This article provides a logical analysis and does not constitute any investment advice. The stock market carries risks, and investment must be approached with caution.
The STAR Market has recently welcomed two hard-tech giants that have captured market attention: ChangXin Memory represents the independent breakthrough of domestic memory chips, while Unitree Technology is being hailed as the "first stock of humanoid robots." A closer analysis reveals fundamental differences in their valuation logic.
How to Understand the Valuation Gap
ChangXin Memory had a static price-to-earnings (PE) ratio of 308.92 times at its IPO, which appears extremely high. This is because the DRAM industry is a classic strong cyclical sector, and 2025 finds it in a trough. Low profit margins have created this inflated static PE; however, shifting to the expected profits for 2026 when the cycle is expected to rise, the dynamic PE quickly falls to the 20-30 times range. ChangXin Memory operates in a mature, multi-hundred-billion-dollar existing market. There is rigid demand for memory in smartphones, servers, and automotive chips. Downstream customers are real, orders are tangible, and the logic of import substitution is clear, with large-scale commercial shipments already underway.
Unitree Technology, which is about to be subscribed, has an IPO price of 150.8 yuan and a total market capitalization of approximately 60.9 billion yuan after issuance. This corresponds to a static PE of 219.23 times for 2025, far exceeding the industry average of 38.56 times, with a price-to-sales (PS) ratio of 35.89 times, significantly higher than comparable companies. Some institutions have set a target market cap of over 100 billion yuan, pricing the grand narrative of a future humanoid robot boom into the current stock price. A striking comparison: ChangXin Memory had a total market cap of 3.28 trillion yuan on its first trading day, with a PS ratio of only 11 times. Unitree Technology, in the still-validating humanoid robot track, already has a PS ratio more than three times that of ChangXin Memory at the issuance stage. The market is applying a more generous valuation to an emerging track that has not yet been commercially adopted on a large scale than to a mature cyclical leader.
Why Just These Two Companies?
The biggest chasm between the two comes from the stark difference in market certainty. ChangXin Memory faces a massive existing market. DRAM is a necessity for modern electronic devices, and regardless of technological iteration, the demand for memory is objectively long-term. Its challenge is the cycle of fluctuations, with risks more from price volatility, competition from overseas giants, and the pace of catching up with advanced technology. Its business model has been repeatedly validated by the global market.
Unitree Technology is in the humanoid robot track, which is a future market where demand needs to be created. A significant portion of its current revenue comes from university research procurement and B-end prototype testing. Consumer-grade and large-scale industrial applications are still far from reality, and there is no widespread rigid demand for humanoid robots. While its 332% revenue growth in 2025 was impressive, its growth rate slowed to 35%-45% in the first half of 2026. R&D and sales expenses have continued to rise, further squeezing profit margins.
Behind the ultra-high valuation, the market implicitly assumes three strong conditions: a near-term explosion in the humanoid robot industry; Unitree Technology can maintain its leading position against competitors like Tesla's Optimus, Xiaomi, and Zhiyuan Robot; and its gross margin will not significantly decline due to industry price wars. Only if this perfect scenario fully materializes can the high valuation be supported. If any one of these conditions falls short, the massive valuation will lack a fundamental backing.
History Repeating Itself
The A-share market has repeatedly seen similar scenarios. Many high-PE tech new stocks on the STAR Market, initially driven by scarcity and thematic hype, see their prices surge. However, as time passes and earnings fail to match the expensive valuation, it is not uncommon for their stock prices to fall below the IPO price within 6-12 months.
Risk and Opportunity
It is important to clarify a misunderstanding: not denying the long-term prospects of the track does not mean accepting a current price that has already discounted three to five years of future performance. The future potential of humanoid robots may be vast, but industrial implementation takes time, and every step—technological iteration, cost reduction, and scenario cultivation—is filled with uncertainty. The capital market, however, tends to discount ten years of imagination into today's stock price.
ChangXin Memory's valuation trap comes from the cycle: using trough-period profits makes the PE look very expensive, while using boom-period profits makes it seem relatively reasonable. Unitree Technology's valuation trap comes from long-term discounting: treating a future industrial explosion that has not yet happened as a realized reality. ChangXin Memory still needs to be wary of cyclical reversals, while Unitree Technology faces the greater uncertainty of whether the industry can truly develop.
From a trading perspective, Unitree Technology has very few shares available for online allocation, and its floating share percentage is low. This makes it easy for the stock to be speculated on and driven up shortly after listing, further detaching the short-term stock price from its true fundamentals. Ordinary investors can easily be attracted by the profit effect and buy at the peak.
Across the entire A-share tech sector, similar phenomena are not uncommon. Many tech companies do not lack technology; they lack the sustained performance to match high valuations. The market is keen on chasing new concepts and "first stocks," willing to pay a high premium for stories, but often forgets: no matter how great the track, stock prices must ultimately be digested by earnings. Hard-tech investment should be about capturing gains from technological breakthroughs, import substitution, and industrial implementation, not simply from thematic sentiment. ChangXin Memory still has the risk of cyclical ups and downs. For an early-stage humanoid robot target, a PE ratio over 200 times means investors have already bought up future growth. If industrial implementation is slower than imagined, the potential for a valuation crash is equally massive.
"The vast sea of stars in technology is certainly alluring, but when the tide goes out, only real revenue and profit are the most solid moats for stock prices." For ordinary investors, facing hot tech IPOs with PE ratios of over 100 times, it is necessary to both see the industrial opportunity and respect the valuation bubble. The future direction will ultimately be revealed by time. A clear opinion: it is recommended to subscribe; the stock should not be heavily sold in the short term; it is not recommended for speculation; the risks outweigh the opportunities.