Hard is Back

The arc of contemporary startups, particularly digital ones, has singularly progressed towards a relentless reduction of barriers of entry. No longer are engineers having to maintain a pet server living in an office closet. Critical software is open source, automatically updated by a dependency manager, and carefully deployed by sophisticated CI/CD pipelines. Cloud computing is elastic and globally distributed all the way to the edge. We have incredible platforms that aggregate billions of people and are built on standards that allow one-size-fits-most distributions. Anyone anywhere can work with anyone else, thanks to virtual collaboration tools.

Yet, startups have avoided the fate predicted by the proverbial invisible hand, which would drag their gross margins to the efficient frontier. Why? It’s because software is eating the world. The world is huge, and growing, hence the demand for software is unbounded to a first degree. Supply meanwhile is constrained by individuals with the wherewithal and agency to direct the tendrils of software into the remotest niches of the economy. This process is fueled through boom/bust cycles of technological bubbles as eloquently articulated by Byrne in Boom: Bubbles and the End of Stagnation.

Thus, now decades into the golden era of startups, we have witnessed a renaissance of nerds who now dominate Forbes lists and direct Washington. Powerful feedback loops led to a dramatic increase in qualified talent and capital allocation, and the inception of championing institutions like Y Combinator. Aside from a few bad quarters here and there, the trend seems unstoppable. More talent leveraging more capital through more versatile platforms has resulted in a Cambrian explosion of monopolies.

While demand created by “software eating the world” is large, it’s not infinite, and AI is eroding the last standing barriers of entry. Trivialization of app production will lead to diminishing returns, and upside will vanish. There will be no more low hanging fruit—at least not the kind that leads to monopolies. Just as Shopify commoditized high-tech e-commerce into vanilla Bourgeois-marts, LLMs will commoditize the rest. Self-identified hustlers who became realtors because coding bootcamps were beyond their grasp will become tech CEOs.

The manic inertia will tolerate these moatless businesses, for a time, which will cause unprecedented fragmentation. For every Florida man building a ChatGPT wrapper for their previous employer, there’ll be hundreds more. Software will have eaten the world, and the world of software will exist in the boring plane of compressed margins.

The only way out of easy is hard. The solution is to tackle hard, messy problems. Hard, in short, is Back. Investors will soon realize that pedigree, prototypes, and even early revenue have become poor predictors of future success. Instead, value will accrue to founders who find some difficult, heavy lifting, or nasty piece of work in the bowels of the economy, for which there are no shortcuts but hard work and likely capex. Gone are the 7% WoW growth rates and 12-week demo days. In is the grind, the sweat, the messy whiteboards, the heroic negotiations, the millisecond advantage.

Good news is that Hard is also Fun. My best memories involve epic struggles. So I’m looking forward to it.