Axios’s analysis is pretty spot-on. Beijing’s policy mix right now is indeed a bit tangled: on one hand, it’s aggressively pushing AI and capacity upgrades; on the other, on the demand side it’s still reluctant to “throw money around” in a big way.
This isn’t a new problem—it’s the AI-version of an old habit: production efficiency keeps soaring, while domestic demand remains the same. In the end, they still have to rely on external demand to absorb the excess capacity, but the global market has its limits.
I’ve seen a similar logic in the 2003–2008 cycle. Back then it was steel and cement; now it’s chips and compute power. The essence is the same: the supply side sprints ahead, the demand side lags behind, and finally it either has to dump exports overseas or the excess capacity can’t be digested domestically.
Now that the AI race is accelerating, this contradiction can only get sharper. You drive DeepSeek-level (and others’) competition to build world-class models, but domestic consumer purchasing power can’t support such a large industrial closed loop—so those capacities ultimately have to be pushed outward. The issue is that the US, Europe, and Japan are all pursuing supply chain rebalancing and tech decoupling, which is narrowing your export room.
So Axios is asking the right question: how much of China’s industrial capacity can the world actually absorb? This isn’t a technical issue—it’s an old bill for structural imbalance—AI just makes it due sooner.
This isn’t a new problem—it’s the AI-version of an old habit: production efficiency keeps soaring, while domestic demand remains the same. In the end, they still have to rely on external demand to absorb the excess capacity, but the global market has its limits.
I’ve seen a similar logic in the 2003–2008 cycle. Back then it was steel and cement; now it’s chips and compute power. The essence is the same: the supply side sprints ahead, the demand side lags behind, and finally it either has to dump exports overseas or the excess capacity can’t be digested domestically.
Now that the AI race is accelerating, this contradiction can only get sharper. You drive DeepSeek-level (and others’) competition to build world-class models, but domestic consumer purchasing power can’t support such a large industrial closed loop—so those capacities ultimately have to be pushed outward. The issue is that the US, Europe, and Japan are all pursuing supply chain rebalancing and tech decoupling, which is narrowing your export room.
So Axios is asking the right question: how much of China’s industrial capacity can the world actually absorb? This isn’t a technical issue—it’s an old bill for structural imbalance—AI just makes it due sooner.