Investment Research & Speculative Infrastructure

Core Thesis

Apple's hardware moat is deepening, not plateauing. As spatial computing, AR wearables, and silicon integration converge, the supply chain feeding Apple's ecosystem becomes a bet on the infrastructure layer beneath one of the world's most locked-in consumer platforms. QCOM supplies modem IP Apple can't fully replace yet; STM and eMagin (EMAN) supply microdisplay components for the spatial layer; Kopin (KOPN) is a dark-horse microdisplay bet on AR headset scaling.

Historical Analogy

Intel inside the PC era, 1990–2000. The brand on the box captured headlines; the component supplier captured margin. Apple is the box. The suppliers are the Intel. This thesis bets on the component layer of the next device category — spatial computing — before mass adoption arrives.

Proof of Concept

Apple Vision Pro launched in 2024 as a $3,499 developer device — seeding the ecosystem before the mass SKU. Sony (SNE) supplies OLED panels for the Vision Pro already. STMicroelectronics supplies motion and sensor ICs across iPhone/Watch/AirPods families. This is not speculation about whether Apple enters spatial computing; it already has.

Primary Risk Factor

Apple aggressively designs out suppliers — they replaced Intel chips with Apple Silicon. QCOM's modem dependency has an expiration date. Smaller plays (KOPN, EMAN) are illiquid microcaps with high binary risk on contract wins.

5-Year Window

If a mass-market Vision Pro SKU launches at sub-$1,500 by 2027, EMAN and KOPN become acquisition targets. STM and QCOM mature steadily. AAPL is the anchor; the others are call options on the spatial layer.

Core Thesis

AI compute demand is doubling every 18 months. The physical infrastructure that houses that compute — data centers, cooling systems, fiber interconnects, power substations — is the constrained resource, not the chips. Equinix (EQIX) and American Tower (AMT) are the real estate layer of AI infrastructure. Digital Realty (DLR) and CyrusOne (CONE) are the hyperscale layer. NVDA is the chip that fills the center; TSM is the foundry that makes the chip. This thesis bets on the physical layer more than the software layer — the landlords, not the tenants.

Historical Analogy

Standard Oil's pipeline network, 1880s–1910s. Rockefeller didn't just drill oil — he controlled the infrastructure through which all oil flowed. EQIX and AMT are building the same position in digital infrastructure: the neutral interconnection points through which all internet traffic and AI inference must pass.

Proof of Concept

EQIX has delivered positive returns in 19 of the last 20 years. AMT's tower portfolio generates predictable lease escalators regardless of which carrier wins the wireless wars. DLR's hyperscale leasing pipeline hit a record in 2024 driven entirely by AI infrastructure demand from Microsoft, Google, and Amazon.

Primary Risk Factor

The binding constraint on data center growth is power — not capital, not land, but megawatts. Utilities cannot deliver power fast enough. Hyperscalers (Microsoft, Google, Amazon) are increasingly building their own data centers, potentially reducing third-party demand. REITs are rate-sensitive.

5-Year Window

This is a front-loaded thesis — the AI infrastructure buildout is happening now, not in 5 years. EQIX and AMT are accumulate-on-dips positions. NVDA is volatile; size accordingly. TSM is a long-duration hold on semiconductor sovereignty.

Core Thesis

Tesla's real moat is not the car — it is the largest real-world driving dataset on Earth, collected by millions of vehicles. As autonomous vehicle infrastructure expands from consumer roads to mining, agriculture, and municipal logistics, the companies building physical road and terrain infrastructure become critical. TRMB (survey/mapping), CAT (autonomous construction), and KMTUY (Komatsu — autonomous mining) are the terrain-layer picks that feed the world Tesla and its competitors are trying to automate.

Historical Analogy

The railroad buildout of the 1860s–1880s. The land surveyors and steel manufacturers (not just the rail companies) captured durable value. As roads, mines, and farms go autonomous, the companies that map and grade the physical world become infrastructure-layer plays with defensible moats.

Proof of Concept

Caterpillar's autonomous mining trucks already operate in Australia and Chile with no human operator. Komatsu's autonomous haul systems have logged over 5 billion tons of material moved. Trimble's positioning systems are embedded in autonomous ag equipment globally. The non-Tesla layer of this thesis is already operating at scale.

Primary Risk Factor

TSLA carries significant idiosyncratic risk tied to Musk's public profile. Robotaxi regulatory approval timelines in the US are unpredictable. The municipal contract catalyst needed to upgrade this thesis has not materialized. The non-TSLA picks (TRMB, CAT, KMTUY) are cleaner risk profiles independently.

5-Year Window

This is the longest-duration thesis in the portfolio. CAT and KMTUY compound quietly regardless of autonomous vehicle timelines. TRMB is a steady infrastructure play. TSLA is a high-variance bet on whether autonomous data moat translates to a licensed platform business.

Core Thesis

Google's search advertising monopoly is structurally eroding as AI answers replace search queries, and as younger audiences route attention through audio, short video, and social feeds instead of search boxes. This thesis bets on the beneficiaries of that re-routing: Spotify (SPOT) as the audio attention layer, The Trade Desk (TTD) as the programmatic advertising infrastructure that doesn't depend on Google's walled garden, META as the social feed layer, and Amazon as the commerce-intent search replacement.

Historical Analogy

Yellow Pages to Google, 1995–2005. A dominant information-retrieval monopoly was displaced not by one competitor but by a platform shift. AI-generated answers are doing to Google's search box what Google did to the phone book. The companies routing around the old monopoly compound the fastest in the transition decade.

Proof of Concept

Google's share of the US search advertising market fell below 50% for the first time in over a decade in 2024. Spotify's podcast and audiobook ad revenue has grown 30%+ YoY for three consecutive years. TTD's CTV and retail media revenue lines are expanding faster than its core display business, capturing budgets migrating away from Google Display Network.

Primary Risk Factor

Google is not dead — it is adapting. Gemini integration into search could re-lock query volume. A DOJ-mandated ad-market breakup could perversely benefit Google by creating a cleaner, more competitive market. TTD's growth depends on open internet advertising remaining viable against walled gardens.

5-Year Window

The attention re-routing is happening now, not in 5 years. This is the most near-term thesis in the portfolio. TTD and SPOT are the cleaner infrastructure plays; META and AMZN are large-cap hedges on the attention shift with less upside multiplier.

Core Thesis

GLP-1 drugs (Ozempic, Wegovy, Mounjaro) are not just obesity treatments — they are the first proven metabolic reprogramming drugs in history, with emerging evidence touching cardiovascular disease, Alzheimer's, addiction, and cellular aging. This thesis bets on the companies at the center of the metabolic revolution (NVO, LLY), the AI-driven drug discovery infrastructure enabling the next generation of longevity interventions (RXRX — Recursion Pharmaceuticals), and the senolytic frontier (UBX — Unity Biotechnology, ARKG — ARK Genomic ETF).

Historical Analogy

Statins, 1987–2000. Lovastatin launched as a cholesterol drug and became one of the most prescribed medications in human history, redefining cardiovascular risk management for a generation. GLP-1 agonists are tracking the same trajectory — a class of drugs initially approved for one condition revealing multi-system benefits that expand the addressable market by an order of magnitude every 3–4 years.

Proof of Concept

Semaglutide (Novo Nordisk) demonstrated a 20% reduction in major cardiovascular events in the SELECT trial — a non-diabetic, non-weight-loss endpoint. This is a paradigm shift: a metabolic drug with cardioprotective effects. Recursion (RXRX) has partnerships with Roche and Bayer deploying AI phenomics at industrial scale.

Primary Risk Factor

NVO and LLY face patent expiration risk and biosimilar competition by the early 2030s. UBX is a pure clinical-stage binary: if Phase 2 senolytic trials fail, the stock approaches zero. RXRX has been burning cash and has not yet demonstrated a drug through to commercial approval.

5-Year Window

This is designated the highest-upside thesis in the portfolio. NVO and LLY compound on GLP-1 global expansion. RXRX has 10x potential if any AI-discovered drug reaches Phase 3. UBX is a 0 or 50x binary. Hold small positions in UBX and RXRX; scale NVO and LLY on any pullbacks.

Core Thesis

Water is the only resource with zero synthetic substitute and guaranteed demand growth. American Water Works (AWK) and Xylem (XYL) are the infrastructure operators and technology providers managing an aging US water system facing $1 trillion in deferred maintenance. The container layer (BALL, CCK, AMCR) captures the packaging shift away from plastic toward aluminum and sustainable materials driven by regulation and consumer pressure. Coca-Cola (KO) is a defensive anchor — the world's largest water purchaser and distribution network, inflation-resistant, dividend-paying.

Historical Analogy

Municipal utility bonds, 1920s–1950s. Unglamorous, essential, politically protected, inflation-adjusted, monopoly-adjacent. Water utilities are the original infrastructure investment — they don't grow fast, but they don't die. XYL is the modernization layer on top: smart water meters, leak detection AI, flood management.

Proof of Concept

The EPA's Lead and Copper Rule revisions mandate pipe replacement across US municipalities through 2030, creating a $45–65B infrastructure spend that flows directly through AWK's service territory and XYL's equipment business. Ball Corporation reported aluminum can demand growing 6% annually in North America as beverage companies shift away from plastic.

Primary Risk Factor

The primary risk is the thesis's own safety feature: AWK's returns are regulated by public utility commissions, capping upside. This is a capital preservation and modest compounding thesis, not a 10x bet. Container picks face risk if plastic tax legislation stalls or is repealed.

5-Year Window

This is the only thesis in the portfolio with no bad entry window. AWK and XYL compound regardless of market conditions. Build this position first; use it as the ballast that lets you hold riskier theses without emotional capitulation.