A significant shift is happening to the concept of software as intellectual property. We're watching, in real-time, as the thing that once defined value in our industry—the unique, proprietary code—becomes something closer to a commodity. This isn't just a technical change; it's a strategic one for anyone building software companies.
For decades, we've operated under a simple assumption: code is valuable because it's hard to create. A startup's codebase was its moat, its treasure, the thing acquirers paid premiums for. But what happens when that assumption crumbles? When a small team with AI tools can replicate years of engineering work in weeks?
We're not just facing a productivity revolution. We're facing a fundamental question about what software companies actually are.
The Great Inversion
The traditional software startup was, at its core, a machine for converting venture capital into code. You raised money, you hired engineers, they wrote software, and that software became your asset. The capital allocation was straightforward: most of it went to salaries.
Today, we're witnessing an inversion. A growing portion of venture funding doesn't go to human salaries—it goes to API calls, model training, and compute time. We're funding silicon, not carbon. Every startup is becoming, in some sense, a reseller of intelligence from a handful of foundation model providers.
This creates a strange new dynamic. VCs, once the kingmakers of Silicon Valley, are increasingly becoming middlemen, funneling capital from LPs to a few dominant AI platforms. The real power is consolidating not with those who fund innovation, but with those who provide the tools for it.
The Paradox of Infinite Leverage
Here's the paradox: as it becomes easier to build software, software itself becomes less defensible. The same tools that give a team of three the power of thirty also mean that any competitor can match your features in a fraction of the time it took you to build them.
We're entering an era of infinite leverage but zero moats. The traditional defensibility of software—the accumulated complexity that made it hard to replicate—is evaporating. What remains?
The answer, I think, lies not in what you build but in how you build it. The new competitive advantage isn't your codebase; it's your ability to continuously rebuild and evolve faster than others can copy. It's not about the artifact but about the process that creates it.
The Velocity Game
In this new world, the only sustainable advantage is speed. Not the speed of writing code—AI handles that—but the speed of understanding what to build, for whom, and why. The speed of iterating based on real user feedback. The speed of pivoting when the market shifts.
This changes what we should value in startups. The interesting metric isn't lines of code or even traditional product-market fit. It's something more like "market-response velocity"—how quickly can you sense a need and deliver a solution? How fast can you incorporate feedback and ship improvements?
Teams that can go from idea to deployed feature in hours, not weeks, will dominate. But this isn't just about using AI tools well. It's about building organizations designed for this new tempo. Flat hierarchies, autonomous teams, and decision-making pushed to the edges become not just nice-to-haves but existential necessities.
The New Primitives
As software creation becomes commoditized, what becomes scarce and valuable? I see a few emerging primitives:
Distribution remains king. Building features is easy; getting them in front of users is not. The startups that win will be those that crack distribution in their specific domains.
Taste becomes crucial. When anyone can build anything, the question becomes: what's worth building? Product sense, design taste, and deep user empathy become the differentiators.
Data flywheels accelerate. While code is replicable, your specific dataset—especially if it's continuously improving based on user interactions—is not. The feedback loops you create matter more than the features you ship.
Operational excellence scales. The ability to run reliable, secure, compliant services at scale remains hard. The boring stuff—the stuff AI doesn't magically solve—becomes the moat.
Where We're Heading
I suspect we're moving toward a future where software startups look more like media companies than traditional tech companies. The value isn't in the content (code) itself but in the audience (users), the brand (trust), and the velocity of creation.
We might also see new funding models emerge. Why should VCs be the intermediaries when the AI platforms have perfect information about which startups are gaining traction? Imagine getting an automated funding offer based on your API usage patterns. The platforms that provide the intelligence could also provide the capital, creating an even tighter loop of dependency.
This could lead to a bifurcation in the startup ecosystem. On one side, venture-scale businesses that are essentially highly leveraged bets on specific AI capabilities. On the other, a new class of software companies that look more like traditional businesses—profitable from day one, growing on revenue rather than funding, and owned by their operators rather than investors.
The New Calculus of Value
If intelligence is becoming a utility, then the calculus for valuing a software company must also change. The question is no longer just "What have you built?" but "What durable assets have you accumulated that AI cannot replicate?"
The answer seems to be a combination of the primitives discussed earlier. A company's defensibility is no longer its codebase, but a function of its:
- Proprietary Distribution: The ownership of a user relationship, channel, or network that cannot be easily accessed by a competitor.
- Data Moat: A unique, compounding dataset that improves the product and is generated by the user's interaction with it.
- Operational Excellence: The proven ability to deliver a reliable, secure, and compliant service at scale, which remains a significant, non-trivial challenge.
These elements, when combined with taste—the deep user empathy to build the right things—and velocity—the organizational capacity to iterate faster than the competition—form the new foundation of value.
The age of AI doesn't diminish the importance of software startups. It re-focuses it. Value is shifting from the static asset of code to the dynamic process of building, learning, and adapting. The most valuable companies of this new era will be those that master this process, treating their codebase not as a fortress, but as a fluid manifestation of their deep connection to the market.