The Scissors

This review lands ten weeks after the book's entries and six weeks after The Scarcity Test sharpened the lens, and the tape since then has one shape almost everywhere: a mid-June peak, then a slide into July. Since May 31: the copper miners fell hard (COPX -14.5%, BHP -10.1%, SCCO -8.8%, FCX -7.9%, with FCX peaking at $68.68 on June 17 and losing 12% in the three weeks after), cooling gave back its spring (MOD -12.7%, NVT -5.4%, VRT +2.6%), and even the transformer names that made new highs pulled back from them (GEV peaked near $1,110 mid-June, ETN at $421). Against entry the book still looks comfortable, transformers +10-15%, cooling +2-21%, but momentum reversed weeks ago, and the interesting question is what reversed it. The answer differs by thesis, and each answer is a version of the same event: the market started pricing the second act this section had already written.

Start with the widest gap, because it is the piece's namesake. Copper the metal ran from a $5.64/lb April low to an all-time high of $6.72 on May 13 and still sits at $6.27, near records. Copper the equity complex fell double digits over the same stretch. Scissors: one blade up, one blade down. The divergence has three documented legs. First, Freeport: on June 30 the company cut its 2026 guidance a second time on the Grasberg ramp-up, sales down to roughly 3.1 billion pounds from 3.4, the block cave now expected at about 65% of nameplate in the second half against the 85% previously promised, full recovery pushed to late 2027 at the earliest. The stock lost 11% in a day. That is the same company-specific execution drag Divergence diagnosed in April, recurring. Second, the metal's own price is partly an artifact: the Section 232 tariff regime has pulled record inventory into the United States (COMEX stocks around 652,000 tonnes in late June against roughly 80,000 in February 2025), and Macquarie now calls the spot strength artificial tightness, forecasting surpluses above 700,000 tonnes a year for 2027-28, while Goldman argues the opposite, that refined-copper tariffs could push LME past $14,000 a tonne. Third, the structural discount: head grades keep falling, strip ratios keep rising, and producer P/NAVs sit compressed even at record metal prices.

Here is what the lens says about those scissors. The Scarcity Test held that copper's geology-bound scarcity is durable but its 2026 price is cyclically rich. The equity market, six weeks later, is voting for exactly that distinction: miners are long-duration claims on future copper prices, and they are discounting a future price lower than a spot propped up by tariff-driven hoarding. The lens survives; the specific expression pays for it. Two conclusions follow, and only one is comfortable. FCX goes from BUY to HOLD, not because the copper thesis broke but because the same single-name execution risk has now been realized twice in three months and the recovery timeline reaches into late 2027; a durable-scarcity holder wants the geology without betting the ramp-up schedule of one block cave. COPX stays BUY: the basket is the geology claim without the single-mine risk, and at -8% from entry some of the cyclical richness the lens worried about has already deflated out of it.

The transformer story produced the strangest week: fundamentals accelerating and the stock falling for the correct reason. The demand facts are not in dispute. GE Vernova's Q1 orders ran $18.3 billion, up 71%, its gas-turbine backlog hit 100 GW, and management reported 10-20% price increases on new bidding; Eaton's data-center orders ran up roughly 240% with a backlog it measures in years; Quanta's backlog set another record at $48.5 billion. And on July 7, Barclays downgraded Siemens Energy with an argument this section's readers have seen before: the sector is booking gas-turbine orders at roughly 50 GW annualized against 80-90 GW of sustainable global demand, so peak orders, peak free cash flow, and peak supply tightness all arrive together in 2026. GE Vernova fell 7-9% in a day on the read-across, with executive share sales near the highs for garnish. That is The Scarcity Test's 2027-28 overbuild wave, no longer a minority distinction but a sell-side headline that moves the tape.

The physical evidence agrees with the timing. Roughly $1.8 billion of announced North American transformer manufacturing investment is under way, and the completion dates cluster exactly where the lens put the risk: Siemens Energy's Charlotte plant in early 2027, Eaton's $340 million South Carolina plant in 2027, ERMCO's Arizona facility in 2027, Hitachi Energy's $457 million Virginia plant, billed as the largest transformer factory in the country, breaking ground on June 29 for 2028. And the first hard normalization datum is in: distribution-class transformer lead times have already improved from 52-plus weeks at the 2023 peak to 26-40 weeks for catalog units, while large power transformers still quote 128-144 weeks. The commodity end of the market normalizes first while the specialized end stays tight, which is precisely the fiber pattern What Stays Scarce described: overbuild arrives at the standardized product and mercy arrives last for the hardest thing to make.

So the transformer signals now enforce the lens's distinction between owning scarcity and owning its normalization. GEV goes from BUY to HOLD at 59 times trailing earnings and roughly 39 times forward EV/EBITDA against a peer median near 15; the business is executing superbly, and the multiple prices years of flawless execution precisely as the second act begins trading. ETN stays BUY: it is the cheaper claim on the same demand, its data-center backlog is measured in years, and its own South Carolina plant makes it a seller of the normalization rather than only a victim of it. PWR stays HOLD at 92 times trailing against a five-year median near 49.

Cooling delivered the review's cleanest specimen of price separating from value. On May 26, Modine signed a long-term capacity agreement guaranteeing more than $4 billion of Airedale data-center cooling supply through 2029 to a single strategic customer, who paid $165 million up front to reserve the capacity. A customer prepaying nine figures for years of future cooling is not a bottleneck trade; it is a moat being contracted in writing, the exact structure The Scarcity Test said cooling should be valued on. The stock is down 12.7% since May 31 anyway, caught in the sector slide and its own spin-off complexity. Evidence improved, price fell: MOD goes from HOLD to BUY. Vertiv and nVent stay BUY on their own facts, Vertiv with a $15 billion backlog and a co-design seat on NVIDIA's 800-volt Rubin architecture, nVent having raised full-year guidance again off 40% order growth, and the platform shift behind all three accelerating: Rubin is a fully liquid-cooled architecture speced for 45C coolant inlet, which moves the thermal bill of materials further from fans and chillers and deeper into the direct-to-chip loop these companies own.

The turbine positions, the book's newest, are one month old and nearly unchanged in price, which is fine, because the news validated their structure rather than their momentum. The capacity wave Build and Serve warned about is now being announced on schedule: Mitsubishi Heavy committing over 100 billion yen to double large-turbine capacity by 2030, Siemens Energy raising output toward 50 large units a year by fiscal 2027. That is the order-book trap forming in public. But the thesis was never the order book; it was the service annuity installed underneath it, and that is compounding: Siemens Energy's record 17.7 billion euro quarter came with a 154 billion euro backlog and orders like Oman's, six turbines wrapped in 20-year service contracts. Barclays' Underweight is aimed at the order-book multiple, and the read-across took SMEGF down with the sector. The annuity accrues either way. SMEGF stays BUY, MHVYF stays HOLD.

Q2 earnings referee all of this within four weeks. July 22, GE Vernova: watch whether order intake decelerates toward Barclays' sustainable-demand line, and whether the 2027 slot pricing holds. July 29-31, Vertiv, Quanta, nVent: Vertiv guided Q2 EPS to $1.37-1.43 and the street sits at the top of the range; nVent needs to show the 28-30% growth it promised. August 4-5, Eaton, Mitsubishi Heavy, Siemens Energy: Eaton's guided 150-basis-point sequential margin build in Electrical Americas is the single number that most directly tests whether pricing power is real, and Siemens Energy's print tests whether the pre-close 17.7 billion euro order figure survives. Freeport reports in there too, but its quarter is already confessed; what matters is whether the Grasberg 65%-of-nameplate schedule holds.

The meta-observation, and the reason this review exists at all: six weeks ago the lens's distinctions were this section's minority positions. In the first week of July the market began trading all three at once. Barclays is now selling the overbuild thesis; copper equities are now discounting the cyclically rich price; Modine's tape now separates the moat from the multiple. When the market catches up to a thesis, the easy part is over and the discipline part begins, because agreement is what a top looks like when it is happening to you. The scissors cut in both directions: they open a gap between price and value in copper miners, and they close one in grid equipment. The book's job in earnings season is to hold what is durably scarce, own what got cheaper while its evidence improved, and decline to chase what everyone now agrees about.