In January we told you China was the most unloved obvious trade of 2025. That thesis has been playing out. As the equity market keeps making new highs, our highest conviction position is now sitting in a vault.

We believe the decade ahead belongs to physical commodities — and most institutional investors are still holding paper proxies while the underlying assets reprice structurally. Here is why gold, copper, uranium and rare earths are our highest conviction 2026 thesis.

What the market is still pricing wrong

Global equity markets are at or near all-time highs entering Q4 2025. Risk assets have performed. AI has driven a remarkable re-rating of technology. The mood among most allocators is cautiously optimistic — perhaps too cautious about the wrong things, and not cautious enough about the right ones.

What we see beneath the surface is a world in which the physical underpinnings of the next decade of growth are severely underpriced. Wars are not ending — they are multiplying. Supply chains that the world assumed were globalised and efficient have proven fragile. The energy requirements of AI at scale are only beginning to be understood. Defence budgets across NATO and beyond are rising at rates not seen since the Cold War. And at the centre of almost every one of these trends sits the same answer: raw materials.

Most institutional portfolios are not positioned for this. They hold equities in mining companies — financial proxies that carry management risk, jurisdiction risk, and execution risk on top of the commodity exposure. We have taken a different view. We are going directly into the physical assets themselves, held in vault, and we believe this will be one of the defining allocation decisions of the next three to five years.

Re-arming, re-shoring, and the AI energy supercycle — all roads lead to physical

Three structural forces are converging simultaneously — and all three are commodity intensive in ways that markets have not yet fully priced.

The first is re-armament. The wars in Ukraine and the Middle East have not resolved — they have entrenched. Defence spending across Europe, Asia, and North America is rising at generational rates. Modern defence systems — precision munitions, drone fleets, electronic warfare, hypersonic development — require extraordinary quantities of specific materials: copper for electronics and wiring, rare earth elements for guidance systems and electric motors, titanium for aerospace structures, and uranium for energy independence. This demand is not cyclical. It is structural and government-mandated.

The second is re-shoring. A decade of globalisation assumptions have collapsed. Supply chains that ran through single geographies — particularly China — are being rebuilt at enormous cost and with enormous material requirements. Every semiconductor fab, every battery gigafactory, every data center being built in the US, Europe, or allied nations requires the same inputs. Copper. Rare earths. Energy. The irony is that the materials required to re-shore away from China are largely still controlled by China.

The third is AI energy demand. The compute requirements of frontier AI models are growing faster than efficiency gains can offset. Data centers are the fastest growing category of electricity consumption globally. That electricity has to come from somewhere — and increasingly, the answer involves nuclear, which means uranium, and grid infrastructure, which means copper. DeepSeek proved in January that China can build efficient models — it did not prove that the world needs less infrastructure to run them at scale. If anything, lower cost per inference means more inference, more deployment, more energy demand.

We were right on China. Now Europe needs to hear the next part.

In January 2025, we published our view that China was the most unloved and most mispriced trade of the year. We argued that Western investors had allowed geopolitical sentiment to completely disconnect from fundamental and macro reality. That thesis has played out through 2025 — Chinese technology has re-rated significantly, Alibaba and the broader AI infrastructure complex have recovered materially, and DeepSeek proved Chinese AI capability in a way that was impossible to dismiss.

But there is a second chapter to the China thesis that most Western allocators have still not fully absorbed — and it is more urgent than the first.

We expect rare earth supply constraints to become a mainstream institutional conversation within the next 12 months. When that happens, the assets will reprice quickly. We intend to already hold them — physically, not on paper.

Physical holdings in vault — gold, copper, uranium and rare earths

Our positioning in this thesis is deliberately structured around physical holdings rather than equities or derivatives. Mining stocks carry management execution risk, jurisdiction risk, and a correlation to broader equity markets that dilutes the commodity exposure we are seeking. ETFs and futures introduce counterparty and roll risk. We want the asset itself — held in secure, allocated vault storage — and we are willing to accept the illiquidity premium that comes with it in exchange for pure, unencumbered exposure.

We also believe that physical rare earth holdings are on the verge of becoming a more liquid and institutionally accessible asset class. The infrastructure for allocated physical rare earth storage and transfer is developing — driven precisely by the supply anxiety we described above. Early movers into physical rare earth positions will benefit both from the price appreciation and from the liquidity premium as the market matures around them.

Defence, aerospace, AI chips — all roads lead to the same physical inputs

Step back from the individual commodities and the picture becomes even clearer. The three largest investment themes of the next decade — defence and aerospace modernisation, AI infrastructure buildout, and the energy transition — share an almost identical bill of materials. Copper, rare earths, uranium, and to a lesser extent lithium and cobalt sit at the foundation of all three.

Defence and aerospace are accelerating their consumption of rare earth permanent magnets for next-generation systems. The F-35 requires approximately 900 pounds of rare earth materials per aircraft. Drone fleets require rare earth motors at scale. Hypersonic systems require rare earth components for guidance and propulsion. Every NATO member increasing its defence budget to 2% of GDP — and many are now exceeding that — is implicitly increasing rare earth demand.

AI chip expansion is driving copper and energy demand simultaneously. The gigafabs being built by TSMC, Intel, and Samsung in the US and Europe are among the most copper-intensive construction projects in human history. The power infrastructure to run them requires further copper investment. And the AI models they will run require uranium-powered baseload energy at scale that renewables alone cannot provide on the required timeline.

The geopolitical conclusion is stark: the Western world is attempting to build technological and military sovereignty using materials it does not control. China understands this. The progressive tightening of rare earth export restrictions through 2025 is not accidental — it is strategic. Europe, which has been the slowest to respond, faces the most acute exposure. We expect this to become a policy crisis within the next 12 to 24 months, and the commodity prices will move well before the policy response arrives.

Physical over paper. Vault over stock. Real assets over financial proxies.