In Q1 of 2026 alone, about 80,000 tech jobs were cut. Roughly half were explicitly attributed to AI replacing those roles. Oracle cut 20,000 to 30,000 people to free up capital for data centers. Jack Dorsey restructured his entire company around AI and wrote a memo about it. Less than 10% of people surveyed have a positive view of AI right now.

This is real. It’s happening faster and at a larger scale than most people expected. And if you’re a founder or operator, you’re probably doing the math on your own P&L — how much of my team could AI systems replace? How much overhead could I cut? How fast?

I want to give you a different frame for thinking about this. Because the layoffs are real, but the story doesn’t end where the headlines stop.

The Lump of Labor Fallacy

There’s an economic concept called the lump of labor fallacy. It’s the idea that there’s a fixed amount of work in the economy, and if machines start doing some of it, humans permanently lose out. Fixed pie. Machines take a slice. Less pie for everyone.

Economists have been refuting this for 200 years.

The cotton gin is the American textbook example. Eli Whitney invented it to reduce the need for slave labor in cotton processing. What actually happened: it made cotton processing so efficient that demand for raw cotton exploded, which meant more fields needed to be planted, which meant more labor was needed to pick the cotton. The slave economy expanded massively. The opposite of what was intended.

The same pattern played out with ATMs. When ATMs appeared in the 1970s, everyone assumed bank tellers were finished. Per-branch teller count did drop — from about 21 to 13. But because operating a branch became cheaper, banks opened more branches. Total bank teller employment went from around 300,000 in the 1970s to about 600,000 by the 2000s. The nature of the job changed — less cash counting, more relationship management and financial product cross-selling — but the total number of jobs grew.

Internet. Cars. Textiles. The pattern repeats across 250 years of technological disruption. The initial displacement is real and painful. The long-term result is more economic activity, more jobs, and jobs that look fundamentally different from what existed before.

Nobody 30 years ago would have predicted that “social media manager” or “digital ad specialist” would be a career. The same is true for whatever roles emerge on the other side of this transition.

Three Acts

I think the AI employment story plays out in three acts.

Act One is the painful one. Companies become AI-native. They discover they can produce the same output — or more — with a fraction of the headcount. Revenue per employee explodes. Look at Anthropic: roughly 5,000 employees, projected to hit $80 to $100 billion in revenue by end of 2026. When Google hit $30 billion in revenue, they had 32,000 employees. Salesforce had 79,000.

The first movers who go AI-native will outcompete everyone in their market. Better margins. Higher speed. More consistent quality. And they’ll do it with dramatically fewer people. This creates pressure across entire industries — competitors either follow suit or fall behind. The layoffs in Act One aren’t a one-time event. They compound as more companies make the transition.

If you’re an operator right now, the question isn’t whether the opportunity exists to reduce headcount. It obviously does. The question is whether you should, and what comes next if you do.

Act Two is when everyone catches up. The first movers’ secrets come out — because impressive results always get shared. Competitors in every industry start their own AI-native transformations. The advantage decays. The playing field levels.

The analogy I keep coming back to: Act One is bringing a gun to a knife fight. You don’t need cover. You just need clear sight. Act Two is bringing a gun to a gunfight. Now you need kevlar. You need positioning. You need a fundamentally different strategy because your opponent has the same capabilities you do.

Once every competitor is AI-native, the technology is no longer the differentiator. You can’t out-AI someone who has the same AI. The question becomes: where does the edge come from now?

Act Three is when humans become the edge again. This is where the lump of labor fallacy breaks. Once AI is universal, the differentiator has to come from what AI can’t provide. Original creativity. Aesthetic judgment. Taste. Relationship building. Trust-based sales. Strategic bets that require conviction under uncertainty. Accountability — you can’t sue an AI system, but someone has to be liable for outcomes.

Here’s the mechanical argument. AI amplifies every human on the team. A 12-person company with AI can produce the output of a 100-person company. But the amplification doesn’t cap at 12. A 24-person team with AI produces more than a 200-person traditional company. A 50-person team can match what a 1,000-person organization used to output.

Nobody knows yet whether this scales linearly, super-linearly, or with diminishing returns. But the principle holds: more humans working alongside AI systems means more output, more creativity, more strategic depth, and more competitive edge.

Human headcount becomes the scaling input again. The constraint flips from “we need more people to do the work” to “we need more people to direct, refine, and add judgment to the work the AI systems are doing.” More judgment calls per day. More relationships being built. More strategic bets being made. Human labor becomes the bottleneck again — just a different kind of labor, applied at a different point in the value chain.

The Timeline

I don’t have a confident timeline. There are too many political variables. Dylan Patel at Semi Analysis predicts massive protests against AI this year, and I think he’s probably right. Elected officials will run anti-AI campaigns because that’s how their constituents feel. Regulation could slow the transition significantly.

Best case, assuming no major political interference: the full cycle — from widespread layoffs through competitive leveling through a hiring boom — plays out in five to seven years. More realistically, probably closer to ten.

What I know for a fact: there will be a hiring boom on the other side of this. Not for the same jobs. Not in the same roles. But the demand for human talent, once everyone is operating on a level AI playing field, will be enormous. The constraint will be finding people who can work effectively alongside AI systems — who have the taste, the judgment, and the domain expertise to be the input that makes the AI output valuable.

What This Means If You’re Running a Business

The companies that go AI-native first capture an arbitrage. Better margins, higher output, faster execution. That advantage compounds quarterly. By the time competitors catch up, the first movers have years of operational learning, data accumulation, and system refinement that can’t be shortcut.

But “going AI-native” doesn’t mean firing your team and replacing them with chatbots. Clara tried that with customer support — optimized for speed of response instead of customer satisfaction. Refund rates spiked. They’re hiring people back.

The right approach is surgical. Identify where AI creates genuine leverage — the bottlenecks where throughput is limited by human capacity, where the logic is proven, where the ROI math is clear. Build systems that execute that logic with higher consistency and throughput. Then watch where the next bottleneck surfaces, because it always does.

Every constraint you remove creates a new one downstream. You speed up your sales pipeline, now your onboarding team is drowning. You automate onboarding, now your service delivery team is the bottleneck. This is the game — and it’s the game that eventually requires more humans, not fewer, to play at the highest level.

The layoffs are coming. They’re already here. But the story doesn’t end with layoffs. It ends with a different kind of company, a different kind of workforce, and — if history is any guide — more economic activity than we started with.

The faster we get there, the better. Which is why we’re trying to help as many companies as possible make the transition now, instead of waiting for the pain to force it.