[The AI Boom Is Real. The Story We Tell About It Is Not.]

China’s customs data for September 2026 landed like a verdict: exports up 36.3 percent year-on-year in dollar terms, imports up 31 percent, both figures blowing past analyst forecasts of roughly 25 percent and 24.5 percent respectively. [1] A trade surplus of $938.11 billion accumulated over the first nine months, on track to shatter last year’s $1.2 trillion record. [1] The number that made economists reach for their calculators was $447.18 billion in monthly exports — the highest ever recorded. [1] The headline explanation arrived with the data: artificial intelligence, according to the customs authority. Customs Vice Minister Wang Jun pointed to electronic components and computer hardware as the key pillars supporting the trade boom. Semiconductor exports have routinely posted year-over-year increases exceeding 100 percent. Integrated circuits, memory devices, the unglamorous intermediate goods that make machine learning possible — these drove the surge.

That explanation is accurate. It is also incomplete in a way that reveals something uncomfortable about how we talk about AI.

What the numbers actually show

The trade figures describe a manufacturing story, not an intelligence story. When the customs authority cites AI as a driver, it means the physical substrate: chips, circuit boards, servers, the hardware that data centers consume by the warehouse-load. The demand is real. The shipments are real. South Korea took 39.1 percent more Chinese goods than a year earlier; Taiwan, 37.3 percent more. Both are deeply embedded in the Asian tech production web where components become systems.

But notice what the AI narrative does here. The AI narrative transforms a straightforward industrial export boom into evidence of technological supremacy. The same numbers could be described as “China’s factories are running hot” or “global hardware demand is strong.” Instead they become a story about AI reshaping the global economy, a phrase that implies something about intelligence, about capability, about a future arriving on schedule.

The gap between what AI appears to be doing and what it actually does starts here, in the language we wrap around trade statistics.

The deception of the obvious explanation

Exports to Germany jumped 31.4 percent. Imports from Germany rose 0.1 percent. Exports to the European Union rose 25.3 percent; imports from the bloc edged up 0.7 percent. The asymmetry is stark, and it has drawn repeated complaints from Brussels and other capitals about the imbalance between Chinese exports and imports.

Here is where the AI explanation becomes a kind of misdirection. Yes, chip demand is surging. Yes, the hardware buildout for AI infrastructure is enormous. But the trade imbalance with Europe has almost nothing to do with AI and everything to do with structural differences in manufacturing costs, industrial policy, and domestic demand. When the customs authority highlights AI as the driver, it is not lying. It is selecting the most flattering frame.

This is the pattern that deserves scrutiny. AI gets credited with economic effects it only partially causes, because “AI” is a more compelling story than “manufacturing cost differentials” or “weak European demand.” The technology becomes a narrative device, a way of explaining outcomes that have multiple, less exciting causes.

Where the mechanism breaks down

Chips must be fabricated, packaged, tested, shipped. Servers must be assembled, cooled, powered. The demand for these goods reflects the buildout of physical infrastructure, not the deployment of autonomous reasoning systems.

When we conflate the two, we make a category error with consequences. We attribute to AI the economic effects of capital expenditure. We treat hardware shipments as evidence of software capability. We assume that because the infrastructure is being built, the intelligence it will host must be arriving.

The trade data does not tell us whether AI systems are becoming more capable. It can only tell us that more hardware is moving across borders. Those are different claims, and the gap between them is where the narrative diverges from the evidence — not deliberate, not malicious, but structural.

The surplus that reverses the story

A trade surplus approaching $1 trillion is not necessarily a sign of technological triumph. It is a sign of imbalance, and imbalances may invite correction. The European Union has repeatedly urged Beijing to address the gap. The United States, which took 9.8 percent more Chinese goods in September, has its own history of responding to such figures with tariffs and restrictions.

The AI narrative makes this correction harder to see coming. If the boom is driven by AI, and AI is the future, then the boom is inevitable and benign. But if the boom is driven by manufacturing capacity and cost advantages, then it is vulnerable to policy, to shifting supply chains, to the very trade measures that a $1 trillion surplus makes more likely.

The data point that reverses everything is not in the export figures. It is in the imports. China’s imports from Germany rose 0.1 percent. From the EU, 0.7 percent. The country selling so much to the world is buying almost nothing from its largest trading partners. That asymmetry may not persist indefinitely. When it breaks, the AI story may not explain what happened.

What we miss when we let AI explain everything

The September trade data is real. The AI-driven demand for chips and components is real. The $447.18 billion in exports is real. What is not real is the implication that these numbers tell us something about artificial intelligence as a cognitive technology, rather than about the hardware that AI runs on.

They tell us about supply chains. They tell us about industrial capacity. They tell us about the physical requirements of building data centers at scale. These are important things. But they are not the same as intelligence, and treating them as evidence of intelligence is the quiet deception at the heart of the AI boom narrative.

The customs authority cited AI as a driver of export growth, pointing to stronger demand for intermediate goods such as integrated circuits and memory devices. That is true in the narrow sense that AI infrastructure requires hardware. It is misleading in the broader sense that it suggests the technology itself is delivering on its promises. The hardware is being built. The intelligence may still be arriving.

Sources

1. Yahoo Finance — Portal copy