ChangXin IPO Imminent, South Korean National Team Enters: Where Else Can Capital Flow in the Memory Supply Chain?
TechFlow Selected TechFlow Selected
ChangXin IPO Imminent, South Korean National Team Enters: Where Else Can Capital Flow in the Memory Supply Chain?
Are there any segments that haven't fully risen yet?
Navigating Web3 tides with focused insightsAuthor: David, TechFlow Research
In the past couple of days, positive news for storage has come one after another.
On June 29, South Korea launched a semiconductor super plan totaling over 1000 trillion Korean won (approximately 650 billion USD), with an official goal to double DRAM capacity within five years;
At the same time, China's DRAM leader CXMT has passed the regulatory review, with the market expecting it to list between mid-July and early August, and institutional valuations reaching 2 trillion to 4 trillion yuan; coupled with the judgment that major storage manufacturers will face shortages until 2028, the reasons to be bullish on storage and semiconductors have never been as clear and aligned as they are now.
At the same time, this sentiment has even spilled over to foreign networks.
Well-known tech investment blogger Jukan (@jukan05) on X posted that the direction most worth betting on in the second half of 2026 might still be Chinese semiconductor self-sufficiency targets.
Citing communications with Chinese sell-side analysts, he stated that CXMT's market cap after IPO will be at least 5 trillion RMB, and most of the raised funds will flow into stocks related to domestic semiconductor self-sufficiency, thus believing that targets like ACMR and NAURA still have prospects.
However, rushing in along with this wind may not always be the right timing.
Currently, the combined market cap of nearly 30 CXMT concept stocks in the A-share market has exceeded 1.9 trillion yuan, and most leading stocks upstream and downstream of the industrial chain are generally hovering near 52-week highs; blindly rushing in is certainly not the optimal solution.
After one round of gains, there are not many segments left where the market hasn't fully priced in expectations.
Prices Are Rising, Sales Volume Has Hardly Moved
US semiconductor research institution SemiAnalysis broke down a set of data in its June 23 report "China's CXMT Is Set to Challenge DRAM Incumbents":
CXMT's bit shipments in Q1 2026 grew only 11% quarter-over-quarter, while the Average Selling Price (ASP) increased by about 57% quarter-over-quarter. Bit shipments calculate the actual volume sold based on storage capacity, measuring "how much was sold"; ASP measures "how expensive it was sold".
Putting these two numbers together means that this quarter CXMT hardly sold more goods; rather, the goods were sold at higher prices.
Therefore, SemiAnalysis's judgment is that CXMT's profit explosion in this round relies on the industry cycle itself, rather than breakthroughs in technology or market share.
In the current price trend, the first to reap profits are the original manufacturers directly selling chips: Samsung, SK Hynix, Micron, and CXMT itself. Their profits expand linearly with ASP, and they are also the group that has surged the most fiercely in the past year.
SK Hynix's stock price once rose over 350% this year. But having risen to this point, expectations for the original manufacturer chain have been priced in very fully: Samsung and SK Hynix currently have expected P/E ratios of only 3 to 5 times, which seems cheap, but behind this is the market having already priced in AI-driven demand and profits from 2026 to 2027 in advance.
The market has basically accepted the profits realized through price increases. The same applies to that batch of storage original manufacturers and module stocks in the A-share market; the gains are already large, and there is not much room to squeeze in. As for whether the upstream expansion chain (such as equipment, materials) has also been bought to the same level, we will look at the data later; no conclusion here for now.
Overall, the author feels that CXMT stands at a subtle crossroads:
On one side, selling DRAM and profiting from price increases makes it a beneficiary of this cycle; on the other side, using the 29.5 billion yuan raised from the IPO to expand production, buying equipment and materials from upstream, makes it the spending client.
In the past month or two, a voice has been popular in the market, which is to directly "buy CXMT upstream and downstream"; in hindsight, blindly going all-in did indeed generate huge returns, but at the current position, if judging that the storage and semiconductor market trend will continue, actually still need to figure out a few things,
First, where exactly in the industrial chain is the target you are eyeing located, and how does it benefit;
Second, is its current price still at the foot of the mountain, or has it already climbed to the halfway point or even the summit?
Let us answer the first question first.
"CXMT concept stock" is an overused label. Breaking it down, the demand driven by CXMT splits into two paths; the beneficiaries are not the same group of companies, and the realization time is also different.
The first is the ordinary DRAM expansion chain. Currently, 99% of CXMT's shipments are ordinary DDR and LPDDR; of the 29.5 billion raised in the IPO, over 22 billion is specified for equipment procurement for wafer production lines and technology upgrades. This money first goes to front-end equipment, that is, machines that make chips, which is the largest part of expansion investment; after the production line runs, materials are continuously consumed.
Representatives of the equipment segment are NAURA (002371), AMEC (688012), Piotech (688072), Hwatsing Technology (688120), ACMR Shanghai (688082); the material segment is Anji Microelectronics (688019), Jiangfeng Electronics (300666), Yoke Technology (002409), NSIG (688126). This chain consumes the money CXMT is spending now, with the highest order certainty.
The second is the HBM chain, not the same group as the first. HBM is high-bandwidth memory used by AI servers, with technology a notch harder than ordinary DRAM. CXMT's HBM is still catching up; the production line will not start production until the end of 2026, one position later than ordinary DRAM expansion. More critically, the value of HBM is not in front-end etching and deposition, but in the packaging segment, that is, stacking, bonding, and molding multiple layers of chips. So the beneficiaries of HBM are another group of companies:
Jingzhi Da (688627) for test equipment, Huahai Chengke (688535), Lianrui New Material (688300), Shanghai Sinyang (300236) for packaging materials, and SJ Semi (688820), Tongfu Microelectronics (002156) for advanced packaging and testing.
Upstream and Downstream, Is It Too High to Stand the Cold?
Laying out the targets of the two chains above, and arranging them according to the position of the current stock price relative to the highest point in the past year, one can intuitively see the traces of capital sweeping through; the following data is as of intraday on the 29th.
There is a clear dividing line in the table.
For ordinary DRAM equipment and materials, these two segments are almost all hovering near 52-week highs, mostly within 3% of the one-year high. On June 29, Hwatsing Technology hit the limit up directly and set a new historical high, Yoke Technology also refreshed its high, and AMEC, Anji, NSIG all had gains around 10%.
This segment is the market-recognized "CXMT expansion shovel sellers", with the hardest logic and highest certainty, so capital has bought it most fully. In other words, the certainty of the expansion chain has already been written into the price.
Lagging behind is the HBM packaging segment.
Lianrui New Material is still about 18% away from the 52-week high, closing down against the trend on June 29; SJ Semi is also about 18% away from the high; testing and packaging factory Tongfu Microelectronics is about 9% away from the high. The gap between them and equipment/materials is not because they are cheaper or ignored, but more because the realization rhythm is later:
CXMT's HBM production line will not start production until the end of 2026; the orders and performance of these companies will only be truly released after the production line runs and yields climb. Their current position is lower, corresponding to "not yet their turn"; simply viewing it as "finding bargains" may incur time costs and opportunity costs.
As for the overall conclusion, it is actually already clear.
The so-called "buy CXMT upstream and downstream", at this position is no longer a question of whether to get on the bus, we have to see whether they are actually at the summit.
Looking at price alone, the equipment and materials on the ordinary DRAM side are basically all standing at the highest position in the year, there are no cheap entries left; the small segment of HBM packaging is slightly lower in position, but the premise is that you are willing to wait.
In addition, price only answers half of "whether it is expensive". The other half cannot be explained by price itself; it depends on whose money is buying and selling at this position.
Hot Money is Supporting, Some Are Retreating
The essence of this layer is that pricing power has shifted from long-term capital looking at fundamentals to short-term hot money gambling on sentiment.
On one side, industrial capital, the National Integrated Circuit Fund, and the National Team are systematically reducing positions at high levels; on the other side, hot money and retail investors are rushing in borrowing the heat of the AI theme. The former perhaps understands this business best and is selling; the latter wants to buy low and sell high and is buying.
First look at the side that is withdrawing; we have done a round of organization from public information:
- GigaDevice actual controller Zhu Yiming reduced holdings by about 6.33 million shares from May 11 to 25 (company announcement). He is also the founder and chairman of CXMT, the person who should be most bullish on the CXMT industrial chain, reduced holdings of his own related leader at high levels.
- National Integrated Circuit Fund has continuously reduced holdings of NSIG since January, cashing out about 3.882 billion yuan cumulatively by early June (company reduction announcement).
- Montage Technology, Hygon Information, Piotech and other multiple semiconductor weight stocks have had shareholders reduce holdings after stock price surges since 2026 (each company announcement).
- The National Team (Central Huijin) made high-level reductions on broad-based ETFs such as CSI 300 (according to "Caijing" calculation of Central Huijin holdings and fund circulating shares). It is not reducing semiconductors itself, but a counter-cyclical withdrawal from the entire market high, and semiconductors are exactly the block with the largest gain in this round and the one most due for realization.
Source:Fulai Index Investment's Xueqiu Column
The motives for these reductions cannot be generalized. The Big Fund itself has an exit cycle, and reducing holdings to recover funds when shares expire is a routine action; the National Team adjusting broad-based holdings may be counter-cyclical rebalancing rather than bearish on a certain sector; industrial capital and executive reductions also have different reasons.
Interpreting them as "collectively bearish" is definitely a bit excessive, but one thing is certain:
At the current price level, these original long-term holders have unanimously chosen to realize some profits. Regardless of the motive, the action itself conveys the message that the current price has reached a position where long-term capital is willing to cash out.
Now look at the side that is buying.
According to market data cited by Sina Finance, the main force speculating on technology in this round is hot money:
Northbound capital added about 400 billion within the year, and the margin financing scale expanded to about 2.8 trillion. This kind of money bets on themes and momentum; whether the target is near 52-week highs or PE is high does not affect their entry; what they buy is the rise itself.
Regulators also noticed the overheating, stepping on the brakes by raising margin financing margin ratios and suspending trading to check consecutive limit-up stocks; chip ETFs frequently suspended trading because exchange prices were far higher than net value, with premiums once exceeding 30%.
Putting buyers and sellers together, the conclusion is not complex:
At the current price level, long-term capital related to storage and semiconductors is realizing profits in batches, and short-term hot money is taking over. The marginal pricing power at this position is increasingly falling into the hands of capital gambling on sentiment.
Note, this does not mean the market trend immediately peaks. Storage shortages continue until 2028, and CXMT expansion is real money. It's just that most institutions' judgment is that the sector will digest profit-taking through violent fluctuations, entering a ranking competition that values performance realization more, rather than a trend decline.
The Author's View
Combining the two layers of judgment and the capital surface above, I think the tendency is this.
Short-term (around CXMT listing), sentiment still has inertia. The heat of new share subscriptions, concept fermentation on the listing day, may give the sector another surge, but this is the last segment driven by sentiment; the later it goes, the more it becomes a relay game, and the risk of chasing highs is greater than the return.
Mid-term (watching CXMT HBM production line realization), what is truly worth waiting for is the differentiated realization of the ordinary DRAM expansion chain and the HBM chain. Equipment and material orders depend on the speed of CXMT's capital expenditure landing after listing; those HBM packaging stocks with lower positions need to wait until the production line starts production and yields climb at the end of 2026 before it is their turn; lying in wait now is gambling on time.
Several Signals to Watch:
- DRAM spot price turning point. The foundation of this round is price increases; once spot prices turn downward, original manufacturers and modules eating the price increase will react first.
- CXMT's capital expenditure rhythm after listing. Order realization mostly depends on these data; if money is spent fast, equipment and materials can sustain high valuations.
- Whether reductions by industrial capital and the Big Fund expand. Smart money continuing to withdraw is a leading signal of sentiment ebbing.
- Whether chip ETF premiums can converge. If premiums do not retreat from high levels, it indicates that those taking over are still retail hot money.
Cheap entries are no longer available; what remains are conditional opportunities, and the conditions may be implied in these several signals.
This article is for information organization and viewpoint analysis; involved individual stocks, ratings, and target prices are from public sources and are time-sensitive, constituting no investment advice. The market has risks; decisions need to be judged by oneself.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@chaoxiangooo
