China's AI Blind Spot: Industrial Profits Cool to 4.2% in August

I was scrolling through Monday's data drop from China's National Bureau of Statistics, and one number jumped out at me. Industrial profits grew just 4.2% in August from a year earlier. That's not a disaster on its own. But it's the fourth consecutive month of slowing growth, and that tells a much bigger story.

The deceleration nobody wants to name

Here's the detail most coverage skips. China's cumulative profit growth, the year-to-date figure officials prefer to cite, peaked at 24.7% through April. By the end of August, it had eased to 15.7%. That's a nearly nine-point slide in four months. Granted, some of that reflects base effects from a strong 2023 comparison. But not all of it.

Meanwhile, the AI capex cycle is handing profits to everyone else. Nvidia's data center revenue keeps climbing past $30 billion a quarter. South Korea's memory chip makers are printing money on high-bandwidth memory demand. Japan's semiconductor equipment vendors are booked out for years. The US, Japan and South Korea are all riding the same wave, and China's industrial base isn't on the board.

Why? Export controls on advanced chips and lithography equipment. That's the blunt answer. Washington's restrictions have kept China from buying the most advanced Nvidia accelerators, and domestic substitutes from Huawei and others are catching up, but they aren't there yet. The question worth asking: can China build an AI stack fast enough to matter before this current boom cycle matures?

What this means beyond Beijing

Pull the camera back and the takeaway for markets is straightforward. The global AI trade has a clear geography, and China's mainland industrials aren't part of it. That matters for anyone holding emerging market exposure, commodity positions tied to Chinese manufacturing, or even just a broad index fund. If Chinese industrial profits keep decelerating, so does the case for a China re-rating this year.

It also matters for the memory and equipment supply chain. South Korean and Japanese suppliers are the biggest beneficiaries of the divergence. To be fair, that's been the trade for 18 months now, and it's crowded. But the gap in profit growth between China and its neighbors is widening, not narrowing.

Regular people feel this through jobs and wages in China's manufacturing hubs, and through prices on everything from EVs to solar panels. Chinese overcapacity in those sectors is a symptom of the same problem. Capital that might have gone into advanced computing went into legacy manufacturing instead.

My honest read

I'm not entirely convinced this is a permanent structural shift. China has a track record of playing the long game, and its domestic chip efforts are real, not just talking points. Huawei's Ascend line keeps improving. SMIC is squeezing older nodes harder than most skeptics expected.

But the timeline matters. The current AI profit cycle is rewarding whoever has the best silicon right now, in 2024 and 2025. China doesn't. That's a gap measured in years, not quarters, and no amount of industrial policy closes it overnight.

Color me skeptical that Beijing turns this around before the next earnings cycle. What I'm watching: September's print, due in late October, and whether cumulative growth holds above 15%. If it breaks below that, the deceleration narrative becomes something more uncomfortable. Time will tell, though.