Goehring & Rozencwajg: Oil's "Demand Destruction" Is Really Lost Refining Capacity, and Gold Likely Has Further to Fall Before New Highs
Adam Rozencwajg on Merryn Talks Money, October 20, 2026
The 6 million barrel demand hole may be a refining artifact
The most useful new argument from Adam Rozencwajg, portfolio manager at Goehring & Rozencwajg, concerns the oil market 215 days or so into the Strait of Hormuz closure. When the strait shut on March 1, his firm estimated that 100 days of lost flows at 10 million barrels a day would remove about 1 billion barrels from the market. That is more than inventories could cover, so the firm warned of "tank bottoms" by summer. It was not alone, since JP Morgan reached a similar conclusion. The closure has lasted more than twice as long as assumed and tank bottoms have not arrived. Bears read that as proof the market was heavily oversupplied going in.
Rozencwajg rejects that reading. The headline data imply that about 6 of the 10 million lost barrels have been offset by demand destruction, a figure he calls worse than the global financial crisis and on par with Covid lockdowns, even though GDP prints remain positive almost everywhere. His point is that oil demand is not measured. "Nobody goes and looks at how much gasoline is going into gas tanks and jet planes," he said. Agencies model it, and the largest input is refinery runs.
That is where he sees the distortion. Since the war began, he said, the world has lost almost exactly 6 million barrels a day of refining capacity: about 2 million in Russia from Ukrainian strikes, about 2 million in the Persian Gulf, and about 2 million in China, where he described the cut as a policy decision to cap domestic gasoline and diesel prices. In his view, consumption of refined products has not fallen. Refinery runs have. The crisis is real and worsening, he argued, but it is showing up in product prices and logistics rather than in the crude market, where visibility is limited.
The positioning is candid. Rozencwajg acknowledged the firm should have owned refiners from the start of the crisis and did not. It judged that crude producers would benefit most, but, as he put it, "we didn't appreciate how long and how severe" the logistics bottleneck could be. He now argues that investors who bought refiners should rotate into upstream names. With diesel near $200 a barrel and a normal crude-to-diesel gap of about $20, he sees crude rising to close the spread, which implies roughly $80 of upside. Export volumes from the Middle East are back near pre-conflict levels, so he sees the constraint as refining and distribution rather than shipments.
The firm favors pure-play producers with long reserve life, and names oil sands operators Suncor and Canadian Natural Resources as large holdings. He called them companies with "100 year assets" whose capital is already spent and which "almost never decline." He is more cautious on integrated majors such as Shell and BP. Higher crude would lift their upstream earnings but erode refining margins, so the effects "somewhat net out." Their shares have also already rallied on refining strength.
Why the firm cut gold from about 30% of the portfolio to 5%
On gold, Rozencwajg remains a long-term bull but a near-term seller. He expects new highs this cycle, possibly by 2035, but said "gold will move lower before it moves higher." The firm cut its gold-equity weighting from close to 30% of the portfolio last year to 5%, and cited three triggers.
The first is relative value. Gold looks expensive against real goods and cheap against financial assets, and the gold-to-oil ratio hit an all-time high in January, when one ounce bought 90 barrels. Historically, a ratio of 30 has been enough to justify rotating from gold into oil. The second is speculative positioning. Western investors, who tend to follow trends, accumulated bullion through ETFs such as GLD, which he fears sets up forced liquidation if prices fall. The third, and the main catalyst, was what partner Lee Goehring calls the silver catch-up ratio. Gold bull markets typically start with silver lagging for years, followed by a violent catch-up rally. Over the full move the two perform about the same. The catch-up rally is the signal to reduce exposure, since both metals tend to retrace afterward. Silver did exactly that in January after lagging for three or four years.
The firm also believes the last leg of the gold rally priced in 2 to 3 rate cuts under incoming Fed Chair Kevin Warsh. Its January letter argued that the market was positioned for cuts and that a hike was the unpriced outcome. "When everyone is expecting cuts, more often than not, the market does what punishes the greatest number of people," Rozencwajg said. With Hormuz-driven inflation, the Fed has begun a hiking cycle, and gold sold off sharply on Warsh's first hike. He said he sees no evidence that further hikes are priced in.
His historical work shows gold retracements after a silver signal typically run 20% to 45% and last about two years on average. Gold is now roughly nine months and 20% into the move, so a bottom here would be tied for the shallowest pullback since gold floated freely in 1971. Rozencwajg conceded that long-term holders using gold as insurance against fiscal stress or financial repression, and willing to endure another 20% to 25% drawdown before a move to roughly double the old $5,000 highs, have little reason to sell. On a two-year tactical view, he thinks the odds favor a better entry point. He does not see the central bank diversification trend as over, despite a pause in buying.
Shale growth has stalled, and Hubbert's logic applies again
The longer-term oil thesis is that US shale has effectively peaked. Rozencwajg frames this through M. King Hubbert, the Shell geologist who correctly predicted in 1955 that US production would peak in 1971. The firm applied the same logic to the North Sea and Cantarell in the early 2000s, and argued that their rollover, which accounted for about 70% of the prior decade's supply growth, tipped the market into deficit and drove prices to $145.
The firm began studying shale in 2019 and 2020 using a neural network it trained itself on shale well productivity. It concluded that growth would continue but slow each year, which has happened from 2020 to 2026. Year-on-year growth is now "flirting with negative." A price spike of 6 to 12 months could produce a brief rebound, but he pointed to the 1970s, when oil rose 15-fold, drillers completed four times as many wells, and US output still fell 30%. He sees demand as steady, at about 1 million barrels a day of annual growth, and said the energy transition narrative has not shown up in the numbers. EV penetration is stalling outside China, which he said is pursuing import independence rather than pure decarbonization.
The capital cycle framework: early, not late
Rozencwajg built the broader bullish case on a mistake. In 2012 the firm argued stocks printing money at 9 or 10 times earnings could not be in a bubble. It missed a capital investment boom that had begun in the early 2000s and peaked around 2010 and 2011, bringing on iron ore, coal and shale supply, and the sector entered an eight-year bear market. His conclusion is that investment takes 10 to 15 years to translate into supply, and that the lag is lengthening because the easy resources are found and environmental baseline studies alone can take five years. Prices therefore run "longer and farther than anyone expects." Everything after the midpoint of the cycle is, by definition, excess capital.
His preferred gauge is the ratio of commodity prices to equity prices. He extended the chart, which he credited to Jeffrey Gundlach and BCA's Tony Boeckh, back to 1900. The ratio moves in a consistent band, and only at the 2020 bottom has it been lower than today. He also noted that upstream oil spending of about $500 billion a year is well below prior peaks in real terms. Lithium illustrates the divergence between commodities. It had its capex boom during the ESG era while everything else was starved of capital, which is why it is lagging now.
Uranium: own the trend, ignore the noise
The firm bought uranium near the bottom at about $8 a pound and expects a steady upward trend punctuated by bouts of retail and hedge fund euphoria and pessimism. Because no one rings a bell at the turns, the firm simply holds. Cameco is the core position, and Kazatomprom is a smaller holding under review. Rozencwajg cited tax regime changes, the geopolitical position of Kazakh uranium between Russia and the US, and, most important, sharply deteriorating geology. "I don't think they have quite the exceptional assets they did five years ago," he said. He pointed to developers such as Denison as the likely next producers.
He is skeptical on listed small modular reactor companies. He considers the technology excellent in some cases but said valuations rest on revenue "sometime in the 2030s," and called some of the work "hype and hope." His view is that the eventual winner is Bill Gates' TerraPower, which is private.
Allocation and diversified miners
The firm avoids BHP, Rio Tinto and Glencore, whose multi-commodity mix dampens volatility but also caps upside capture. For generalist investors, Rozencwajg suggested 10% to 15% in commodities across a full cycle, with 20% to 25% possible for those who can tolerate drawdowns. He argued sizing should reflect risk tolerance rather than opportunity, given the asset class's volatility.