Investors obsess over moats, but in a rapidly changing AI landscape, a startup's ability to quickly build and ship products that unlock latent demand is a more reliable predictor of success than any theoretical defensibility.
The middle layer of the AI stack (software infrastructure for data movement or frameworks) is a difficult place to build a company. Foundation models are incentivized to add more capabilities from below, leaving little room for defensible platforms in between applications.
Creating frontier AI models is incredibly expensive, yet their value depreciates rapidly as they are quickly copied or replicated by lower-cost open-source alternatives. This forces model providers to evolve into more defensible application companies to survive.
The future of software isn't just AI-powered features. It's a fundamental shift from tools that assist humans to autonomous agents that perform tasks. Human roles will evolve from *doing* the work to *orchestrating* thousands of these agents.
The emergence of high-quality open-source models from China drastically shortens the innovation window of closed-source leaders. This competition is healthy for startups, providing them with a broader array of cheaper, powerful models to build on and preventing a single company from becoming a chokepoint.
Unlike previous tech cycles where early revenue was a strong signal, the current AI hype creates significant "experimental demand." Companies will try, pay for, and even renew products that don't fully work. Investors must look beyond revenue to assess true product-market fit.
Major platform shifts like AI reward founders who are not burdened by historical context or "how things have been done before." This creates an environment where young, inexperienced teams working with high intensity (e.g., "9-9-6") can out-innovate incumbents with existing business models.