The moat illusion – China’s bid to deflate US AI valuations
China’s low-cost AI models are narrowing the US AI moat, reshaping competition and raising questions over valuations, compute and the economics of AI scale.
The global AI order is taking shape as two partly competing governance and distribution systems rather than a single winner. In our view, the US-led Pax Silica is built around access to trusted chips, compute, and tech supply chains, while the China-led WAICO is premised on a broader offer of lower-cost, more open models and AI capacity-building.
China is running its familiar overcapacity playbook on its AI models, just like infrastructure, housing, solar panels or electric vehicles before. Because AI models cross borders digitally at zero marginal cost, the competitive impact of China’s AI build-out could be considerably larger than prior manufacturing cycles.
The US AI moat is shrinking unevenly as it quickly compresses at the model layer, holds up reasonably well at compute, and is increasingly taxed at home by state-level power and permitting friction.
Can the US AI moat hold? We assess where US advantages are eroding, where they remain resilient and what China’s low-cost AI push could mean for hyperscaler valuations.
Download the report for more detailed insights.
You might also be interested in
This report was brought to you by the Macro team
Reliable and consistent economic forecasts, analysis, models and scenarios provide the insight necessary to make informed decisions in a fast-changing world.
