Divergence

The first portfolio review on April 16 ("Week One") found 9 of 10 positions green and the three theses moving as a block. The framing then was that AI infrastructure is one interconnected stack, and confirmation anywhere benefits everything. Two weeks later, that block has separated.

Equal-weighted across 10 positions since entry on April 9-11, the portfolio is up 2.6%. The average hides a real divergence:

Transformer bottleneck: +12.2% (ETN, GEV, PWR)

Heat wall: +2.7% (VRT, NVT, MOD)

Copper squeeze: -4.6% (FCX, SCCO, COPX, BHP)

The macro thesis is intact. The implementation story is the news.

Cooling: thesis confirmed by earnings

Vertiv reported Q1 on April 22. Revenue $2.65 billion, up 30% year-over-year. EPS $1.17 against $1.01 expected. Adjusted operating margin 20.8%, expanded 430 basis points. Full-year 2026 guidance: $13.5-14.0 billion in revenue (29-31% organic growth) and $6.30-6.40 EPS, implying 50-52% earnings growth.

The numbers behind the numbers matter more. Vertiv announced an 800V DC power architecture co-developed with NVIDIA, timed to NVIDIA's 2027 Rubin Ultra rollout — keeping VRT a generation ahead of the silicon. They are capacity-expanding via 3-4% of sales capex. Hyperscaler capex consensus for 2026 has climbed to $527 billion (some estimates run to $650 billion), up roughly 71% year-over-year.

The thesis from "The Heat Wall" was that the January CES sell-off had misread Jensen Huang's remarks: liquid cooling becomes mandatory, not optional, when racks pass 100kW. Vertiv's print is the cleanest possible validation. Revenue acceleration, margin expansion, multi-year visibility through Blackwell into Rubin.

So why is VRT only +3% as of yesterday's close, off from a +9% intraweek peak? Two factors. The stock had run hard into earnings on positive expectations, and beats need to be very large to keep momentum. The broader market has rotated out of names that already moved.

None of that is thesis breakdown. It is market action around an earnings print that confirmed the underlying story is stronger than at entry. No signal changes. BUY VRT, BUY NVT, HOLD MOD.

Transformers: still the cleanest expression

The transformer leg has been the steadiest of the three. ETN at +12%, GEV at +16%, PWR at +9%. Earnings season provides ongoing confirmation. GE Vernova reported April 22 with continued backlog growth. Eaton holds momentum from its earlier investor day. Quanta Services reports April 30; the book-to-bill is what to watch.

The fundamental story has only gotten stronger since Week One. AI capex continues climbing. Transformer lead times remain at 3-5 years for large units. Domestic manufacturing investments are accelerating, with Eaton's $340M South Carolina facility on track for 2027. The Bloomberg piece on China's 60% share of global transformer capacity has gotten more attention; tariff risk on transformer imports is now actively discussed in earnings calls.

No signal changes. BUY ETN, BUY GEV, HOLD PWR.

Copper: thesis valid, equities struggling

The copper commodity thesis is intact. Copper averaged $5.80/lb year-to-date, briefly exceeded $6/lb in Q1. Analysts still project a 330,000-tonne deficit for 2026. Demand from electrification and AI infrastructure continues building. The macro story has not changed.

But the equities have caught problems the macro does not capture.

FCX reported Q1 on April 23 and beat earnings expectations. The stock has not held the beat. The Grasberg mine, which produces a substantial share of FCX's output, faces material handling bottlenecks limiting the second-half production ramp to 60,000 tonnes per day. That is well below pre-incident capacity. Company-specific operational drag is pulling the stock down despite favorable commodity prices.

SCCO was moved BUY→HOLD in Week One after CEO Oscar Gonzalez Rocha's death. The position is now down 4% as the leadership transition continues without a named successor. The Tía María expansion project remains "under review."

COPX, the broad copper miners ETF, is essentially flat — -0.4% at Week One, -4% today. It tracks the equity space, which is suffering even as the commodity holds up.

BHP, the more diversified position, has held up better at +1.5%. Its broader portfolio cushions the copper-specific drag.

The lesson from this divergence: the macro thesis (copper demand from electrification and AI infrastructure) is correct, but the equity expressions of that thesis are caught between commodity prices that work for them and company-specific issues that do not. FCX with Grasberg, SCCO with leadership transition. The April tariffs partially offset by supporting domestic prices, but not enough.

No signal changes — yet. The fundamental case for copper exposure remains. But if FCX's Grasberg issues persist into H2 or SCCO's succession lingers, those become equity-thesis questions rather than commodity-thesis questions. Watch for FCX guidance updates over the next 60 days and SCCO's CEO announcement.

What the divergence means

The Week One framing was that the three theses are entangled, one interconnected system where each constraint amplifies the others. That framing remains correct at the macro level. AI capex still drives all three. Infrastructure constraints still cascade. But equities do not always price the macro cleanly. Equities price macro plus company-specific everything.

Transformers and cooling are expressed through companies that benefit cleanly from the macro. ETN, GEV, VRT, NVT build the things in shortage; the AI buildout flows directly into their order books. Copper is expressed through companies that need to extract from troubled mines, manage CEO successions, and navigate political risk in Peru and Indonesia. The macro tailwind is real, but it has to fight through company-specific friction.

This is a useful clarification of the original framing. The "physical infrastructure as bottleneck" thesis is not a single trade. It is three trades with different beta to the same underlying story. The transformer and cooling expressions are higher-quality. The copper expression is messier and more dependent on operational execution at individual mines.

The portfolio remains long all three theses. Position weighting has not changed. But the recognition that equity expressions of the same macro can diverge sharply, even within two weeks, is itself a finding worth holding. The next portfolio review will revisit whether copper's company-specific issues resolve, and whether transformer earnings (PWR April 30) confirm or complicate the leg that has been working.