China in 2025: why we believe this is the year the most unloved trade re-rates — and why DeepSeek just lit the fuse

While Western investors panic over DeepSeek and debate China’s investability, we see something different — a textbook monetary stimulus cycle at the moment of maximum pessimism. This is where asymmetric opportunities are found.

The past week has been instructive. DeepSeek’s R1 release sent Nvidia down 17% in a single session and triggered a broad reassessment of AI infrastructure spending in the US. The narrative has shifted fast. But while attention is fixed on what this means for American tech, we think the more important question is what it signals about China. 

What the market was missing

China equities are structurally underowned by Western institutional capital entering 2025. Three years of regulatory crackdowns, property sector stress, delisting fears, and geopolitical tension have pushed allocators to near-zero positioning. The prevailing view remains: China is uninvestable.

We have learned to pay close attention when consensus becomes this uniform. Maximum pessimism rarely coincides with maximum risk — it more often marks maximum opportunity. And right now, something important is changing beneath the surface that most Western allocators have not yet priced.

In late 2024 Beijing signaled a material policy shift — PBOC reserve requirement cuts, fiscal stimulus measures, and an explicit prioritisation of growth. We began positioning into that shift when the market was still dismissing it. We believe 2025 is the year that positioning proves correct.

PBOC money printer, historic valuation gap, asymmetric setup

Monetary stimulus cycles in China have historically produced lagged but powerful equity responses — particularly in technology and consumption. The transmission is slower than Western markets, typically 6 to 18 months. That lag means the full effect of late 2024 stimulus is still ahead of us. We are in the early innings of what we expect to be a multi-quarter re-rating.

What makes this setup particularly compelling is the valuation gap it is occurring against. Chinese technology names are trading at a fraction of comparable US platforms on every fundamental metric. The discount is not justified by fundamentals — it reflects a sentiment and positioning overhang that has become completely disconnected from business reality.

Policy shift plus extreme valuations plus a stimulus cycle whose full effect is yet to transmit — this is the setup. We expect Chinese technology to be one of the best performing asset classes of 2025. That view was contrarian when we formed it. It is becoming less contrarian by the week.

China tech and AI infrastructure — and why DeepSeek changes the urgency

Our primary expression is Chinese technology — specifically names where the valuation discount to US peers is at or near historic extremes, and where the domestic AI buildout provides a second, independent catalyst. China’s AI infrastructure investment — data centers, model deployment, cloud computing, and domestic semiconductor development — is accelerating with explicit state backing. This is not yet priced into Chinese equities the way the AI narrative has been priced into US names. We see a double opportunity: valuation mean reversion and an AI re-rating that is only beginning.

Seven days ago DeepSeek released R1 — a model matching frontier US AI performance at a fraction of the compute cost. Nvidia fell 17% in a session. The Western consensus read it as a threat to AI infrastructure spending.

We read it as the single most powerful confirmation of our China thesis we could have hoped for. DeepSeek proves that China’s AI capability is real, advancing faster than anyone expected, and increasingly independent of US hardware constraints. Efficient models do not mean less infrastructure — they mean faster adoption, wider deployment, and more demand for the cloud, data center, and energy infrastructure that runs them. That infrastructure will be built in China, by Chinese companies, with state backing. We added to our positions on the dip.

Real risks — but already in the price

We are not dismissing the risks. Geopolitical escalation, renewed regulatory action, ADR delisting pressure, and stimulus failing to transmit into real economic activity are all live scenarios. We hold these in mind actively.

But at current valuations, we believe they are more than reflected in the price. Alibaba trading 72% below its 2021 peak, with accelerating cloud revenue and a refocused management team, is not priced for a recovery — it is priced for continued deterioration. We do not believe that is the right base case for 2025, and nothing we have seen changes that view.

The principal risk we are managing is timing. We do not expect an immediate re-rating across the board. Stimulus cycles transmit over 6 to 18 months. We have sized these positions for a 12 to 24 month horizon and are communicating that clearly. We are not trying to catch a quarter — we are positioning for the full move.

Structural Repositioning