Zero-Point Software
In physics, there’s this concept called zero-point energy, the idea that even in a perfect vacuum, even at absolute zero, there’s still energy permeating all of space. The quantum fields never fully settle down. There’s always something there, vibrating away in the background, an inexhaustible sea of energy woven into the fabric of reality itself. In science fiction, zero-point energy has become shorthand for the ultimate fantasy: unlimited, virtually free energy, extracted from the vacuum of space. It’s the power source of the Ancients in Stargate. It shows up in half the hard sci-fi novels you’ve ever read. The idea is intoxicating. What happens to civilization when energy becomes so abundant and so cheap that it’s essentially free? When the constraint that has shaped every human society since fire is simply… removed?
I’ve been thinking about this, because I think we’re approaching the software equivalent of that moment.
Software and the internet have always had this interesting property that Ben Thompson likes to write about, which is that software has zero marginal cost to distribute. If you can put enough capital into building the initial product, and have some minimal capability to maintain it and add features over time, distributing it to a thousand people is effectively the same as distributing it to a million. The incremental cost of the next user approaches zero in a way that no physical product can match. And this has been the source of enormous business leverage, particularly throughout the 2010s, as the SaaS and mobile eras took the early internet and put software in every conceivable aspect of our lives.
What we’re seeing now with AI is that software distribution is not the only zero marginal cost aspect of the software lifecycle. Increasingly, software *creation* is becoming zero marginal cost too.
What once took a team of engineers a year is now able to be done in an afternoon. I don’t mean a toy demo. I mean a functional, deployable product with real utility. Yes, there remain bottlenecks around code review, security, scalability, the kinds of things that separate a weekend hack from a production system. But if at this point you don’t believe that those bottlenecks are going to be substantially solved within six months, I don’t think you’re really paying attention. The trajectory is not subtle.
This is what I’m calling zero-point software: the approaching event horizon where the cost of creating software converges on zero. Like the sci-fi dream of zero-point energy, it’s not that the cost literally disappears. There will always be some irreducible minimum of human judgment, taste, and domain understanding required. But that minimum is plummeting, and fast. The distance between “idea” and “working product” is collapsing in a way that changes the fundamental economics of the entire industry.
What this amounts to is the death of the software moat.
This doesn’t mean the death of software. To the contrary, Jevons’ paradox certainly applies here. When something becomes radically cheaper, you don’t use less of it. You use dramatically more. Software will be more abundant and woven into more aspects of our lives than we can possibly comprehend at this point. Everyone will be able to create it. The nontechnical employee spinning up an internal tool. The most senior engineer who now has 100x capacity. The fourteen-year-old in Lagos who has an idea and a laptop.
But precisely because everyone can create it, software itself can no longer be a moat. In the past, the ability to ship software was a genuine competitive edge in many parts of the economy. It explains why Silicon Valley’s upstarts were able to extract blood from so many slower, old-economy incumbents. If you could recruit and retain great engineers, move fast, iterate, you had an advantage that was hard to replicate. The incumbents were too bureaucratic, too slow, too removed from the craft of building to compete.
That’s changed. Lines of code and engineering hours can no longer be a moat. Your ability to execute fast on software is no longer a competitive advantage. It’s table stakes.
This is going to have very interesting second-order implications.
Consider the classic startup David vs. Goliath narrative. What happens when the small, scrappy startup that was able to execute faster than the big incumbent is now faced with an incumbent that can move almost as fast as they can? When the incumbent’s army of mid-level product managers can each spin up and test a new product concept in an afternoon? When the Fortune 500 enterprise with the massive distribution footprint can also ship software at startup speed?
The traditional moats, distribution, brand, customer relationships, network effects, domain expertise, regulatory capture, vertical integration, these don’t go away. They get *amplified*. When everyone can build the product, the product is no longer the differentiator. Everything around the product becomes the differentiator. The things that were always important but secondary to “can you actually build the thing” become primary. And likely, new sources of moats will emerge that are hard to predict today. Taste, maybe. Curation. Trust. The ability to orchestrate AI systems rather than just deploy them. Community. Data gravity. Things we don’t have names for yet.
One of the most interesting implications of zero-point software is what it does to business models. If building software costs next to nothing, then charging for the software itself starts to look like the wrong move. Why would you monetize the thing that’s becoming free when you could give it away and monetize something else entirely?
We’re going to see an explosion of models where software is completely given away, not freemium, not free-with-ads, but genuinely free as a strategic choice, in order to capture value somewhere else in the stack. The software becomes the customer acquisition channel, and the real revenue engine is transactions, or data, or some adjacent service that the software enables.
This isn’t entirely new. Payments companies have been doing a version of this for years: give away the POS terminal, make money on the take rate. Robinhood gave away the trading app, monetized order flow. But what zero-point software does is generalize this pattern across the entire economy. When any business can spin up a full-featured software product for almost nothing, the rational play is often to treat that software as a loss leader (or more accurately, a no-cost leader) and build your margin into something stickier. Transactions. Financing. Insurance. Logistics. The physical world stuff that can’t be conjured from a prompt.
Think about what this means for SaaS. The $99/month subscription model that has defined software for the last fifteen years starts to look precarious when your competitor can offer an equivalent product for free and make their money on the 1.5% they clip on every transaction flowing through it. The entire SaaS pricing model was predicated on the assumption that building and maintaining the software was expensive enough to justify recurring fees. When that assumption breaks down, so does the model. What replaces it is something more like the payments or marketplace model, software as infrastructure for capturing transactional value.
What’s really happening here is that the software industry is going through a transformation so fundamental that calling it a “transformation” undersells it. The software industry is essentially disappearing, not because software is going away, but because it’s becoming the substrate that is common to all industries. It’s becoming the water, not the fish.
This is typically how industrial revolutions progress. Something that is once valuable as a breakthrough capability becomes so commoditized that it’s too cheap to meter. And when it becomes too cheap to meter, it becomes essentially invisible, and it becomes the backbone of everything. Electricity went through this. Before it was commoditized, “the electricity industry” was a thing, you could build a massive competitive advantage just by having access to it. And then everyone had access to it, and the advantage shifted to what you *did* with it. Nobody today starts a company and lists “we use electricity” as a competitive advantage on their pitch deck.
Software is on the same trajectory now. We’re watching it happen in real time, just compressed from decades into years.
The software industry is dead. Long live software.