Every major technology cycle produces a handful of companies that don't just build better products—they define entirely new markets. These category creators capture 76% of the value in any space they create, and hold it for decades. This portfolio exists to find those companies during the 3–7 year formation window, before the market fully understands what they are. Right now, that window is in AI infrastructure.
The framework is durable. The arena changes.
The pattern repeats across every major technology cycle. The railroad era had companies that created the "integrated rail network" category before anyone had language for it. The oil era had Standard Oil. The PC era had Microsoft owning "operating system," Oracle owning "relational database." The internet era had Google owning "search," Salesforce owning "cloud CRM," AWS owning "cloud infrastructure." Each time, a small number of companies defined entirely new categories and captured 76% of the economic value. Everyone else competed for scraps.
The insight is not about AI. The insight is about category creation — finding companies in the 3–7 year formation phase, when institutional capital hasn't yet priced what they actually are, and holding long enough to let the compounding happen. AI infrastructure is simply where that pattern is most active right now. When this cycle matures, the framework travels to wherever the next category formation is occurring.
Current deployment: enterprise AI infrastructure — data security, observability, data platforms. The picks and shovels that every AI application requires, regardless of which AI models ultimately win. Domain advantage: 10+ years as a Software Product Manager in enterprise software, providing direct experience evaluating technical moats, category positioning, and authentic category creation versus marketing narrative.
The enterprise AI stack has seven layers. Category creation is most active at L3 and L6—data foundation and security/observability—where new problems are being defined that didn't exist five years ago.
Two time horizons. Different return targets. The same framework threshold of 8.0+ to qualify.
Goal: 10x return over 10 years. Requires starting market cap small enough for that math to be realistic. Maximum conviction, minimum activity.
Goal: 3–5x return over 5 years. These are proven category leaders with deep moats—already too large for 10x, but with clear compounding ahead. Framework score still required.
Capital held in money market instruments, earning yield while waiting for specific deployment opportunities. Cash is a position, not laziness.
Documented publicly for accountability. Cannot be revised with hindsight.
| Month | Portfolio | S&P 500 | Notes |
|---|---|---|---|
| Jan 2026 | −19.3% | +1.4% | First full month · Thesis unchanged |
| Dec 2025 * | +11.8% | +2.4% | Partial (Dec 23–31) |
A four-filter system. All filters must score 7.0+, and the weighted total must reach 8.0. The average evaluated company scores 5–6. Most are rejected.
Companies that define new categories capture 76% of category economics and hold it for decades. A company with a better product competing in an existing category is interesting. A company that makes the old category obsolete is investable. The framework exists to tell the difference.
This is not an AI portfolio. It is a category leader portfolio currently deployed in AI infrastructure because that is where the formation phase is most active. When this cycle matures, the same framework evaluates the next one — biotech infrastructure, energy tech, whatever category creation looks like in 2030.
Too early (0–2 years): the category doesn't exist yet, risk is existential. The window (3–7 years): category is forming, the leader is emerging, but institutional capital hasn't fully priced it. Too late (8+ years): commoditization begins, margin compression follows. The framework targets only the window.
For every position, the arithmetic is verified before entry: what market cap does this company need to reach for 10x (or 5x), and is that realistic? ServiceNow and Cloudflare are exceptional businesses. At $113B and $60B respectively, the 10x math doesn't work. Framework score is necessary but not sufficient — valuation runway is the final gate.