While Everyone Panics About AI Layoffs, Some Companies Are Hiring

Whoop announced it's hiring 600 people—nearly doubling its 800-person workforce—and CEO Will Ahmed framed it as the strategic bet of 2026. "Right now, companies are debating whether to hire more people or just invest in AI," he said. "And we are doing both."

This feels... wrong, doesn't it? Every headline right now is about displacement, about how many fewer people companies need. And here's Whoop going the opposite direction, hard.

Strategy consultant Nate B Jones argues in a new video that the companies aggressively hiring—not cutting—might actually understand what's happening better than the doom-scrollers. His thesis: we're asking the wrong question about AI entirely.

The prevailing question is: "How many fewer people do we need?" You can model this in a spreadsheet. It feels sophisticated. CFOs love it.

Jones's alternative: "Given that execution cost just dropped by an order of magnitude, what can we do right now that was previously impossible?"

These aren't variations on the same theme. They're fundamentally different orientations toward what's happening.

The Fixed Pie vs. The Expanding Pie

The cost-cutting frame assumes a fixed amount of value in the world and optimizes for capturing your slice more efficiently. The expansion frame assumes the total opportunity was artificially constrained by execution costs—and removing that constraint creates more value than all the savings combined.

History suggests the expansion frame wins. When steel got cheap, industries didn't just make existing products more efficiently—they built skyscrapers, railroads, cars. When computing got cheap, we didn't just do accounting faster—we got personal computing, the internet, mobile, cloud. When distribution got cheap, media companies that played defense died while new categories exploded.

The pattern has a name: Jevons Paradox. When efficiency dramatically increases, consumption goes up, not down, because cheaper resources make entirely new applications viable.

Jones points out something genuinely surprising: there are more software engineering jobs now than there were a year ago. If AI were purely displacing work, you'd expect the opposite.

His argument isn't that displacement will never happen—it's that we're obsessing over the wrong half of the equation. "The companies that break out of the doom frame first aren't just going to get a head start," he says. "They're going to get the whole race because everybody else is still arguing about headcount while customers move to the companies that dream bigger."

Six Structural Shifts Already Happening

Jones outlines six changes that don't require any future breakthroughs—they're available right now:

Speed as strategy. When you can compress product iteration from months to days, you're not doing the same strategy faster. You're doing different strategy. Right now, a product bet costs a quarter minimum, so most companies rationally copy competitors rather than explore. But if you can run 200 learning cycles a year instead of four? "The bottleneck shifts from 'can we build it?' to 'should we build it?'—and that's a human question."

Cursor's recent cloud agents update lets developers spin up 20 parallel agents working simultaneously on different branches. About a third of Cursor's pull requests are now written autonomously by agents. The tech exists. The constraint isn't technical—it's whether organizations can handle operating that fast.

Domain experts become builders. There are roughly 40 million software developers globally. There are hundreds of millions of domain experts—doctors who know what software their patients need, logistics managers who can draw the warehouse algorithm on a whiteboard, teachers who understand exactly what adaptive learning looks like.

All of them have been blocked by the translation layer: the gap between knowing what should exist and making it exist as software. That translation has been slow, expensive, and lossy.

"That translation layer is going away," Jones argues. "It's gone." When a doctor can describe what she needs and an AI agent can build it in an afternoon, you're unlocking an entirely new class of builder. Platforms like Lovable, Bolt, and Replit are already putting production-quality development in non-coders' hands.

The scale of this is hard to overstate. We might go from tens of millions of builders to hundreds of millions.

Quality becomes baseline. Most software has been mediocre not because engineers are bad, but because execution capacity has been scarce. Great testing, documentation, security review, performance optimization, accessibility, visual polish—all on time, all under budget? That's been impossible for most teams.

Now agents can handle testing, security review, and documentation as standard procedure rather than expensive add-ons. The baseline quality of all software goes up. "For so long, the gap between the top 5% of engineering teams and everyone else has been polish," Jones notes. "Not anymore."

This pushes differentiation toward product vision and customer experience—human questions.

Everyone's a platform now. Building and maintaining integrations has been nightmarish. But when agents can access systems anyway—through APIs or browser automation—every company becomes a platform whether they plan for it or not. The question shifts from "should we become a platform?" to "is our platform sticky and valuable?"

This also means platform strategy thinking, which typically lives at the VP level, needs to cascade down. If individual contributors can roll out integrations in an afternoon, they probably need to understand corporate strategy better.

The CFO math flips. Companies currently walk away from $10 million markets because the engineering team costs $3 million a year. They skip R&D projects with 20% success odds because failure costs two quarters of roadmap.

When execution costs drop 10x or 100x, all these calculations flip. The $10 million market becomes viable. You can run five experiments instead of one. "CFOs need to change their mindset," Jones argues. Capturing expanded opportunities requires people—different people doing different work, but people nonetheless.

Organizations move at insight speed. Not just execution speed—insight speed. When you get reliable customer insight, the default should be getting it into code immediately, not writing documentation and escalating through layers. This requires structural changes to how companies operate, and it's deeply uncomfortable for people who've been told code is dangerous.

What's Actually Hard Here

None of these shifts require technical breakthroughs. They're all available today. The hard part is entirely human.

"The hardest work ahead isn't technical," Jones says. "It's figuring out what upskilling looks like when the job isn't 'do the same thing faster' but 'do something you've never been asked to do before.'"

That's a radically different challenge. You can't train people to work faster at their current job when their current job might not exist in the same form—or might exist at 10x the scale with completely different constraints.

Jones's argument is that we're living through a constraint shift, not just an efficiency improvement. We weren't limited by ideas or ambition—we were limited by the cost of turning ideas into products. Remove that bottleneck and the constraint becomes our capacity for good ideas, deep domain knowledge, customer empathy, creative vision.

Those are all human capacities, and they're in short supply.

Think about how much bad software people currently tolerate. Think about how many niches can't be served because building for them costs too much. Personalized education. Clinical decision support for individual patients. Financial planning for the two billion adults with bank accounts but no adviser.

These aren't technical problems—they're economic problems. The cost of building has been too high.

Maybe not anymore.

The companies cutting headcount aren't wrong that AI increases efficiency. They're just making a bet about what happens next—that the pie stays the same size, and their job is to capture their slice with fewer people.

The companies hiring are making the opposite bet: that dramatically cheaper execution unlocks so much latent demand that the constraint becomes finding people who can see opportunities and act on them.

Only one of these bets accounts for what happened every other time execution costs dropped by an order of magnitude.

Which one you're making might matter more than whether you're using AI at all.

— Zara Chen