As we step into 2024, it is undeniable that generative artificial intelligence (AI) stands as a dominant force within the global technological landscapeThe rapid advancements in AI computational power and supportive hardware are propelling the commercialization of AI applications at an unprecedented pace.
Recently, the US Department of Defense added SenseTime, a prominent Chinese AI firm, to its "Chinese Military Companies List." Shortly after, on January 7, SenseTime publicly expressed its strong disagreement with the decision, labeling it unfoundedThe impact of this list primarily restricts included entities from becoming suppliers for the US Department of Defense, which SenseTime asserted would have no significant effect on its businessThe company aims to engage in proactive discussions to secure fair treatment and preserve the interests of its shareholders.
This incident reflects a critical shift in the AI domain, where the gap between different nations, particularly the US and China, is narrowing swiftly
Advertisements
According to recent reports by Japanese media, China's capabilities in AI research are rapidly catching up to those of the United StatesAn analysis conducted on the top research papers adopted by three major international societies revealed that four of the top ten institutions, based on author rankings, hail from China.
Positioning itself as the sole major AI cloud platform independent of internet giants, SenseTime has emerged as a heavyweight contender in China’s AI arenaSoon after its public stance regarding the US Department's actions, SenseTime unveiled its latest achievement: the "SenseNova," a novel generative AI model.
As SenseTime races toward the AI 2.0 era, its trajectory embodies what the future holds for the industry at large.
Reflecting on the past decade, SenseTime has witnessed significant transformations in the global AI landscape and has been part of China's AI sector's ascendance since its inception as one of the earliest AGI startups.
The storm of generative AI took the world by surprise in 2022, and SenseTime quickly aligned itself with this burgeoning trend
Advertisements
In April 2023, SenseTime led the domestic market by launching the “SenseNova” architecture, undergoing several iterations over the following year, currently reaching version 5.5.
This reflects SenseTime's nimbleness in responding to advancements in algorithm capabilities, demonstrating quick adaptation and robust learning aptitudeNotably, the performance of the SenseNova 5.5 model aligns with multiple key indicators comparable to GPT-4o and is regarded as one of the leading domestic models equivalent to GPT-4 Turbo.
These developments indicate that the disparity among leading AI entities concerning large-scale model capabilities is diminishing, attributed to China's heightened focus on infrastructure development alongside the technological accumulations of AI firmsOver the past decade, SenseTime has equipped itself with a comprehensive AI technology stack, encompassing hardware infrastructure, AI cloud services, models, and applications.
Nonetheless, amid technological advancements, a pragmatic dilemma looms over SenseTime: it has yet to achieve profitability since its IPO, putting sustainable viability of its business model under scrutiny
Advertisements
Consequently, this has adversely affected the company's stock performance.
In response, Xu Li, Chairman and CEO of SenseTime, articulated in an interview that the company holds a vital responsibility as a public entity towards its shareholdersTherefore, advancing towards AGI necessitates the development of mature commercial monetization models.
Consequently, over the last year, SenseTime underwent a series of strategic transformations, including layoffs, the formulation of new strategies, and organizational restructuring.
In terms of organizational structure, SenseTime emphasizes adaptability, resilience, and rapid transformationIn December 2024, Xu announced that SenseTime completed its "1+X" structural reorganizationThe internal communication indicated that the company would focus on its core business within the AI cloud and generalized visual models, while its endeavors in smart transportation, home robotics, smart healthcare, and smart retail would be spun off into independent entities, each with its own CEO.
Strategically, SenseTime has embarked on a tripartite new direction encompassing large infrastructure, extensive modeling, and application
- Nasdaq Futures Down 1.15%
- Chain Hotel Price Surge Lasted One Year
- Sales Know Best the Intensity of the Car Price War
- Unicorns Seek Breakthroughs Amid Controversy
- Challenges Ahead for the U.S. Stock Market in 2025
By positioning itself as the most knowledgeable service provider for computational power in large models and vice versa, SenseTime aims to become a major player in the AI landscape.
These strategic shifts send a clear message: SenseTime intends to concentrate on its core business represented by generative AI, betting that this transformational wave in technology can foster sustainable profitability and robust growth prospects for the company.
The rapid advancement of generative AI requires no embellishment, yet it is essential to recognize that the future landscape of artificial intelligence is one marked by global competition and uncertaintyEach player in this arena faces the delicate balance between technological investments and commercial returnsHow SenseTime navigates this terrain and fosters a positive cycle between technology and business remains a question with significant implications.
In unraveling this complexity, three keywords emerge, highlighting the core mission.
Solution lies in the tangible
The tripartite strategy illustrates SenseTime's clear vision: extensive infrastructure will support large models, which in turn will receive increased investment, ultimately empowering practical applications.
This vision aligns with current market trendsIn 2024, the commercial prospects for AI applications are gradually gaining more attention, prompting market participants to anticipate the emergence of more large-scale model applications.
From this rationale stems the remarkable growth of companies like Applovin, which saw a nearly sevenfold increase in shares, and Palantir, tripling in valueSuch instances underscore the AI application's pivotal role in catalyzing a surge in technology stocks, particularly as the three major US indices witnessed historical highsConversely, the top ten performers in the Hong Kong stock market reveal a lower representation of "AI" driven companies
As international AI applications explode, the market eagerly awaits the value release from domestic AI applications.
In this context, after a lackluster year in 2024, SenseTime anticipates leveraging its strategic framework to realize revenue growth and valuation resurgence by 2025.
Initially, the sheer computational power is paramount.
In the realm of AI, the significance of computational power cannot be understated; it directly impacts the training and inference of AI models and their practical applicationXu Li noted that the shift from traditional AI 1.0 to generative AI 2.0 has transitioned the cost structure from "labor-intensive R&D" to a "computational resource-intensive" model.
Thus, SenseTime's substantial infrastructure underpins its new tripartite strategy, capable of delivering robust computational services essential for training and operationalizing large models, thereby accelerating the commercialization of AI applications.
As one of the earliest companies to commercialize intelligent computing centers in China, SenseTime has established a solid foundation in AI infrastructure, which serves as a pillar for advancing generative AI application
By the end of 2024, SenseTime has deployed 46,000 GPUs, surpassing 23,000 petaFLOPS of total computational capacity, achieving interlinked cluster training capabilities with 90% acceleration efficiency, showcasing abundant computational resources alongside superior engineering capabilities.
Secondly, large models are a critical focus.
SenseTime has consistently maintained a position among the top players in domestic large model researchOn January 10, the company officially launched the “SenseNova” integrated large model, which has achieved native modal integration, markedly enhancing deep inference and multi-modal information processing capabilitiesAccording to the latest report by SuperCLUE evaluating domestic large models for 2024, the SenseNova integrated large model scored 68.3, tying for first place in the country with DeepSeekV3.
The introduction of the SenseNova model signifies a substantial breakthrough in native modal training, challenging the current common separation of large language models and multi-modal models, effectively initiating the realization of unified models.
Lastly, the vast application ecosystem is vital.
Just as a building comes alive once inhabited, SenseTime's generative AI "edifice" is rapidly taking shape
Within this framework, both B2C and B2B users are paramount for successful commercial deployment and the establishment of a closed-loop ecosystem.
In the ongoing process of applying generative AI technology, SenseTime has enhanced service delivery for clients such as Kingsoft, Haitong Securities, Xiaomi, and Yuedu Group, culminating in a remarkable revenue of 1.051 billion yuan for its core generative AI business in the first half of 2024, a staggering year-on-year growth of 255%.
With enhanced AI capabilities and decreased inference costs, SenseTime’s AI applications are poised for increased product penetration and user base expansion, consequently driving the company’s revenue growth.
In summary, the trifecta of substantial computational power, large models, and an expansive application ecosystem delineates SenseTime’s competitive advantage in the generative AI sphere
This strategic focus should ideally accelerate the release of commercial potential through its significant ecological positioning within the sector.
In conclusion, following the upheavals of 2024, SenseTime is set to bask in a prosperous year, marked by synergistic benefits from policy initiatives, industrial dividends, and technological advancements in the forthcoming year.
The Central Economic Work Conference in December 2024 spotlighted "AI+", advocating for large-scale application demonstration initiatives featuring innovative technologies, products, and scenariosConcurrently, over a dozen provinces and municipalities have enacted specific plans for the advancement of artificial intelligence, promising to maintain an accelerated pace in the industrial application of large models across various sectors by 2025.
As the era of AI applications reaches critical mass, an array of both consumer and commercial scenarios will inevitably undergo transformative remolding by AI