Reflecting on 2025: A Year of Framework Validation

2025 was a year that tested the Bottleneck-to-Ticker framework against a complex and at times counterintuitive market environment. AI infrastructure spending continued to accelerate, but the market's response to that spending was uneven — rewarding some supply chain positions heavily while dismissing others that we viewed as equally structural. Energy markets navigated a cycle of geopolitical supply disruption and demand uncertainty that tested conviction in our Commodities pool thesis. Through both, the framework performed its intended function: providing analytical discipline that prevented reactive repositioning in favor of structural conviction.

The most significant development in GZC's research process during 2025 was the expansion of our coverage universe into purpose-built AI infrastructure — specifically, the class of operators building HPC data center campuses optimized for GPU compute rather than general enterprise workloads. This expansion reflected our observation that as semiconductor supply constraints eased and GPU availability improved, the binding constraint in the AI infrastructure cycle was shifting downstream to power delivery and physical compute capacity. Applied Digital and Nebius Group are the two names that emerged from this expanded coverage as conviction positions.

Portfolio Construction Refinements

We made several deliberate refinements to our portfolio construction methodology during 2025. The most consequential was the introduction of a more systematic position sizing framework that explicitly accounts for bottleneck severity — the degree to which a specific company's product or service represents a supply chain choke point that cannot be bypassed or substituted over the relevant investment horizon.

Under this framework, NVDA and VRT receive maximum conviction weightings because they occupy supply chain positions with the fewest viable substitutes at the scale required. AMD receives a meaningful but smaller weighting reflecting the existence of a viable (if still disadvantaged) competitive alternative in NVDA. APLD and NBIS receive early-stage conviction weightings that will scale as execution track records develop.

We also refined our exit framework. We now maintain explicit monitoring criteria for each position that would cause us to reduce or exit: substitution risk events, pricing power deterioration, or the resolution of the structural bottleneck that underpins the thesis. Having these criteria defined in advance — before we have a position to defend emotionally — has improved the quality of our position management decisions.

New Coverage Names Added in 2025

Beyond APLD and NBIS in the Technology pool, we added several names to the Commodities pool coverage universe that reflect the broadening of our structural demand thesis beyond oil and gas:

Energy Fuels (UUUU) entered coverage as our primary uranium and critical minerals position. The nuclear energy renaissance thesis is not new — we have tracked uranium for years — but the convergence of AI data center power demand with nuclear policy support reached a threshold where conviction warranted a position.

Brookfield Renewable Partners (BEP) entered coverage as our clean energy infrastructure position. The hyperscaler clean energy PPA thesis is straightforward: hyperscalers have net-zero commitments and are executing multi-decade renewable power contracts. BEP is the counterparty at scale.

The industrial metals positions — ICOP and SLV — replaced legacy commodity exposure that did not have the structural demand characteristic we require. Copper and silver are bottlenecks in the AI infrastructure and energy transition buildout in a way that crude oil generalists and midstream operators are not.

Outlook for 2026

The macro environment entering 2026 is one of continued AI infrastructure investment with increasing visibility of the power and physical infrastructure constraints that will define the next phase of the cycle. We expect:

The nuclear energy procurement cycle to accelerate as data center operators move from PPA discussions to signed contracts. UUUU's revenue visibility should improve materially as this procurement activity crystallizes.

Copper supply constraints to become more visible as the pace of new mine development relative to demand growth becomes clearer. The supply gap thesis has been building for years — 2026 may be the year it becomes consensus.

European AI sovereignty policy to create the regulatory environment that validates NBIS's market positioning. EU AI Act implementation is still early, and the full demand implications for European AI compute infrastructure are not yet reflected in valuations.

We look forward to continuing to serve GZC clients through what we believe is a uniquely rich structural investment environment.

— Ahijah Ireland, Founder & CIO