The past thirty days have tested investor nerve. A joint US-Israeli military strike on Iran on February 28, 2026 triggered the largest disruption to global oil supply in recorded history. The Strait of Hormuz, through which twenty percent of the world's seaborne oil flows, ground to a near standstill. Brent crude peaked above $126 per barrel. Global equity markets repriced. Stagflation replaced soft-landing as the dominant narrative. Portfolios bled.
For most investors, this is a crisis. For Nova One Capital, it is a confirmation.
We write this not as reassurance. We write it as a public record of conviction. A document that we intend to stand behind, that we will be measured against, and that we believe, twelve and thirty-six months from now, will be recognized as having read the moment correctly.
The five structural themes underpinning the Nova X Convergence Fund are not hypotheses. They are observable realities, measured in hundreds of billions of committed capital, signed contracts, federal legislation, and physical infrastructure. The market sell-off has temporarily obscured that signal. It has not changed it.
"There may be more near-term downside. We are not calling a bottom. What we are calling is this: investors who hold or accumulate positions in these structural themes over the next one to three years, through the noise, will look back at this period as one of the most consequential entry windows of the decade."
Markets are extraordinarily efficient at pricing the present. They are far less efficient at pricing structural transformations that play out over years rather than quarters. This gap, between what the market prices today and what the world will look like in three years, is where we operate.
The Iran conflict has created genuine macro disruption. The IEA has called it "the greatest global energy security challenge in history." That is not hyperbole. Twenty percent of global seaborne oil was cut off at the source. LNG supplies from Qatar fell by over twenty percent. Commodity markets reacted violently, and broad equities followed.
None of that changes what the next three years look like for AI infrastructure, nuclear power, or digital financial rails.
Microsoft, Amazon, Alphabet, and Meta have collectively committed to spending between $635 billion and $665 billion on AI infrastructure in 2026 alone. That is a 67 percent increase over 2025 levels. These are not aspirational announcements made in bull market euphoria. They are binding capital allocation decisions made by four of the most financially disciplined organizations on earth, made after the conflict began, on earnings calls where analysts pushed hard on whether spending would slow. It will not slow.
What the macro environment has done is create a temporary disconnection between price and value across the names we hold. That disconnection, historically, is where the best long-term returns are made.
The Nova X Convergence Fund is constructed around five structural megatrends where multiple independent tailwinds converge on the same names at the same time. No name is held for macro exposure alone. Every position earns its place through a specific, falsifiable thesis.
| Names | Thesis |
|---|---|
| NVDA · AMD · AVGO | AI chip supercycle. NVDA owns the CUDA moat. AMD is the highest conviction value in the portfolio. AVGO is the custom silicon architect for every hyperscaler. |
| VST · CEG · LEU | Nuclear power is the only carbon-free solution to AI's 24/7 baseload requirement. The Iran conflict has made domestic, dispatchable clean energy more valuable, not less. |
| COIN · CRCL | Stablecoin legislation is codifying the digital dollar. USDC supply held at $78B while broader crypto fell 44%. This is infrastructure, not speculation. |
| PLTR | Enterprise AI operating system for classified and regulated environments. 70% Q4 revenue growth. 137% US commercial growth. |
| CRWV | Specialized GPU cloud with $66.8B contracted backlog. NVIDIA-backstopped at $87.20. The margin between GPU cost and compute revenue at scale is extraordinary. |
| TSLA · MU | Long-duration binary bets on Robotaxi/Optimus and the HBM memory supercycle. High asymmetry. |
AI Chips and Hardware: NVDA · AMD · AVGO
Every dollar spent building the AI economy passes through semiconductors. There is no version of the AI future that does not run on chips. The question is not whether the infrastructure buildout continues. It is who captures it.
NVIDIA is the anchor. The CUDA software ecosystem is not a product feature, it is a decade-long switching-cost moat that compounds with every model trained on NVIDIA silicon. Jensen Huang estimates hyperscalers will spend around $600 billion annually on AI compute infrastructure. Every iteration of that number benefits NVIDIA first.
AMD is our highest conviction value opportunity in the entire portfolio. The market is pricing AMD as a company growing at fifteen percent. AMD is guiding to 35 percent-plus revenue CAGR. Its forward PEG ratio sits below 1.0. The MI450 GPU architecture is shipping into contracted capacity at Meta and OpenAI. When markets recover their composure and refocus on earnings growth trajectories, the re-rating will be violent to the upside.
Broadcom is the most misunderstood name in the fund. It does not compete with NVIDIA, it wins from the same demand signal via a completely different mechanism: it is the custom silicon architect for the hyperscalers themselves. Google's TPUs, Meta's MTIA, and next-generation proprietary AI chips are Broadcom designs. As hyperscalers build custom silicon to diversify away from NVIDIA dependency, Broadcom captures that spend. AI revenue has already doubled year-over-year. It is one of the most misunderstood beneficiaries of the buildout relative to its role in it.
The thesis across all three names is simple: approximately $450 billion of 2026 hyperscaler capex is directed at AI-specific infrastructure. That capital flows through chips, and these three names sit at different points in the stack that capture it.
Nuclear and Power Infrastructure: VST · CEG · LEU
Here is a fact the market has not fully priced: the AI buildout has a hard physical constraint that chips cannot solve. It needs electricity. Continuous, dispatchable, carbon-free electricity, at a scale the existing US grid was never designed to deliver.
A single hyperscale campus can consume 500 megawatts or more. The US grid has no meaningful spare capacity. Renewables cannot provide the 24/7 baseload that AI workloads require, the sun goes down and the wind stops, but the inference servers do not. There is one technology that solves this cleanly: nuclear.
The hyperscalers already know this. Constellation Energy's 20-year, 1.1 GW power purchase agreement with Meta starting in 2027 is the template for what AI power contracts will look like for the next generation. Microsoft's deal to restart Three Mile Island under a 20-year PPA is another data point in the same direction. These are not green-washing announcements. These are legally binding, decade-spanning commitments to nuclear baseload power because the hyperscalers have determined there is no alternative.
Centrus Energy is the most asymmetric bet in the nuclear pillar. It is the only domestic producer of High-Assay Low-Enriched Uranium, the fuel for next-generation reactors. It holds a multi-billion dollar contracted backlog extending to 2040. It is a monopoly in a supply chain the US government has explicitly committed to securing.
Here is the geopolitical overlay that the market is missing: the Iran conflict, the one event causing the current sell-off, directly strengthens the nuclear thesis. When the largest oil supply disruption in history reminds every government and corporation on earth that fossil fuel supply chains are fragile, the premium attached to domestic, fuel-secure, dispatchable clean power rises. VST, CEG, and LEU are direct beneficiaries of the very event creating short-term volatility in the broader market.
President Trump's executive orders directing a quadrupling of US nuclear generating capacity by 2050 codify the policy support. The contracts are signed. The infrastructure is operating. The demand is accelerating. This is not a prediction, it is a supply chain that is already executing.
Digital Finance Rails: COIN · CRCL
The stablecoin market passed $320 billion in March 2026. Adjusted transaction volumes grew more than 90 percent year-over-year. Visa now supports stablecoin-linked cards across more than 50 countries. The GENIUS Act, signed in July 2025, created the first comprehensive federal framework for dollar-backed stablecoins. The US government has explicitly chosen private-sector regulated issuers to lead the digitization of the dollar.
This is not a crypto thesis. This is a financial infrastructure thesis.
Circle is the infrastructure layer of this shift. USDC supply held at $78 billion while the broader crypto market fell 44 percent from October 2025. That is the behavioral signature of infrastructure, not speculation, demand that survives market cycles because it is driven by utility, not sentiment. Circle's Payments Network now enables institutions to send USDC cross-border and settle in local currencies, with annualized volumes growing rapidly. Bernstein maintains a price target implying 60 percent-plus upside from current levels.
Coinbase is the compliance-grade exchange and custody platform that institutional adoption runs through. As digital assets move from speculative to systemic, regulated on-ramps with government relationships and robust custody infrastructure become structurally valuable. COIN is that platform in the United States.
A recent draft of the Clarity Act spooked the market with proposed yield restrictions on stablecoins. The sell-off was significant and, in our view, overdone. Citi described the market reaction as a "scaling setback, not a thesis killer." Regulatory clarity, even imperfect clarity, is a long-run positive for compliant, regulated issuers. It raises the barrier to entry and codifies the advantage of incumbents like Circle.
AI Software and Platforms: PLTR
Palantir is the software layer that sits above the AI hardware stack and transforms raw compute into enterprise decision intelligence. It is not a chatbot. It is not a wrapper on top of a foundation model. It is a purpose-built AI operating system for organizations that operate in classified, regulated, and high-stakes environments, the most defensible and highest-value segment of enterprise AI.
Q4 2025 results were exceptional: 70 percent total revenue growth. 137 percent US commercial revenue growth. A government contract base that provides structural revenue floors no startup can replicate. The AIP platform is winning enterprise deployments not because it is the cheapest option, but because it is the only platform that can handle the data governance, classification, and decision-audit requirements of serious institutions.
Palantir remains one of the highest-conviction names in the enterprise-software pillar, and May 2026 earnings are the next major catalyst.
AI Infrastructure: CRWV
CoreWeave is the fund's highest-conviction emerging position and the name most investors have not yet fully understood. It is not a general-purpose cloud provider. It is a specialized GPU cloud, built from the ground up for AI workloads, operating in a segment that AWS, Azure, and Google Cloud are structurally unable to serve as efficiently.
The investment case rests on three independent pillars: a $66.8 billion contracted revenue backlog with large hyperscalers, 140 percent revenue growth, and a NVIDIA backstop investment that provides both a structural floor and the most powerful endorsement possible from the most important company in the AI supply chain.
Q1 2026 earnings execution in May is the next major catalyst for the name. The unit economics of specialized GPU cloud are among the most interesting in the technology sector. CoreWeave captures the spread between what NVIDIA charges for GPUs and what hyperscalers pay for AI compute time, at scale, with a locked-in backlog that gives extraordinary revenue visibility.
Intellectual honesty matters more than comfort. So here is the unvarnished view:
There may be more downside. The Strait of Hormuz crisis is not resolved. Inflation could remain sticky. The Federal Reserve may delay rate cuts. Earnings multiples across growth names could compress further. The second quarter of 2026 could be worse for markets than the first. We are not calling a bottom. We never do.
"What we are calling is something different and more durable: the structural case for every position in this portfolio has not been weakened by the events of the past thirty days. In several cases it has been strengthened."
Consider the math of structural convergences at depressed entry prices. AI chip demand that doubles every eighteen months, hitting names already at discounted valuations. Nuclear power contracts with decades of locked revenue, priced as utilities rather than infrastructure monopolies. A digital dollar settlement layer that grew transaction volume 90 percent year-over-year during a crypto bear market.
Investors who look back at March 2026 will not remember the precise bottom. They will remember whether they held their thesis when it was uncomfortable to do so. The ones who did will have the better story.
- AI Capex Reversal. If hyperscalers simultaneously conclude that AI ROI does not justify the current spend trajectory, the chip thesis collapses. We assign low probability. All four major hyperscalers are already reporting measurable revenue improvements from AI. The spending is self-reinforcing because early movers are seeing returns.
- Geopolitical Escalation. The Iran conflict is the most significant geopolitical event in decades. Further expansion, particularly anything affecting Taiwan or the broader semiconductor supply chain, would introduce new risk to the chip pillar. To date, TSMC and the semiconductor supply chain have not been materially affected.
- Regulatory Uncertainty in Digital Assets. The Clarity Act's proposed yield restrictions created real volatility. Crypto regulatory risk remains asymmetric. The digital finance pillar is sized to survive a negative regulatory outcome while still delivering meaningful upside if the thesis plays out.
- Nuclear Timeline Risk. The nuclear renaissance is real but measured in years, not quarters. The thesis centers on operating capacity and fuel supply, not construction-stage projects. Contracted revenue backlogs provide visibility regardless of new build timelines.
- Valuation Compression. Growth multiples across the portfolio remain elevated even after the sell-off. A prolonged risk-off environment could compress them further. Valuation discipline and a long time horizon are the guard against this scenario.
The fund models three scenarios across a 5-year horizon. The 30-percent-plus CAGR target requires the bull case. The base case requires only partial execution across two to three pillars. The bear case requires simultaneous failure of all five structural themes.
| Scenario | Odds | 5-Yr CAGR | Required Condition |
|---|---|---|---|
| Bear | ~20% | -3% to +5% | AI capex reversal + nuclear stall + crypto winter, all simultaneously |
| Base | ~55% | +14% to +22% | 2–3 of 5 pillars deliver on thesis over 5 years |
| Bull | ~25% | +30% to +45% | All five pillars converge. Geopolitical tailwinds persist. |
The asymmetry is clear. Bear downside is modest. Bull upside is generational. Base case is competitive with any diversified growth portfolio. The fund is constructed around this asymmetric distribution, not around minimizing tracking error to a benchmark.
Every position in the Nova X Convergence Fund is held because of a structural argument measured in years, not quarters. The arguments for AI infrastructure, nuclear power, digital finance rails, enterprise AI software, and specialized GPU cloud are not weakened by the Iran conflict. Most of them are strengthened by it.
The hyperscalers are not pausing AI capex because of Hormuz. The data center power problem is not getting easier because oil is expensive. Stablecoins are not becoming less useful because BTC is down. Palantir's government contracts are not at risk from a macro selloff. CoreWeave's contracted backlog does not evaporate because markets are fearful.
Conviction is not stubbornness. It is the output of a thesis that has been stress-tested against current reality and held up. The fund has defined the specific conditions that would invalidate each position. None of those conditions are present today.
When markets give lower prices on names where the structural conviction has not changed, the disciplined response is patience, not reaction.
That is what Nova One Capital is doing. We believe the record will show we were right.
This document is for informational purposes only and does not constitute investment advice, a solicitation, or an offer to buy or sell any security. All investments involve risk, including possible loss of principal. Past performance does not guarantee future results. The views expressed represent the opinions of Nova One Capital as of March 30, 2026 and are subject to change without notice.