The Leverage Shift
When one person with AI tools can do what a team of ten once did.
The Leverage Shift
"Give me a lever long enough and a fulcrum on which to place it, and I shall move the world."
— Archimedes
Naval Ravikant said something a few years ago that I keep coming back to. He was talking about wealth creation, and he made a distinction that sounds simple until you think about it: there are two kinds of leverage. The old kind — labor and capital — requires permission. You need someone to hire you, or someone to fund you. The new kind — code and media — is permissionless. You can create it while you sleep. Nobody has to approve it.
When Naval said this, it was an interesting framework. Now it's a description of something that's actually happening at scale, and the speed at which it's happening is the part most people underestimate.
Let me describe what I mean by leverage in concrete terms, because the word gets used loosely.
Leverage, in the sense I'm using it, means: the ratio between what you put in and what you get out. A person with no leverage trades an hour for an hour's worth of output. A person with leverage trades an hour for something that continues to produce value after the hour is over.
For most of employment history, leverage belonged to organizations. A company could hire a hundred engineers, build a product, and sell it to a million customers. The company had leverage. The individual engineers had jobs.
What's changed — what's actually changed, not as a theory but as a measurable fact — is that the tools now available to individuals provide leverage that used to require organizations to generate.
Five years ago, building a working software application required, at minimum, a developer with several years of experience, or enough money to hire one. The process took months. Infrastructure decisions alone — hosting, authentication, payments, databases — required specialized knowledge that took years to acquire.
Today, I watch people who have never written a line of code describe what they want to an AI system and receive a working prototype. Not a mockup. A working application with authentication, a database, payment processing, and deployment — in an afternoon. The first time I saw this happen, I understood it intellectually. The tenth time, I understood it practically. By the hundredth time, I stopped being surprised and started doing math.
The math looks like this.
Gartner projected that by the end of 2025, 70% of new applications would be built using low-code or no-code platforms — and that projection was made before the current generation of AI coding tools existed. The low-code market itself grew from $29 billion to a projected $264 billion. Those numbers describe the formal tools. They don't account for the informal ones: the people building with Claude, with Cursor, with Lovable, with tools that didn't exist eighteen months ago and that are now part of how real products get shipped.
But the code is just one dimension. Consider what else compressed:
A person who wants to reach an audience used to need a publisher, a distributor, or a marketing budget. Now they need a keyboard and a perspective. The infrastructure for distribution — social platforms, email, search engines, podcasts — is free. The cost of reaching your first thousand people is zero dollars and some judgment about what's worth saying.
A person who wants to design a product used to need a designer, or years of training in design tools. AI design systems now generate professional-quality interfaces from text descriptions. The output isn't always perfect. But it's good enough to test, to iterate on, to put in front of real users and learn from their responses.
A person who wants to produce music used to need a studio, equipment worth tens of thousands of dollars, and years of technical skill. I've produced tracks using AI tools that would have required a professional studio session five years ago. The quality ceiling has risen while the cost floor has dropped to near zero.
Each of these compressions, individually, is significant. Together, they describe something structural: the minimum viable capability to create, distribute, and sell something valuable has dropped by an order of magnitude.
I want to be careful here, because there's a version of this argument that's dishonest. The version that says: anyone can build anything now, no skills required, AI does everything. That's not true, and pretending it is will waste people's time and money.
What AI gives you is not the ability to build something good. It gives you the ability to build something real — a working version of your idea — fast enough that you can discover whether it's good by showing it to the people it's meant to serve. The judgment about what to build, who to build it for, and whether the thing you built actually solves their problem — that's still entirely human. AI didn't automate judgment. It automated the execution gap between judgment and reality.
That distinction matters enormously. The people I've watched succeed with these tools are not the ones who delegated everything to AI and hoped for the best. They're the ones who had a clear idea about a specific problem, used AI to collapse the time between idea and prototype, and then iterated based on what they learned from real users. The AI was the lever. The judgment was the hand on the lever.
This is what Naval's framework looks like in practice. Permissionless leverage doesn't mean effortless output. It means you don't need anyone's approval to start. You don't need a co-founder to write the code. You don't need an investor to fund the first version. You don't need a publisher to reach your audience. The gates that used to stand between "I have an idea" and "people are using my product" have largely dissolved.
There's a historical parallel that's clarifying here, even if it's imperfect.
When the printing press arrived in Mainz in the 1450s, the immediate reaction was fear for the scribes. Scribes had spent years developing the skill of handwriting manuscripts — beautiful, accurate, painstaking work. The press made that skill commercially obsolete within a generation.
But here's what happened next, and it's the part that matters for this moment. The cities that adopted the printing press saw economic growth rates 60 percentage points higher than those that didn't, measured between 1450 and 1600. The reason wasn't just that books became cheaper. It was that the press created a new type of worker. Historians describe the emergence of "a new man" — someone "adept in handling machines and marketing products while editing texts, founding learned societies, and promoting artists." The press didn't just redistribute existing work. It created categories of work that hadn't existed before.
The pattern repeated with the steam engine, with electrification, with the personal computer, with the internet. Each wave displaced practitioners whose skills had been made abundant. Each wave created new forms of independent economic participation that were impossible before the wave arrived.
What's different about the current wave — and I think this difference is underappreciated — is the breadth of compression. The printing press compressed one thing: the production of written text. The internet compressed distribution. AI is compressing production across essentially every cognitive domain simultaneously. Writing, coding, design, research, analysis, music, video, legal reasoning, medical diagnosis — all at once.
That breadth means the leverage shift isn't concentrated in one industry. It's available to anyone who has a specific problem they understand well enough to build a solution for. The lever is universal. What varies is whether you pick it up.
The question I keep getting from people who sense this but haven't acted on it is: okay, but leverage to do what?
It's the right question. Leverage without direction is just capability without purpose. The tools exist. The barriers have dropped. But dropped barriers don't tell you which door to walk through.
The answer, as far as I can tell from watching the people who are actually building, is not a single answer. It's a pattern. And the pattern is more specific than "start a business" but less specific than "build exactly this thing." The people who are building successfully right now share something, and it's not a particular skill set or a particular industry. It's a particular relationship to a particular group of people.
That's where the next part of this starts.
Next: The Builders Already Building — what the people who moved early actually have in common.