[Red packet] 7.13 Pre-market: July’s mainstream domestic AI compute power?

High-end AI technology is currently mostly in a correction. After a big rally on Thursday, Friday’s correction is relatively severe: the ChiNext Index (创指) turned bearish by engulfing the prior day’s candle directly. After Zhao “Yi” released earnings on Thursday night, Friday saw a gap up, then a fade, accompanied by heavy volume and a large bearish candle. [Taoqiu Ba]

For high-position stocks with outstanding earnings, it’s easy for capital to realize profits. There are plenty of stocks that have risen tenfold or even dozens of times. After the big move, they often use good news to distribute.

Meanwhile, for low-position stocks, the feedback after earnings has been quite good. Once low-position earnings stocks are released, their price action generally isn’t bad—for example: Ruijie Network, Xinwang Ruijie, and after the earnings of Founder Lianxin (浪潮信息), there were one-word limit-ups: two consecutive limit-ups on Thursday, and on Friday it surged to near the limit-up again.

High-position stocks: good news comes, and capital uses the opportunity to quantify and sell/distribute in volume. Low-position stocks: good news comes, and capital strengthens, with quant funds pulling. This is essentially a matter of where the stock sits—capital rotates between “high” and “low.” This is the style the market has shown this week. Next, we’ll see when the high-position names start to have reduced volatility, and whether the low-position names can break out with sustained momentum.

Based on Friday’s stock moves so far, the big-cap stocks’ charts are basically similar—especially in the afternoon—mostly one-way declines, with intraday patterns that look nearly identical. The dominant forces are institutions and quant funds; institutions are also mostly quant programs right now. When they fall, there’s no real follow-through or support—just a one-way drop. Compare that with Thursday afternoon’s rally—also one-way up. In a quant-dominated backdrop, it’s better to think more about quant behavior. As for the logic, the market’s basics are essentially transparent; the question is whether you believe early or believe late.

Low-position directions: Domestic compute power: servers, switches, supernodes, compute leasing, AIDC, etc.; innovative drugs; brokerage; and the liquor segment is universally hated (万人嫌). For the mid-year results: after the earnings of Shede (舍得) come out, check whether it continues to fall—if it doesn’t and instead rises, it could also come for an oversold rebound from below.

Technology is still the mainstream right now. With capital rotating from high to low, low-position names will catch up.

On domestic industrial lines—for example, wafer fabs, semiconductor materials, and semiconductor equipment—this is a logic that benefits from capacity expansion, and their positioning isn’t low either. This direction is worth separating out on its own.

Founder Lianxin (浪潮信息): After earnings that beat expectations, it responded with a one-word limit-up, hit another limit-up again on Thursday, and on Friday surged high then pulled back with a volume expansion! The feedback is very strong. The earnings-line varieties relative to low-position names received nearly three limit-ups, and they’re also large-cap—within this week’s market environment, this kind of feedback is quite remarkable. It’s in sharp contrast to Zhao “Yi” and the like! The key difference is that it’s low-position—before, the market hadn’t overly “blasted” it. For this year’s stocks that can rise tenfold or even ten-plus times almost at the drop of a hat, this provides room for catch-up and炒作 (speculative) play.

Uni-President? (紫光股份): On Friday, its intraday performance was relatively resilient compared with most stocks in the market—i.e., it was dragged down only because of the index. Tomorrow, we’ll see whether it merely catches down (补跌) or whether it can stay strong. If domestic compute power doesn’t end, it won’t end—once domestic compute power dies, this one is easy to “get killed” (噶).

Right now, the only area where the market is relatively strong is domestic compute power—plus switches, servers, and supernodes. Even if we continue to see adjustments in the short term, I still look positively ahead: in July, continue to call on domestic compute power. For low-position AI applications, AIDC, and innovative drugs: sector rotation rallies showed better performance. For commercial aerospace, we’ll see tomorrow whether it can strengthen—otherwise it’s just another rotation expectation. That’s pretty much it~