The Dark Matter of AI in China
China's AI programme didn't start with DeepSeek. It started in 2016, with Demis Hassabis. R1 was the visible flame. The wildfire is AI for materials.
Companion essays where the analytical call gets made before the evidence is fully in. Speculative by design. Predictions still graded publicly.
China's AI programme didn't start with DeepSeek. It started in 2016, with Demis Hassabis. R1 was the visible flame. The wildfire is AI for materials.
UK coal. US petroleum. China sovereign electricity. The supply side of the first electrostate, assembled.
A note on first principles. Why the Smil-Rosling synthesis is the analytical engine of Plain Sight Research.
My quarterly prediction was a disaster. The quarter itself was the strongest forward guidance Eddie Wu has ever given.
Chinese miners expanded copper 16% and gold 43% during the supercycle. Western majors contracted 18–60%.
Permanent structural surplus by 2028–2029. Five demand-destruction forces the consensus underestimates.
Three answers to the coal question. The third one — coal-to-olefins — is the thesis nobody in Western energy markets has priced.
A $295B megacap trading at 40–45% of fair value. The algorithm reads collapse; it's actually a pivot.
The FYP as deployment order, not speculative bet. Why TFP is China's new organizational scorecard.
China export consensus misses 2024–2026. The same structural misforecast the IEA made on solar, applied to trade surplus.