class life.

AI just kicked the bottom rung off that ladder.

And these countries have no fiscal cushion. No UBI debates. No "Transition Economy Act." No safety net at all. They absorb the shock with zero policy tools.

What the Real Number Looks Like

So let's add it up.

Freelancers losing work globally — millions. Not thousands. Millions. On Fiverr, Upwork, and dozens of regional platforms. Income declining with no visibility in any official report.

Hiring freezes — 500,000+ positions in the US alone that were budgeted and never filled. Multiply globally.

Underemployed graduates — hundreds of thousands per year entering a job market that has structurally less demand for what they trained to do.

International knowledge workers — India, Philippines, Eastern Europe, Latin America — watching their rates collapse and their order flow dry up.

Discouraged workers leaving the labor force entirely — not unemployed, just... gone. Not looking anymore. Not counted.

The official "AI displacement" number is 55,000 to 78,000.

The real number — people experiencing meaningful income decline or job loss connected to AI — is conservatively 2 to 3 million globally. As of right now. February 2026.

The official statistics capture maybe 5-10% of what's actually happening.

Why This Matters

I'm not sharing this to be a doomer. I don't think the world is ending. I think the world is restructuring, and the restructuring is happening faster than the measurement systems can track.

That gap — between what's actually happening and what the numbers say is happening — is dangerous. Because policy follows data. Governments respond to statistics. If the statistics say everything's fine, then there's no urgency to act.

Meanwhile, a freelancer in Manila just watched her last client switch to ChatGPT.

A 23-year-old in Ohio just got his fourth rejection from an entry-level marketing role that doesn't exist anymore.

A senior developer in Bangalore just got told his team is being "restructured" — which means replaced by an AI tool his manager learned about at a conference three months ago.

None of them show up in the number that says 55,000.

I work in cross-border mobility. I sit across from founders and professionals navigating these systems every day. I watch the doors narrow in real time. I've seen people sell everything to move to a country for a job that disappeared before they got there.

This isn't theoretical to me.

And the gap between what's happening on the ground and what the official numbers reflect is the widest I've ever seen.

The Tremor

One more thing.

Everything I just described — the freelancer collapse, the graduate wipeout, the international displacement, the invisible hiring freezes — this is the tremor. Not the earthquake.

Enterprise AI adoption is at 88%. But only 39% of companies report any meaningful bottom-line impact. And of those, most say less than 5%.

That gap — 88% adoption, less than 5% impact — is where the next two to four years live. As companies move from "we have AI tools" to "we've restructured our workflows around AI," the impact on headcount scales.

The shallow cut is happening now. The deep cut hasn't started.

When it does, the numbers I just showed you — the ones that are already 50x higher than what's being reported — get much, much bigger.

The people being displaced today are the canary. They're telling us exactly what's coming.

Nobody's listening.