The Hang Seng Tech "bear market" is really two stories — only one has broken math

The Hang Seng Tech Index has slid into a "technical bear market" again, roughly 20% below its October 2025 peak — the second time it has worn that label this year. Headline writers want the reader to conclude that Chinese tech is broken. It isn't, not as a single thing. "Technical bear market" is a price label, and it is doing two jobs at once: it is dragging the index's cash engines down alongside the corner of the market where the bear case is actually earned.

Start with what the label does not say. A technical bear market means nothing more than a 20% drawdown from a recent high; it says nothing about whether earnings broke. This particular reset erased the premium that liquidity front-running built into the index in early 2026. Price came down. Earnings are a separate question, and they answer differently depending on which half of the index you are measuring.

Where the panic is rational

The part of the complex that deserves its selloff is the hardware and AI segment, because China's own breakthroughs are the problem. Within a two-week stretch, the country put a homegrown DRAM champion on the market, moved a domestic deep-ultraviolet lithography tool into production, and released an open-weight frontier AI model. Each milestone is a source of national pride and an immediate threat to pricing power. The competition — an "involution" dynamic where every player races to add capacity and undercut — turns capacity races into price wars, and open-weight models turn tokens into a commodity export that generates little profit at the point of production. That is a structural margin problem, not a fixable ramp.

Baidu is the live exhibit of what a cheap multiple hides here. It trades at roughly 7.5x forward earnings, but only because near-term earnings are near zero: revenue is growing about 2% and operating margin is negative. A single-digit multiple with no earnings floor is not a value signal. It is the value-trap version of the index's hardest story.

The pocket the headline hides

The same index also holds profit engines that grow faster than their prices would suggest — and they have been pulled down with everything else. PDD trades at about 8.8x forward earnings while revenue grows 17%, returns on equity run near 23%, and free-cash-flow margin is about 24%. JD goes for roughly 7.6x forward earnings on double-digit revenue growth, a 3.8% dividend yield, and a price-to-sales ratio under 0.2. A double-digit grower at a single-digit earnings multiple is the divergence that flips the question from "should I buy China tech?" to "which part of this index is mispriced?" — so long as the margin erosion stays in the commodity-AI corner and does not leak into e-commerce operating margins.

That fence is the entire judgment. The distinction between a growing pain and a structural problem lives exactly at the boundary between the AI-commodity complex and the commerce-and-internet core. In February the index was down over 20% and the same cash engines were being offered at single-digit multiples; the rally back to the spring confirmed that falling prices and falling earnings need not be the same thing.

Buy the index and you buy both stories at once — which is why "buy the dip" is the lazy version of this trade. The useful read is the divergence. The drawdown re-rated the cash engines to below their own growth rates while the AI names fell for a reason. If deflation stays inside the chips-and-models segment, the "technical bear market" is a mislabeled valuation reset, and the risk is on the side of the owners selling PDD and JD at a discount to their own growth. If commodity pressure crosses into commerce margins, those cheap multiples become the same trap Baidu already sits in. The index's story has been one label; the decision inside it has always been two.