The embodied intelligence track, valued in the trillions, is witnessing a fierce contest between genuine value and inflated valuations. This piece examines Unitree's post-listing share price volatility to understand the short-term versus long-term value proposition for robotics companies.

Before Unitree's listing, one investor had a plan: with an issue price of 150.80 yuan, if the opening price stayed below 600 yuan, he would buy; if it exceeded that, he would wait. On August 19, Unitree opened at 1,100 yuan, surging 629%, leaving him stunned. By afternoon, the price fell to around 800 yuan; still deeming it expensive, he hesitantly purchased 200 shares, unwittingly becoming what his friends joked was a "bag holder." That day, Unitree closed at 845 yuan with a massive 231.60 billion yuan in turnover and an 85.28% turnover rate. The next day, it closed at 687 yuan, down 18.7%; on the third day, it settled at 672.41 yuan. As of the fourth trading day, its market cap has lost over 190 billion yuan from its peak.

"It's insane, absolutely insane, and no one around me understands it," one fund manager remarked, referencing Unitree's peak market cap of 440 billion yuan. "The money has gone mad; those trapped deserve it." In his view, a reasonable valuation range would be 100-150 billion yuan. Unitree founder Wang Xingxing's stoic expression on listing day has resurfaced as a hot topic. At the bell-ringing ceremony, surrounded by smiling guests, Wang's lips were tight, his face devoid of any emotion. Why wasn't this 90s-born newly minted billionaire happy? A day later, netizens concluded from the declining share price, "He saw it all coming, which is why he couldn't smile."

Wang appears unfazed. On August 20, he appeared at the World Robot Conference in Beijing, sharing his thoughts on technology commercialization. He also extended his timeline for the "ChatGPT moment" in embodied intelligence. A year ago, he predicted it would arrive in 1-2 years at the earliest and 3-5 years at the latest. Now, he's revised it to 2-3 years at best and 5-10 years at worst, making him one of the most conservative forecasters at this year's conference. For comparison, Galaxy General's founder Wang He predicts this GPT moment will arrive by 2028, while Xinghaitu's founder Gao Jiyang doubts there will be a sudden, ChatGPT-like proliferation.

On the other hand, they may be the ones hoping most for Unitree's stock to remain strong. The embodied intelligence sector is approaching a concentrated wave of IPOs. Tencent Technology has learned that nearly 50 robotics companies are preparing for listings, with about 10 ontology manufacturers planning to file within the year. A CEO of a robotics company valued at over 10 billion yuan told Tencent Technology that, similar to how NIO's stock price fell after its IPO and hindered Xpeng's subsequent fundraising, the leading unlisted embodied intelligence companies are currently valued at 20-30 billion yuan. If Unitree, as the "anchor," drops, it would be detrimental to subsequent projects.

While everyone acknowledges that the industry's current fundamentals can't support the capital market's valuations, Unitree's evaporation of 190 billion yuan in market cap within days has starkly presented two challenges to the sector: How can a hardware company that's reached the secondary market continue its embodied intelligence narrative? And to what extent will companies with inferior scarcity, revenue, and profits be affected? On August 24, the fourth trading day post-listing, Unitree opened at 640.26 yuan. As of press time, the latest price is 615 yuan, with a market cap of 249.11 billion yuan.

Trading Scarcity Before Sufficient Float

On listing day, when the market cap hit 440 billion yuan, a robotics practitioner excitedly posted on social media: "The anchor is set, we can rest easy, let the music play on." Unitree's listing was one of the most anticipated events in China's capital market in the second half of 2026. The company issued only 40.446 million new shares, representing 10% of the post-listing total share capital, with no secondary offering of existing shares. The total post-listing share capital stands at 404.464 million shares. Its online subscription hit rate was 0.01809759%, even lower than ChangXin Technology's. A staggering 9.78 million households participated in the subscription, with valid subscriptions reaching 53.637 billion shares, translating to an 8,288-fold oversubscription.

Prior to listing, Nomura initiated coverage with a "Buy" rating and a target price of 370 yuan. Compared to the issue price of 150.80 yuan, this target implied over 100% upside. This target was based on a projected 25 times price-to-sales ratio for 2027; Nomura estimated Unitree's revenue for 2026, 2027, and 2028 at 2.687 billion yuan, 5.396 billion yuan, and 13.184 billion yuan, respectively. However, on day one, trading reflected a different story: the first humanoid robot stock, a scarce asset, limited float, and an unmissable tech frenzy.

By the close of August 19, divergence between buyers and sellers had widened. Unitree's price retreated from 1,100 yuan to 845 yuan, slashing its market cap from 444.9 billion to 341.8 billion yuan. The following day, trading volume expanded, closing at 687 yuan. A US-dollar VC partner who has invested in multiple robotics companies told Tencent Technology that with shares not yet fully unlocked and the float and tradable shares relatively unchanged, price fluctuations may not reflect true operating conditions. "Short-term market cap shouldn't be the primary basis for judging a company's long-term value." In his view, Unitree's market cap holds little reference value until the lock-up period expires, making its fundamentals more critical.

Bridging the Gap Between Aspiration and Reality

Humanoid robot valuations with hardware company revenues—this is Unitree's current predicament. Its revenue, profits, and operating cash flow are among the top in the industry. From 2023 to 2025, operating revenue grew from 159 million yuan to 1.699 billion yuan; during the same period, gross margin on main business rose from 44.22% to 60.13%. In 2025, net profit attributable to the parent was 278 million yuan, with non-GAAP net profit at 591 million yuan and net operating cash flow at 670 million yuan. The revenue mix is shifting rapidly. In 2023, quadruped robot revenue was 119 million yuan, accounting for 75.78% of main business revenue, while humanoid robot revenue was a mere 2.9671 million yuan, or 1.88%. By 2025, humanoid robot revenue surged to 868 million yuan, comprising 51.78%, with quadruped revenue at 698 million yuan, or 41.62%.

Within two years, Unitree transformed from a company primarily selling quadruped robots to one where humanoid revenue surpasses quadruped. The company reports that in 2025, it shipped over 5,500 pure humanoid robots, excluding wheeled dual-arm robots; cumulative quadruped sales from 2023 to 2025 exceeded 33,000 units. The critical question, however, is who is buying the humanoids. According to Unitree's prospectus, in 2025, scientific research and education accounted for 73.60% of humanoid revenue, with industrial applications at 9.01%. The primary application scenarios listed include research, education, application development, teaching, cultural performances, and intelligent services. Universities, developers, and tech companies purchase robots, generating revenue and helping expand the company's software and hardware ecosystem. Factory deployments, however, require calculations on production line takt time, failure rates, maintenance costs, labor replacement, and repurchase rates.

In Q1 2026, Unitree reported revenue of 423 million yuan, a 68.49% year-over-year increase, but non-GAAP net profit attributable to the parent fell 52.55% to 40.2536 million yuan, and net operating cash flow dropped 85.65% to 34.3996 million yuan. The company attributes this to faster growth in R&D and sales expenses, with cash outflows for purchases and period expenses outpacing inflows from sales. It projects H1 2026 revenue between 1.052 billion and 1.128 billion yuan, a 35.62% to 45.41% increase year-over-year, with non-GAAP net profit attributable to the parent expected to be between 236 million and 283 million yuan, a decline of 6.43% to 21.97%.

Unitree's net proceeds of 5.917 billion yuan from this offering are earmarked for intelligent robot model R&D, robot ontology development, new intelligent robot product development, and manufacturing base construction. These projects address the most capital-intensive aspects of a humanoid robot company: models, hardware, product iteration, and capacity building. The prospectus highlights risks including slower-than-expected commercialization and demand growth for general-purpose robots, technology failing to meet downstream needs, short-term heat from rentals fading, intensified competition leading to proactive or passive price cuts, and production line overhauls and R&D investment falling short of expectations, all of which could impact revenue growth, gross margin, and profits. Overseas sales previously accounted for over 40% of main business revenue. The company warns that foreign trade, tariffs, technology export restrictions, and exchange rate fluctuations could affect this segment. For a company reliant on hardware shipment growth, overseas sales are not just a growth driver but also a matter of supply chain, compliance, and channels.

In comparison, UBTech, listed in Hong Kong, posted 2025 revenue of 2.001 billion yuan, a net loss attributable to the parent of 703 million yuan, R&D expenses of 507 million yuan, and negative operating cash flow of 784 million yuan. As of August 21, its market cap was approximately 42.235 billion HKD. Several analysts covering the robotics sector point out that Tesla remains the key driver for the entire humanoid robot industry. The rationale is analogous to the EV market: pure electric vehicles only entered mainstream life after Tesla's Model 3 scaled up, which shifted domestic perceptions and paved the way for other brands. Against this backdrop, the sector's trajectory hinges on Tesla's ability to set expectations: "If Tesla fails to open up that expectation, with its own stock price falling and industry progress consistently missing forecasts, Chinese companies will find it hard to establish an independent upward trend."

Impact Exists, But Perhaps Not as Severe as Feared

Several investors believe Unitree is a closely watched listing sample this year, but its "first stock" status means its capital market performance shouldn't be directly equated with the broader embodied intelligence sector's health. A fund manager told Tencent Technology that he believes Unitree's reference value for subsequent embodied intelligence IPOs isn't as significant as outsiders think. "It will certainly serve as a benchmark, but a benchmark isn't a pricing method." He outlined different scenarios. First, later entrants may benchmark against Unitree and discount their own valuations based on technical barriers, production capacity, and revenue quality. The issue is that many followers lack Unitree's profitability, shipments, and brand premium; if Unitree continues to fall, the benchmark drops, and discounts only worsen. Alternatively, later entrants might forgo high premiums from the start, pricing themselves based on their own revenue, losses, customers, and delivery cadence rather than trying to replicate Unitree's debut day price.

The aforementioned investor told Tencent Technology that the secondary market quickly prices in macro liquidity, interest rates, economic conditions, external markets, financial performance, and future expectations. In contrast, the primary market is influenced by available capital scale and the number of investable tracks, meaning the two won't always move in lockstep. The heat in the embodied intelligence primary market isn't solely driven by short-term technical progress from individual companies—investable tracks like AI, robotics, and commercial aerospace are relatively concentrated, and institutions still need to deploy capital. As long as no major new track emerges to divert funds, the primary market may not immediately cool due to short-term secondary market weakness. However, he cautioned that secondary market performance does transmit to the primary market: when listed companies perform well, unlisted peers find fundraising easier; conversely, subsequent financing and valuation expectations in the primary market face pressure. Overall, he believes high primary market valuations are reinforced by fundraising success, backing from renowned institutions, and market attention. If high-valuation projects aim for further valuation increases, they must demonstrate technological breakthroughs, product advancements, or commercialization proof.