The AI Jobs Paradox: OpenAI Doubles Headcount While Industry Lays Off 150,000

Last Updated on April 5, 2026 by Eytan Bijaoui

I need you to hold two facts in your head at the same time.

Fact one: over 150,000 tech workers have been laid off in 2026 so far. At least 20% of those layoffs explicitly cited AI as the reason. Block cut 40% of its people. Oracle is gutting 30,000 positions. Managers across every industry are running the math on what a $30-per-month AI subscription can do versus what a $90,000-per-year employee does, and they’re reaching the same conclusion.

Fact two: OpenAI, the company that built the tools everyone is using to justify those cuts, just announced it’s hiring 3,500 more people. Nearly doubling its workforce from 4,500 to 8,000 by December. That’s 12 new employees every single day for the rest of the year.

The company most responsible for eliminating jobs is on the biggest hiring spree in AI history. And nobody seems to find this weird.

The Redistribution Nobody Wants to Talk About

Let me be clear about what’s happening because I think most people are misreading the situation.

AI is not destroying jobs. AI is redistributing them. Upward, inward, and out of reach for most people.

The jobs disappearing are the ones that involve executing defined tasks: writing standard code, handling tier-one support tickets, generating reports from templates, processing data. These were traditionally entry-level and mid-level positions. The starting rungs of every tech career ladder.

The jobs being created? Research scientists. ML infrastructure engineers. “Technical ambassadors” (that’s a real title OpenAI is hiring for). People who can bridge the gap between what AI models can theoretically do and what specific industries actually need done. These are senior roles that require years of specialized experience.

So the labor market isn’t shrinking. It’s polarizing. The top is expanding. The middle is compressing. And the bottom is disappearing.

Sam Altman said something interesting about this in February. He told Fortune that some companies are “AI washing” their layoffs, blaming AI for cuts they would have made anyway. And he’s probably right about some of them. But then he added that real displacement is also happening, and that he expects it to accelerate. Coming from the CEO of the company building the displacement engine, that’s about as close to a confession as you’re going to get.

The Junior Developer Crisis That Should Terrify You

This is where the story gets genuinely concerning, and where I think most founders aren’t paying attention.

Junior developer hiring has collapsed. Not declined. Collapsed. Job postings labeled “entry-level software engineer” grew 47% between late 2023 and late 2024. But actual hiring into those roles? Down 73%. Companies are posting junior roles and then quietly filling them with experienced engineers who can ship with AI from day one.

The share of juniors and graduates in IT employment has dropped from roughly 15% to 7% in three years. That’s half the pipeline gone.

And I get the logic. I really do. If you’re a startup with 18 months of runway, and you’re choosing between a junior developer who needs 6-12 months of ramp-up and an AI coding tool that costs $30 a month, the spreadsheet makes the choice for you. In the short term, it’s the right call for survival.

But nobody is thinking about what happens in 2029.

Because junior developers aren’t just cheap labor filling tickets. They’re the future mid-level engineers. The future senior architects. The future CTOs. Every experienced developer you’ve ever worked with was once a junior who got mentored, who broke production, who learned from debugging something stupid at 2 AM, and who eventually became the person you can’t live without.

When you stop hiring juniors, you’re not saving money. You’re eating your seed corn. And the entire industry is doing it simultaneously.

The Paradox Gets Worse

OpenAI isn’t the only company on an AI hiring spree. Google DeepMind is expanding. Anthropic is growing. xAI just merged with SpaceX and is building a $25 billion chip fabrication facility in Austin that will need thousands of specialized engineers.

But these companies aren’t hiring the people who just got laid off. They’re hiring a completely different talent pool. Researchers with PhDs. Infrastructure engineers with a decade of experience in distributed systems. People who understand reinforcement learning from human feedback at a mathematical level.

The 150,000 people who lost their jobs at Block, Oracle, and dozens of other companies? Most of them don’t qualify for the 3,500 openings at OpenAI. The skills gap isn’t a crack. It’s a canyon.

And it keeps widening. Because the companies building AI need more senior talent. The companies using AI need fewer junior and mid-level employees. And nobody is training the next generation because the apprenticeship model that created experienced engineers for the past 40 years just broke.

This is a market failure happening in slow motion. Everyone can see it. Nobody is fixing it.

What This Means If You’re Building a Startup

I talk to pre-seed founders almost every day. And increasingly, the conversation around hiring has shifted from “who should I hire?” to “should I hire at all?” Which, as we’ve been tracking with the rise of solo founders building million-dollar companies, is a reasonable question right now.

But there’s a nuance that gets lost in the “AI replaces everything” excitement.

The talent you need most is getting more expensive, not cheaper. Yes, AI tools can replace the work that junior developers used to do. But the senior engineers, the domain experts, the people who can architect systems and make judgment calls about what to build? Those people are in higher demand than ever. OpenAI isn’t the only one competing for them. Every company that cut its junior pipeline is now fighting over the same shrinking pool of experienced talent.

A year from now, the median salary for a senior ML engineer is going to be higher than it is today. Because the supply is constrained and every company on earth suddenly needs them.

Your hiring strategy needs a 3-year horizon, not a 3-month one. If you’re a founder thinking about your team in 2029, you should be worried. The mid-level developers you’ll need don’t exist yet because nobody is training them. The companies that quietly maintained junior programs, that invested in mentorship and ramp-up time even when the spreadsheet said don’t, will have a structural advantage in 3-5 years that money can’t buy.

I’m not saying every pre-seed startup should hire juniors right now. That’s not realistic with limited runway. But if you’re planning to scale, think carefully about where your senior talent will come from in a world where the entire industry stopped growing it.

The real competitive moat might be people, not technology. This sounds counterintuitive in 2026. But think about it. When every startup has access to the same AI tools, the same APIs, the same foundation models, what differentiates them? Not the technology. The technology is increasingly a commodity. The differentiator is the judgment, the domain expertise, and the institutional knowledge that lives in experienced people’s heads.

That’s why Bezos is spending $100 billion to buy old factories and the expertise embedded in them. He understands that in the AI era, the scarcest resource isn’t compute or capital. It’s human knowledge that doesn’t exist in any training dataset.

The Altman Confession, Revisited

I keep coming back to what Altman said about AI washing. He’s acknowledging, publicly, that some companies are using AI as an excuse to cut costs they wanted to cut anyway. That post-pandemic over-hiring is being laundered through an AI narrative.

And that’s probably true. But there’s a darker reading of his statement that I think deserves attention.

If you’re the CEO of a company whose entire business model depends on companies adopting your AI tools, you have an incentive to say that the displacement is overblown. Because the alternative narrative, that your product is genuinely eliminating hundreds of thousands of jobs, is politically and economically dangerous for your business.

Altman is in an impossible position. He needs companies to adopt AI aggressively (that’s how OpenAI makes money). But he also needs the public narrative to not turn against AI (that’s how OpenAI avoids regulation). So he threads the needle: some of it is real, some of it is washing, the future will create new jobs.

Maybe he’s right. I genuinely don’t know. But I do know that the 73% collapse in junior hiring is real, the 150,000 layoffs are real, and the skills gap between jobs eliminated and jobs created is real. And somebody needs to figure out who trains the next generation of engineers, because the current answer appears to be “nobody.”

The Question Nobody in This Industry is Asking

Here’s what I’d love to hear a single VC, a single CEO, a single accelerator leader address at their next conference.

If we stop hiring juniors across the entire industry, if we replace the apprenticeship pipeline with AI coding tools, if we eliminate the bottom rungs of the career ladder because the spreadsheet says it’s efficient, where do senior engineers come from in 2030?

Because they don’t materialize from thin air. They’re built, slowly, through years of mentorship, mistakes, and accumulated judgment. And right now, we’re systematically dismantling the system that builds them.

OpenAI can hire 3,500 people because there’s a generation of engineers who got trained the old way, before AI coding tools existed. They got hired as juniors, they struggled, they learned, and they became the senior talent that’s now worth its weight in gold.

Who replaces them? What happens when the last generation trained through traditional apprenticeship retires, and there’s nobody behind them because we decided $30 a month was a better investment than mentoring a 22-year-old?

That’s not a workforce question. That’s a civilization question. And the company that just hired 4,000 people doesn’t have an answer for it.

Nobody does. Yet.