Read More《云知声今起招股:获商汤等近亿港元基石认购，将成「港股AGI第一股」》
Positive Comments: Unisound’s IPO Provides a “Benchmark Sample” for AGI Commercialization, Driven by Both Technological Accumulation and Commercialization Capabilities to Release Value
Unisound’s listing on the main board of the Hong Kong Stock Exchange this time is not only a phased achievement of its 13 – year technological in – depth exploration and commercialization efforts but also marks the arrival of the first “benchmark for Hong Kong – listed companies” in the field of Chinese AGI (Artificial General Intelligence). From technological accumulation, commercial implementation to capital recognition, Unisound’s IPO has released at least three positive signals, injecting confidence into the industry’s development.
I. The First – mover Advantage in Technology and R & D Investment Build a “Moat” and Establish Discourse Power in the AGI Field
Unisound’s core competitiveness is first reflected in the forward – looking and continuity of its technological layout. As one of the earliest companies in Asia to commercialize large AI language models, its technological evolution path is clear: from being the first to apply deep learning to speech recognition in 2012, to building the Atlas AI infrastructure in 2016 (with a computing power of over 184 PFLOPS in the intelligent computing cluster and a storage capacity of over 10PB), and then to launching the Shanhai large model with 60 billion parameters in 2023. This full – chain R & D route of “algorithm + computing power + data” has enabled it to form a profound accumulation in fields such as natural language processing (NLP) and multi – modal interaction.
What is even more noteworthy is its “practical combat ability” in technological implementation. The Shanhai large model ranked first with 82.2 points in the MedBench (Medical Large Model Evaluation) and ranked first in science and second in liberal arts in the SuperCLUE evaluation. The results on these authoritative lists directly verify the industry adaptability of its technology. For example, the medical scenario has extremely high requirements for the accuracy of professional knowledge and logical reasoning ability of the model. The leading performance of the Shanhai large model means that it already has the potential to replace some manual decision – making in vertical fields.
II. The “Lighthouse Customers + Long – Tail Scenarios” Model Verifies the Feasibility of Commercialization, and the Resilience of Revenue Growth is Prominent
If technological leadership cannot be transformed into commercial value, it will ultimately become a “laboratory result”. Unisound’s commercialization strategy – “first bind industry – leading customers (lighthouse customers), and then use the scenario experience to support long – tail demands” – has been proven to be an efficient path. Its partners such as Gree (home appliances), Ping An Technology (healthcare), Peking Union Medical College Hospital (medical), and People’s Insurance Company of China (medical insurance) are all “benchmarks” in their respective fields. They can not only provide high – value scenario data (such as clinical cases in hospitals and claims cases in insurance companies) but also attract other customers in the same industry through the leading effect.
Financial data directly reflect the effectiveness of this strategy: from 2022 to 2024, the revenue increased from 601 million to 939 million, with a compound annual growth rate of 25%. The adjusted net loss rate narrowed from 30.5% to 17.9%, and the loss was significantly reduced. Especially in the context where the AI industry generally faces the contradiction of “high R & D investment and slow commercialization”, Unisound’s revenue growth rate and profit improvement ability provide a reference sample for the market on “how technology – driven companies balance investment and output”.
III. Endorsement by Top – Tier Capital and Industry Synergy Strengthen the Certainty of Long – Term Development
Unisound’s shareholder list is quite “luxurious”: it includes early – stage venture capital firms such as Qiming Venture Partners and Trustbridge Partners, industrial capital such as JD.com, Qualcomm, and 360, and strategic investors such as China Internet Investment Fund (a national – level fund) and CICC. Among the cornerstone investors introduced in this IPO, the participation of SenseTime (AI computing power and model ecosystem) and Runjian Co., Ltd. (communication and computing power infrastructure) has sent a clear signal of “industry synergy”. For example, SenseTime’s accumulation in the fields of AI computing power and multi – modal models may complement Unisound’s large language model. Runjian Co., Ltd.’s layout in communication networks and edge computing can provide a more efficient implementation carrier for Unisound’s AI solutions. This dual recognition of “capital + industry” not only guarantees Unisound’s capital reserve after listing but also broadens the boundaries of its technology application scenarios through resource integration.
IV. Supported by Industry Dividends, the High Growth of the AGI Track Provides Imagination Space for Valuation
According to Frost & Sullivan data, the market size of AI solutions in China is expected to increase from 180.4 billion in 2024 to 1174.9 billion in 2030, with a compound annual growth rate of 36.7%. The emergence of AGI will further stimulate the demand in various vertical industries. As the “first AGI stock in Hong Kong”, Unisound’s scarcity will enable it to obtain a higher valuation premium in the secondary market. At the same time, its market position in sub – fields such as consumer AI (third) and medical AI (fourth) also lays a foundation for it to seize shares in the high – growth track.
Negative Comments: Hidden Worries Behind Unisound’s IPO – Triple Tests of Technological Iteration Pressure, Profitability Challenges, and Market Competition
Although Unisound’s IPO is regarded as a milestone in AGI commercialization, from the perspective of the industry environment and the enterprise itself, its future development still faces multiple challenges. We need to be vigilant against the potential risks of “technological leadership ≠ market monopoly” and “revenue growth ≠ sustainable profitability”.
I. The Competition in the AGI Track is Intense, and the Sustainability of Technological Advantage is Doubtful
Unisound’s technological leadership is based on the foundation of “first – mover + focus”, but the competitive environment it faces has changed greatly. Internationally, giants such as OpenAI and Google DeepMind continue to launch large models with larger parameter scales and stronger capabilities (such as GPT – 4 and Gemini) with more abundant capital and data resources. Domestically, technology giants such as Baidu (Wenxin large model), Alibaba (Tongyi large model), and Huawei (Pangu large model), as well as emerging AI companies such as SenseTime and ByteDance, are all accelerating their AGI layout.
Taking the medical large model as an example, although Unisound temporarily leads in the MedBench, other players such as Tencent (Tencent Medical Large Model) and Ping An Technology (Ping An Medical Large Model) are also rapidly iterating. If Unisound cannot maintain the intensity of R & D investment (its R & D expense ratio in 2024 was not clearly disclosed, but it is generally above 30% for AI companies), or is surpassed in key technologies such as multi – modality and reasoning ability, its technological advantage may be quickly diluted.
II. The Profit Model has not been Fully Established, and Continuous Losses are still the “Sword of Damocles”
Although Unisound’s loss rate is narrowing (the adjusted net loss rate has decreased from 30.5% to 17.9%), it still has not achieved profitability as of 2024. The “money – burning” characteristic of AI companies is clearly reflected in it: on the one hand, the training and iteration of large models require continuous computing power investment (the maintenance cost of Unisound’s intelligent computing cluster is high); on the other hand, to expand the market, its sales and service expenses (such as the cost of customized solutions for lighthouse customers) may increase with the increase in the number of customers.
More importantly, its current revenue structure may hide a “growth ceiling”. According to the news, Unisound’s revenue mainly comes from providing AI solutions for vertical industry customers. This “project – based” model can quickly increase the volume, but there are two major problems: first, the gross profit margin of a single project is greatly affected by the bargaining power of customers (the gross profit margin decreased slightly from 40.5% to 38.8% in 2024, which may be related to this); second, the replicability of the solutions is limited. If customized services cannot be transformed into standardized products (such as the SaaS model), revenue growth will highly depend on the number of customer expansions, and the scale effect will be difficult to release.
III. The Risk of Customer Concentration and the Commercialization Efficiency of Long – Tail Scenarios Need to be Verified
Although Unisound’s “lighthouse customer” strategy has helped it quickly accumulate industry experience, it may also lead to a relatively high customer concentration. For example, if the revenue contribution ratio of its top five customers (such as Gree and Ping An Technology) is too high, once the demand of a leading customer shrinks or turns to a competitor, it will have an impact on its revenue stability. In addition, the logic of “covering long – tail scenarios through the experience of lighthouse customers” needs to verify the commercialization efficiency of long – tail scenarios. Although there are many long – tail scenarios, the payment ability of a single scenario is weak. If the development cost (such as model fine – tuning for each long – tail scenario) is higher than the revenue, it may lead to “diseconomies of scale”.
IV. The Valuation Fluctuation Risk of Technology Stocks in the Hong Kong Stock Market
The valuation of technology stocks in the Hong Kong stock market has become more rational. Especially after a round of speculation on AI concepts, investors pay more attention to the “actual profit realization ability” of enterprises rather than just the technology story. Unisound’s market value in this IPO ranges from HK $11.7 billion to HK $14.5 billion, corresponding to a price – to – sales ratio (PS) of 11 – 14 times for its revenue of 939 million in 2024 (about HK $1.04 billion), which is higher than the average PS of technology stocks in the Hong Kong stock market (about 5 – 8 times). If it cannot achieve profitability in the short term after listing or its revenue growth slows down, it may face the pressure of valuation correction.
Suggestions for Entrepreneurs: Insights into the Survival Rules for Entrepreneurship in the AGI Era from Unisound’s IPO
Unisound’s IPO is both a successful case and reveals the key points of entrepreneurship in the AGI field. Combining its experiences and challenges, entrepreneurs can focus on the following directions:
I. Technological Layout Should be “Focused + Iterative” to Avoid Blindly Pursuing “Big and Comprehensive”
Unisound’s success began with its focus on “interactive AI”, and its technological evolution has always revolved around “large language models + vertical scenarios”. For entrepreneurs, the technological threshold in the AGI field is high and the resource consumption is large. They need to clarify their core advantages (such as specific algorithms and understanding of vertical scenarios) and avoid blindly following the pursuit of surface indicators such as “parameter scale”. At the same time, they need to establish a “rapid iteration” mechanism – the competition of large models is essentially a competition of “continuous evolution ability”. They need to continuously optimize the model through customer feedback data to maintain technological leadership.
II. Commercialization Should be Driven by Both “Benchmark Scenarios + Standardized Products”
Unisound’s “lighthouse customer” strategy is worthy of reference, but entrepreneurs need to be vigilant against over – reliance on customized services. While binding leading customers, entrepreneurs should precipitate scenario experience into standardized products (such as API interfaces and industry SaaS platforms) to reduce marginal costs. For example, Unisound can encapsulate the diagnostic ability of its medical large model into a standardized API for small and medium – sized hospitals to call, rather than only providing customized solutions.
III. Capital Operation Should Balance “Strategic Synergy + Capital Efficiency”
Among Unisound’s shareholder structure, the participation of industrial capital (such as JD.com and Qualcomm) has brought scenario and resource support to it. When financing, entrepreneurs should give priority to introducing industrial capital that is synergistic with their own business, rather than only pursuing financial investment. At the same time, they need to strictly control the efficiency of capital use – AGI R & D requires continuous investment, but they need to avoid “burning money for the sake of burning money” and should focus the capital on links that “can be directly transformed into commercial value” (such as scenario implementation and product standardization).
IV. Be Vigilant Against the “Technological Leadership Trap” and Strengthen the Insight into Market and Customer Demands
Technological leadership is the foundation, but market demand is the ultimate orientation. Entrepreneurs need to avoid falling into “technological self – indulgence”. They should clearly understand the real needs of customers (such as “accuracy > generation speed” in the medical scenario) through in – depth interaction with customers (such as cooperation with lighthouse customers) and adjust the technological R & D direction accordingly. For example, Unisound’s strengthening of the professional knowledge reasoning ability in its medical large model is based on the insight into the actual needs of hospitals.
V. Build the Full – Chain Capability of “Computing Power + Data + Algorithm”
The competition in AGI is a full – chain competition, and computing power (intelligent computing cluster), data (industry scenario data), and algorithms (large model optimization) are all indispensable. Entrepreneurs need to layout computing power resources in advance (such as cooperating with cloud service providers and building small – scale clusters by themselves) and accumulate high – quality industry data through compliant means (such as signing data sharing agreements with customers). At the same time, they should focus on algorithm optimization (such as model compression and fine – tuning technology) to reduce the dependence on computing power.
Unisound’s IPO is an important node in the AGI commercialization process. It not only proves the feasibility of the “technology + commercialization” dual – wheel drive but also warns of the severity of industry competition and profitability challenges. For entrepreneurs, its experiences and lessons together form a “practical guide”. In the wave of AGI, only those with both technological determination, business acumen, and resource integration ability can go more steadily and further.
