History shows pioneers who fund massive infrastructure shifts, like railroads or the early internet, frequently lose their investment. The real profits are captured later by companies that build services on top of the now-established, de-risked platform.

Instead of building AI models, a company can create immense value by being 'AI adjacent'. The strategy is to focus on enabling good AI by solving the foundational 'garbage in, garbage out' problem. Providing high-quality, complete, and well-understood data is a critical and defensible niche in the AI value chain.

Current AI investment patterns mirror the "round-tripping" seen in the late '90s tech bubble. For example, NVIDIA invests billions in a startup like OpenAI, which then uses that capital to purchase NVIDIA chips. This creates an illusion of demand and inflated valuations, masking the lack of real, external customer revenue.

In the current market, AI companies see explosive growth through two primary vectors: attaching to the massive AI compute spend or directly replacing human labor. Companies merely using AI to improve an existing product without hitting one of these drivers risk being discounted as they lack a clear, exponential growth narrative.

Unlike the speculative "dark fiber" buildout of the dot-com bubble, today's AI infrastructure race is driven by real, immediate, and overwhelming demand. The problem isn't a lack of utilization for built capacity; it's a constant struggle to build supply fast enough to meet customer needs.

In a new, high-risk category, betting on infrastructure ('shovels') isn't necessarily safer. If the category fails, both app and infra lose. But if it succeeds, the application layer captures disproportionately more value, making the infrastructure a lower-upside bet for the same level of existential risk.

Vincap International's CIO argues the AI market isn't a classic bubble. Unlike previous tech cycles, the installation phase (building infrastructure) is happening concurrently with the deployment phase (mass user adoption). This unique paradigm shift is driving real revenue and growth that supports high valuations.

The enduring moat in the AI stack lies in what is hardest to replicate. Since building foundation models is significantly more difficult than building applications on top of them, the model layer is inherently more defensible and will naturally capture more value over time.

AI favors incumbents more than startups. While everyone builds on similar models, true network effects come from proprietary data and consumer distribution, both of which incumbents own. Startups are left with narrow problems, but high-quality incumbents are moving fast enough to capture these opportunities.