Lonis Hamaili: Launching an AI Supply Chain Hedge Fund

Lonis Hamaili shares his background as a Swedish entrepreneur who dropped out of college in 2019 to start an AI company in e-commerce, which attracted exceptional talent from programming and physics olympiads. Several co-founders went on to found billion-dollar companies. After selling a crypto platform in 2022 and building an early AI agent product (God Mode) that reached one million users, Hamaili became obsessed with positioning for fast AI timelines. He concluded the best approach wasn't betting on individual AI labs due to government intervention risk and execution uncertainty, but rather being long the entire AI supply chain—the semiconductors, memory chips, silicon wafers, power infrastructure, and related inputs that all players will require.

Hamaili co-founded Cambrian Management with Kuram, who had been trading this thesis successfully for four years with ~4000x returns. Launching with ~$70 million in September, the fund invests across Asia and global markets, including micro-cap Japanese suppliers. The core thesis is that AI supply chain oligopolies trade at severely depressed multiples (3-5x sales, 5-15x earnings) compared to software companies (15-50x sales) because the market hasn't priced in the operating leverage these companies gain during supply crunches. As volumes increase with limited ability to quickly expand capacity (fabs take years to build), pricing power compounds profits dramatically.

Hamaili emphasizes preferring memory and wafer companies that avoid long-term agreements with price ceilings, as these lock in pricing that will soon command massive premiums. He likes supply layers with structural constraints—memory can only double supply by decade's end; silicon wafers take years to scale. The fund evaluates companies using AI-powered research tools to analyze earnings reports, scrape supply chain data, track new technologies, and model scenarios. Half the portfolio is in Asia, with roughly a third in small and mid-cap companies including a $30 million Japanese firm.

On broader market dynamics, Hamaili argues the market isn't bullish enough on AI and is bullish in the wrong ways. Enterprise agent adoption—not consumer—represents the true opportunity, as the $30 trillion annual white-collar workforce is 100x larger than consumer markets and could see massive displacement and transformation. He notes token costs continue dropping while capabilities improve, driving exponential usage growth (Jevons paradox), and predicts the next major consumption wave will be persistent agents running 24/7.

On macro implications, Hamaili discusses how AI corporate debt issuance ($150B in 2023, ~$400B in 2024, potentially $1T in 2025) competes with US Treasury purchases, forcing yields higher. Since AI CapEx bonds offer 6-10% returns while Treasuries offer much less, pension funds and insurers are reallocating. Combined with lower savings expectations as people anticipate AI-driven growth, he expects Treasury yields to spike significantly—potentially to 8-10% or higher, echoing Blake Byers' prediction of a rates-driven reshuffling from savings to productive CapEx.

He avoids the neo-cloud providers (CoreWeave, Lambda Labs) preferring hyperscalers like Meta and SpaceX, which can finance GPUs through cash flow and charge $50B per gigawatt annually without long-term agreements, versus $10B for constrained providers. He also discusses shorting Treasuries as part of the macro strategy.

On risks, Hamaili acknowledges that strong government intervention—regulatory crackdowns prompted by AI-related hacks, misuse, or deaths—could slow development and hurt supply chain demand. However, he notes this could actually benefit supply chain holders by forcing a more decentralized frontier with multiple competing labs. The most likely near-term threat is regulatory backlash, not technical saturation, given AI's consistent improvement over multiple orders of magnitude.

On investment approach, the fund operates with a 12-15 position concentrated portfolio, primarily long equities (~two-thirds) and options (~one-third), deliberately unlevered to avoid getting liquidated during inevitable volatility. Positions are sized toward target fair values derived from deep analysis of pricing structures, competitive dynamics, technical risk, and Chinese competition. They build positions over time and may use options to maintain exposure during downturns. The fund avoids day trading, places trades weekly to monthly, and maintains a multi-year time horizon, insulating itself from daily sentiment swings.

Hamaili emphasizes the importance of becoming a deep subject matter expert in specific supply chain technologies, using AI tools (Claude, custom scrapers, tracking systems) to digest thousands of pages of research, earnings calls, and academic papers. He reads every Druckenmiller interview, earns reports obsessively, and uses AI as a collaborative thinking partner to understand complex technical architectures and find alpha in overlooked niches and non-English sources.