At 9:30 a.m. on April 17, Huang Xiaohuang and Chen Hang sounded the gong at the Hong Kong Stock Exchange. Fifteen years earlier, they had returned to Hangzhou from the United States with Zhu Hao to start a business, enduring over a year without securing any external funding. Fifteen years later, Qunhe Technology became the first among the ‘Hangzhou Six Little Dragons’ to go public. On the day of its debut, Huang Xiaohuang mentioned ‘survive’ and ‘staying alive’ 11 times in interviews.

This mindset reflects Qunhe’s 15-year journey. Huang has witnessed numerous companies rise swiftly as industry darlings, only to fade into obscurity a few years later. Over time, Qunhe has explored GPU cloud rendering, O2O, home improvement SaaS, Industry 4.0, internationalization, and now spatial intelligence. Huang describes the company’s strategy as ‘fortifying defenses and advancing step by step,’ prioritizing longevity over short-lived brilliance.

The capital market’s initial enthusiasm, however, was anything but cautious. Qunhe priced its shares at the upper limit of HK$7.62, with the Hong Kong public offering oversubscribed roughly 1,591 times. On its first trading day, the stock closed at HK$18.6, up 144%, and continued rising the next day. By April 21, the share price peaked at HK$48.5, valuing the company at over HK$80 billion—a more than sixfold increase from its issue price in just three trading days.

Over the following three months, the stock nearly erased all those gains. On July 30, Qunhe fell to a low of HK$7.4, breaking its issue price for the first time, with a maximum retracement exceeding 80% from its peak. On September 7, inclusion in the Stock Connect triggered a noticeable rebound, but the stock soon stabilized around HK$10. During this period, Qunhe neither lost 80% of its clients nor 80% of its decade-plus accumulation of 3D data. What truly shifted was market perception. Before listing, narratives around the ‘Hangzhou Six Little Dragons,’ ‘first spatial intelligence stock,’ and ‘Physical AI’ created a cohesive story. Post-IPO, investors began scrutinizing quarterly financials: what had converted into revenue, what remained technical reserves, and what belonged in current versus future valuations.

The ‘Six Little Dragons’ moniker quickly answered ‘who might this company become,’ but listed Qunhe Technology now faces the existential question of ‘who am I.’

Qunhe’s first interim report post-IPO juxtaposes two timelines. In the first half of the year, revenue reached RMB 405 million, up 1.5% year-on-year; gross profit was RMB 336 million, with an 83% margin; adjusted net profit was RMB 55.42 million, up 210.9% year-on-year. Under IFRS, the company still reported a loss of RMB 146 million, primarily due to redeemable liabilities, share-based payments, and IPO-related expenses.

Qunhe can generate profits but has yet to reignite growth. Subscription services contributed RMB 393 million in revenue, up 0.8% year-on-year, accounting for 97.1% of total revenue. This primarily comes from Kujiale and its overseas version, Coohom—businesses validated over the past decade.

Kujiale, a cloud-native 3D space design software launched by Qunhe in 2013, remains the company’s most critical cash cow. Designers and home furnishing firms use the platform for floor plan design, 3D modeling, rendering, and whole-house customization, converting design data into production drawings. It has evolved from a rendering tool for designers to an integrated workflow solution for home furnishing companies.

As of June 30, Qunhe had 48,617 enterprise clients, a net increase of 1,633 since the year’s start; 411 large clients contributing over RMB 200,000 in annual revenue accounted for 46% of total revenue.

However, another metric better explains the sluggish 1.5% revenue growth: overall Net Revenue Retention (NRR) fell from 100.3% a year ago to 94.4%; enterprise client NRR dropped from 101.6% to 96.4%. Only large clients remained above 100%, at 101.8%. This indicates Qunhe can still acquire new clients and retain large ones, but revenue from existing clients has stopped growing organically.

Setting spatial intelligence aside for a moment, Qunhe today is a company with clear characteristics: high gross margins, emerging profitability, nearly 50,000 enterprise clients, but a core SaaS business entering a low-growth phase. Clearly, the HK$48.5 price tag wasn’t just valuing this company; investors were also paying for another—a spatial intelligence firm.

Qunhe has been preparing for this transition for years. Around 2018, it began exploring spatial data to train AI models; after the rise of large language models, it prioritized spatial understanding, generation, and robot training. Now, it offers SpatialLM, SpatialGen, and SpatialVerse for robot training and simulation, along with the AI video product LuxReal launched this year.

These products have started generating revenue. In the first half, ‘new AI applications and products’ contributed RMB 31 million, up 177% year-on-year. Orders for Kujiale’s AI-powered design platform exceeded RMB 20 million; SpatialVerse orders reached RMB 6.8 million; LuxReal’s monthly revenue surged over 470% from May to June.

While RMB 31 million is a promising start, it accounts for just 7.7% of total revenue. This figure includes both AI-enhanced design within Kujiale’s existing business and the standalone AI video product LuxReal. The professional services segment, more directly tied to robot training and spatial simulation, generated RMB 11.81 million, or 2.9% of total revenue.

Thus, Qunhe faces a classic valuation mismatch. As a SaaS company, it boasts an 83% gross margin and positive adjusted profits, but 1.5% revenue growth hardly aligns with typical high-growth tech expectations. As a Physical AI infrastructure firm, its decades of 3D data, GPU infrastructure, and spatial models are tangible assets, and AI revenue is growing rapidly—just not yet enough to reshape the company’s revenue structure.

This dilemma reflects a broader trend in the software industry over the past year. Following significant stock adjustments, investors have refocused on cash flow and genuine AI commercialization. Companies like Salesforce and Snowflake now proactively disclose the ARR, client counts, and revenue contributions from their AI products. For Qunhe, the question is even more specific: When will institutional investors move ‘spatial intelligence’ from a terminal value on their Excel sheets to a present-day reality?

Huang Xiaohuang has set a clear benchmark. In April, he stated his next goal is to truly transform Qunhe into a spatial intelligence company. Success won’t require new technical metrics—just when half of revenue comes from spatial intelligence. In the first half, this figure was RMB 31 million, with a long road ahead.

When Qunhe bet on spatial intelligence, a crucial detail was later obscured by stock volatility: initially, even Qunhe doubted it.

This relates to Huang’s operating philosophy over 15 years. In 2011, after leaving NVIDIA, he aimed to move GPUs to the cloud for 3D rendering. For over a year, Qunhe struggled to secure funding and took on projects while searching for applications. Later, IDG showed interest in home improvement, and Kujiale launched in late 2013. Business grew in 2014, and by 2015, it was profitable. Back then, Qunhe also discussed O2O; Huang later recalled that ‘not talking about O2O meant no investment’ in those years.

Kujiale initially hoped to attract renovation users first, then sell furniture and facilitate transactions. But before reaching the second step, the O2O financing environment shifted around 2015. Qunhe pivoted quickly: instead of waiting for advertising and transactions, it directly charged designers and renovation firms for software, making SaaS its core business for the next decade.

The first major crisis came around 2018 when a group of O2O home improvement clients served by Qunhe collapsed due to internet finance risks, with eight of the top ten clients going under. Meanwhile, large internet firms entered the home improvement sector, expanding from e-commerce and supply chains to design software while aggressively poaching talent, nearly halting some operations.

Qunhe didn’t confront them head-on; instead, it shifted focus to industrial software, handling furniture production drawings and manufacturing scheduling, while accelerating overseas expansion. Huang set extreme scenarios: if design software was lost entirely, manufacturing would remain; if the Chinese market was lost entirely, overseas markets would remain.

Qunhe was accustomed to leaving itself an escape route—a habit that saved it but also caused it to miss opportunities. In 2018, Qunhe had already begun researching spatial AI. However, by around 2020, while Kujiale and real estate-related businesses thrived, AI struggled to find clear commercial paths. Internally, doubts arose about whether this was just technologists’ self-indulgence. At its peak, Huang recalled, only a few remained committed to AI, while hundreds worked on real estate and construction.

This changed with ChatGPT and Copilot in 2023. Huang began worrying that AI would erode the value of process-oriented SaaS. They ruled out general-purpose language models and image generation—areas where giants would inevitably dominate—and chose spatial models, where Qunhe could leverage its decades of 3D data, rendering capabilities, and industry expertise. In 2024, Qunhe significantly increased its spatial intelligence investment—a rare move in Huang’s career: proactively pushing a profitable company toward an unproven business.

Resistance soon emerged. Huang recalled that some investors opposed the transition; algorithm talent was scarce; the sales team worried clients would reject a shift from SaaS subscriptions to usage-based pricing; internally, some questioned why a company like theirs needed to train its own models when agents would suffice.

The ‘Hangzhou Six Little Dragons’ label made Qunhe’s need for this shift more concrete. Among the six, Qunhe might have been the one most reliant on this tag to facilitate its identity transformation. By then, DeepSeek had gained fame with its models, Unitree had robots, and Game Science had Black Myth: Wukong. In contrast, many still knew Qunhe primarily for Kujiale. The gap between a ‘home improvement design SaaS’ and ‘spatial intelligence’ was vast and required explanation.

The ‘Six Little Dragons’ label bridged this distance. Huang later said that after joining the group, opposition to spatial intelligence quickly faded. In 2025, Qunhe received nine times as many resumes from C9 universities as the previous year and 20 times more overseas applicants; the spatial intelligence algorithm and model team grew from fewer than ten to over 60, with annual computing power investment reaching tens of millions of yuan.

Thus, if we must quantify what Qunhe gained from the ‘Six Little Dragons’ label, the first benefit wasn’t its stock price but an organizational revaluation. A long-term project Huang once had to repeatedly explain to employees, investors, and candidates suddenly gained an externally understandable reference point. Spatial intelligence shifted from a niche technologist pursuit to a frontier technology a ‘Hangzhou Six Little Dragon’ company was expected to invest in.

This was the label’s most tangible benefit, but it came at a cost. Previously, the market evaluated Qunhe as a SaaS company: Are clients staying? Are renewals strong? When will it turn a profit? After becoming the ‘first spatial intelligence stock,’ investors began asking: When will spatial intelligence generate significant revenue? Can SpatialVerse scale? When will these RMB 31 million in new business revenue become RMB 300 million and eventually surpass Kujiale?

The ‘Six Little Dragons’ helped Qunhe answer ‘why pursue spatial intelligence’ but also pushed the question of ‘when will it succeed’ into the spotlight earlier than expected. This is the most intriguing aspect of Qunhe’s relationship with the label. Without it, Qunhe might have taken longer to convince talent, employees, and the outside world why a home improvement SaaS company should train spatial models; with it, the identity shift accelerated, but so did the pressure to prove the new identity quickly.

Unitree quickly became the second case study. On August 19, Unitree Technology listed on the STAR Market with an issue price of RMB 150.8, opening at RMB 1,100, up 629%, and reaching a market cap of RMB 444.9 billion; the stock then quickly retreated, falling about 45% from its peak in a short time.

Of course, one cannot simply compare the declines of Unitree and Qunhe, as their issuance mechanisms, free float, and investor structures differ entirely. However, both companies serve as a reminder: Scarcity can rapidly push prices into the future, but the public market will eventually dissect the future item by item. This recalculation has even begun affecting later entrants.

On September 9, The Information disclosed that following Unitree's tumultuous market debut, domestic regulatory authorities have set stricter criteria for the listings of certain humanoid robot firms. These companies are now required to demonstrate sustainable revenue streams, a reduction in losses, or possess clearly defined technological innovations. Just a day afterward, the Financial Times reported that regulators had recently cautioned investment banks to enhance the quality of their IPO projects and adopt a more prudent approach when pricing new stocks.

Consequently, the 'Hangzhou Six Little Dragons' have embarked on a fascinating transformation. Initially just a media moniker, it is now evolving into six distinct entities, each necessitating individual assessment. And Qunhe was merely the first to undergo this detailed scrutiny.

Huang Xiaohuang had patiently awaited this listing for five long years. In 2021, Qunhe initially planned to go public in the United States, even scheduling the gong-ringing ceremony, but ultimately scrapped the plan. Two years later, Huang and Zhu Hao journeyed to New York to seek funding. Zhu once calculated that after years of entrepreneurial endeavors, had they remained at Amazon and NVIDIA, their financial returns might have been even greater. By 2026, Qunhe finally made its entry into the public market. However, upon its arrival, the manner in which capital markets valued tech companies had already undergone a transformation.

The mantra of 'forecast the future first, then await profits' is not a novel concept born out of the AI era. Earlier, Uber endured over a decade of losses to capture cities, drivers, and users, finally reporting its first full-year operating profit of $1.11 billion in 2023, after incurring a loss of $1.83 billion the previous year.

Yet, today's public market is less inclined to accept 'scale will inevitably lead to profits' as a sufficient response. Internet companies once had a decade to demonstrate network effects, but AI firms, upon listing, must almost immediately address more specific inquiries: What proportion of current revenue stems from the new narrative? Can gross margins be sustained? When will the next revenue stream genuinely manifest on the profit and loss statement?

Several other firms are also headed down this path. Bloomberg reported in January of this year that BrainCo had confidentially filed for a Hong Kong IPO. On September 9, Reuters, citing individuals familiar with the matter, stated that DeepSeek had engaged CITIC Securities to prepare for a listing on the STAR Market, with intentions to commence the process within the year. Once these companies enter the public market, the moniker 'Hangzhou Six Little Dragons' will retain its value, aiding ordinary investors in swiftly comprehending why a company merits attention.

Qunhe has simply preemptively delineated the limits of this label. While labels serve to make a company visible, the income statement ultimately determines its worth post-visibility. In April of this year, Huang Xiaohuang set a target for Qunhe, aiming for spatial intelligence to eventually account for half of its revenue. Four months later, in its inaugural financial report post-IPO, revenue from new AI applications and products stood at 31 million yuan, out of a total revenue of 405 million yuan.

Thus, for Qunhe, the next truly pivotal re-pricing is unlikely to coincide with the emergence of another spatial intelligence concept. It will transpire when the 31 million yuan figure no longer requires separate emphasis and explanation.

*The title image and illustrations featured in the text are sourced from the internet.