A concentrated book on the AI buildout.
Our core activity is a concentrated long/short book of public equities expressing a directional view on the downstream buildout of artificial intelligence. The model layer is only the visible tip; behind it is a much larger reorganization of the physical and industrial economy, electricity, data centers, memory, networking, and the firms whose businesses are being rewritten by what AI now makes cheap.
Every major technology cycle has played out the same way. The personal computer didn't only beat the typewriter, it rewrote how every business operated. The internet didn't only beat the newspaper, it absorbed advertising, retail, and distribution. Mobile didn't only beat the camera, it took value from carriers, music labels, and entire categories of physical hardware.
Each cycle moved value from incumbents who depended on the old physical infrastructure to platforms that scaled with near-zero marginal cost.
AI is the next cycle, and it is broader than the prior ones. The compute, the power, the memory, and the buildings the buildout requires are physical and capital-intensive. The applications, by contrast, are leveraged and asset-light. The pattern is the same: the businesses harnessing the wave compound; the businesses being disrupted erode.
We position on both sides of that transition. You're not betting on the market rising; you're betting on the future replacing the past.
A concentrated, long-biased book.
Roughly 10–25 longs, a smaller short book against structurally weak models. Net exposure is meaningfully positive; we are not market-neutral.
Position sizing tied to conviction and price.
Maximum long position sizes are set conservatively at cost basis. Maximum short positions are sized smaller because the cost of being early on a short is unbounded.
Risk discipline.
Diversification within the book to avoid clustering on a single sub-thesis. Continuously updated rankings and exposures as data and prices evolve.
We focus on companies whose business models have clear unit-economic asymmetry: cheap, scalable inputs producing disproportionately valuable outputs. Examples in our current thinking include the picks-and-shovels of compute and energy, fintech and consumer platforms with network effects that compound with scale, and operators in heavy industry whose costs fall as AI is applied to their workflows.
We've publicly written about several names we follow, including Robinhood, Hims & Hers, IREN, and Eos Energy. These are illustrative of the kind of companies we work on, not a recommendation. Positions change.
Illustrative · Not a recommendation · Positions change
The strategy is run by Milton Maxwell with the firm's own capital. We write publicly about how we think and the companies we follow, see Writing for our essays and notes.