The Vertical Wall Series — New to the series? Start with What Is the Wall? — then follow the remaining parts in order. View all five pieces →
By Tango
In 2023, a group of researchers sat down and made a bet. Not a casual "I think AI will be big" bet — a detailed, month-by-month prediction of how artificial intelligence would develop through 2027. How good it would get at writing software. How much money the companies building it would make. How fast it would spread into real businesses.
Three years later, an independent review checked their homework. The verdict: reality is tracking at about 65% of their most aggressive predictions. That might sound like they overshot. But think about what that actually means — even the slow version of their forecast has AI rewriting entire industries within two years. And in some areas, like revenue and coding ability, the real world is actually ahead of what they predicted.
When a three-year-old roadmap is still mostly right, that's not a lucky guess. That's a signal that the window to act is wide open — measured in product releases, not years.
The Trillion-Dollar Proof
If you want to know whether the wall is real, don't ask a technologist. Ask the stock market.
In late January 2026, Anthropic released Claude Cowork — digital tools that do the work of financial analysts, legal researchers, and data scientists for $20/month. The companies selling those services for millions watched their stock prices collapse. FactSet fell 10.5%. Thomson Reuters dropped 18% in a single day. RELX lost 14%. The London Stock Exchange Group fell 8%.
Three days later, Anthropic released an upgraded AI that could manage teams of other AIs. Cumulative damage: roughly one trillion dollars in stock market value, gone.
This wasn't speculation. These were real products doing real work at prices that made entire categories of professional services obsolete overnight. The trillion dollars didn't disappear — it moved. To whoever figures out what comes next.
When Companies Fire Workers at Peak Profit
Block — the company behind Cash App and Square — cut roughly 4,000 jobs (about half its workforce) on February 26, 2026. Quarterly gross profit was up 24%. The reason: AI could do what those people had been doing. Wall Street's reaction: the stock jumped 25%.
This wasn't a recession-driven cut. It was a company at peak health announcing, in the plainest terms, that it had found a better engine. The same dynamic that makes this disruptive also makes it generative: capability that used to cost hundreds of millions now costs almost nothing.
The Intelligence Explosion Isn't a Metaphor Anymore
AI systems now design, write, test, and deploy their own improvements. This self-reinforcing cycle — where each improvement makes the next improvement easier and faster — is the heart of why the wall is vertical. It's compound interest, but for intelligence. The interest rate itself keeps climbing.
The systems don't need to be perfect to be transformative. They just need to be good enough and relentlessly cheaper. And every month, they're both.
What Survives — and What Gets Built
The old division of labour is collapsing. But collapses make room. The roles that matter now aren't just survival plays — they're the seats at the table where new industries get defined.
The guardrail builders. Someone has to decide what AI systems are and aren't allowed to do. That's not a constraint role — it's a power role. The people drawing those lines are writing the rules of the next economy.
The translators. Experts who understand a specific field well enough to tell AI systems what "good" looks like in that context. A great translator in oncology or contract law or supply chain risk isn't being replaced — they're being multiplied.
The capital decision-makers. Leaders who can redirect budgets from human teams into AI-powered operations and show clear results. They're not cutting costs — they're reallocating to create capabilities that didn't previously exist.
At Davos in January 2026, Elon Musk — the person arguably most associated with building this future — put it simply: "I'm very optimistic about the future. I think we're headed for a future of amazing abundance, which is very cool." That's not a PR line. That's the read from someone who's watched the models train in real time.
What to Do About It
This is where the data points somewhere specific.
Become the supervisor, not the laborer. The highest-leverage position in any field right now is the person who knows how to direct AI systems toward good outcomes. Think air traffic controller, not pilot. The controller doesn't fly the plane — they orchestrate a system that couldn't function without their judgment. That role isn't being automated. It's being elevated.
Treat AI as infrastructure, not tooling. Your security posture, purchasing decisions, and strategic planning should treat the major AI labs the same way previous generations treated the internet or the cloud — as foundational to everything else. The companies that integrated cloud early didn't just save money. They built things that were previously impossible. That's the game right now.
Plan in sixty-day windows, and use them aggressively. Align your planning cycles to the release cadence of the major AI labs. Not because longer-term thinking doesn't matter — but because the sixty-day window is where action is available. Every release is an invitation. The companies gaining ground right now are treating it that way.
As Musk said in the same Davos session: "I would encourage everyone to be optimistic and excited about the future good." The data supports it.
The Wall Is Behind Us — and So Is the Dread
We crossed the wall. That's true. The acceleration is no longer driven by human effort alone — it's driven by systems that improve themselves faster than we can track.
But here's what the dread narrative misses: every major technological inflection point in history looked like this from the inside. Electricity. Containerization. The internet. The people who won weren't the ones who waited for the dust to settle. They were the ones who recognized the shift early and moved toward it.
The trillion dollars that left FactSet and Thomson Reuters didn't vanish. It's waiting to flow into whatever gets built next. The researchers who made that 2023 forecast didn't predict collapse — they predicted transformation. The data says they're right.
The evidence isn't ambiguous. Neither is the opportunity.