12/05 2024 374
In the new era of AI ushered in by large models, the technology industry is undergoing unprecedented changes.
Recently, Xu Bing, co-founder and executive director of SenseTime, publicly stated that we may be at the best starting point in history, amidst perhaps the largest wave of technological change ever. SenseTime must become more adaptable, resilient, and capable of rapid transformation.
These words presage a major internal restructuring underway at SenseTime. On December 4, SenseTime announced on the Hong Kong Stock Exchange that it had completed a strategic organizational restructuring to better address the opportunities and challenges presented by the AI 2.0 era.
Specifically, SenseTime has established a new '1+X' structure. Here, '1' represents the group's core business, focused on building an industry-leading AI cloud that seamlessly integrates large devices, foundational models, and AI applications; 'X' represents the ecosystem enterprise matrix, encompassing smart cars ('Jueying'), household robots ('Yuanluobo'), smart healthcare, smart retail, and more. Each ecosystem enterprise will be led by an independent CEO responsible for business development and will benefit from more flexible incentive mechanisms and financing channels.
This restructuring signifies that SenseTime will focus more on generative AI, embarking on a risky but promising bet on the future.
Behind the strategic organizational restructuring lies SenseTime's quest for a better commercialization solution
SenseTime's big moves have long been evident.
At the 2024 SenseTime 10th Anniversary International Forum, Xu Li, chairman and CEO of SenseTime, stated, 'We are currently at a turning point for AGI. In the era of AI 2.0, SenseTime's core strategy is to seamlessly integrate digital infrastructure, AI models, and applications.'
Simultaneously, SenseTime's recent large-scale layoffs were clearly part of this adjustment. Multiple insiders and media reports indicated that the layoffs affected various business departments, including security, autonomous driving, healthcare, and research. In response, SenseTime stated, 'SenseTime is actively promoting strategic transformation, focusing on key businesses and strategic growth areas of "large devices - large models - applications" and making corresponding optimizations and adjustments to its organization and talent structure.'
From the layoffs, it is evident that SenseTime is determined to transform its strategy, which is also related to its current predicament.
The half-year report revealed that SenseTime achieved revenue of 1.74 billion yuan in the first half of 2024, a year-on-year increase of 21%, but still incurred a loss of 2.46 billion yuan. In fact, profitability has been a persistent challenge for SenseTime since its listing.
Image source: Baidu Stock Market News
Due to these challenges, SenseTime's share price has remained subdued. However, this year, the company recognized the enormous commercial potential of generative AI, leading to a significant rebound in its share price. In the first half of the year, SenseTime's generative AI revenue reached 1.051 billion yuan, a year-on-year increase of 256%, accounting for 60% of the group's total revenue.
Benefiting from strong market demand for large models and intelligent computing services, the leap in capabilities of the 'SenseNova' large model, and the rapid decline in inference costs, model invocations have increased rapidly, driving a substantial growth in SenseTime's generative AI business revenue, making it the group's dominant business.
From an industry perspective, as the large model industry accelerates its implementation, global AI application companies have exceeded expectations, with their share prices continuing to perform strongly. For example, Palantir's third-quarter revenue exceeded expectations and revised its full-year guidance upwards; AppLovin's net profit increased by 300% year-on-year; and Shopify's third-quarter revenue increased by 26% year-on-year. The continuous fulfillment of performance under the support of AI large models has become an industry trend.
Against this backdrop, it is easy to understand SenseTime's repositioning of its strategic direction and core areas, accelerating its comprehensive transformation and upgrading towards generative AI.
With the completion of this restructuring, SenseTime can focus on its core business represented by generative AI, driving the group's business growth, profitability, and financial stability, providing enduring momentum for technological innovation.
At the same time, each ecosystem enterprise within the group can independently advance financing matters while sharing the group's infrastructure and foundational model development achievements, achieving better development and 'lightening the load' for the group.
Notably, SenseTime's ambition is hidden within its layout from '1' to 'X'. The company is fully committed to generative AI as its primary business and leveraging additional capital to develop ecosystem enterprises, thereby accelerating in the AI 2.0 era.
Opportunities for Positive Commercialization Cycles in the AI 2.0 Era
With the emergence of ChatGPT, AI has transitioned from the 1.0 to the 2.0 era. Human understanding and perspective on AI have suddenly broadened, amplifying commercial imagination.
Generative AI has brought about significant changes in cost structure, triggering changes in production models for AI companies.
Xu Li, chairman and CEO of SenseTime, believes that in the AI 1.0 era, companies primarily invested in the wisdom of R&D personnel. However, in the AI 2.0 era, the investment in computing resources has become crucial. SenseTime's strategy is to achieve seamless integration of large computing devices, large models, and applications, driving model optimization through application-driven models and ultimately reducing the production and usage costs of large models.
This is known as the 'trinity strategy' of 'computing infrastructure - large models - applications.' In Xu Li's words, it is about being the most knowledgeable computing service provider for models and the most knowledgeable model service provider for computing in the AI 2.0 era.
Based on this strategy, SenseTime is building one of China's largest AI cloud services. This not only addresses the rapidly growing global demand for AI and computing power but also realizes the company's commercial growth, fulfilling SenseTime's stated goal in its announcement: 'Firmly establish the core business represented by generative AI and accelerate the achievement of profitability and stable cash flow.'
Regarding computing demand, the current global AI application traffic is increasing, and the commercial potential of AI computing infrastructure is gradually being realized. According to Questmobile data, in September 2024, the scale of the AI-native application industry increased by 393.9% year-on-year, ranking first in growth among industries with a scale of tens of millions.
Image source: Questmobile
Faced with surging demand, SenseTime's early layout underscores the company's forward-thinking vision. As one of the earliest domestic enterprises to engage in the marketization and commercialization of intelligent computing centers, SenseTime initiated the construction of the AI large device SenseCore, a computing infrastructure, in 2020, providing powerful computing support for the development of general AI models.
Simultaneously, under the trinity strategy of 'computing infrastructure - large models - applications,' SenseTime is gradually advancing the commercialization of applications. According to the mid-year report, SenseTime's 'SenseNova' edge-side model has penetrated various industries, initiating commercial engagements with over 150 customers. More than 3,000 leading industry enterprises have utilized SenseTime's large models and intelligent computing services, with the overall invocation volume of the 'SenseNova' large model achieving a 400% increase.
SenseTime's strategic organizational restructuring will leverage the synergistic effects of the group's infrastructure advantages, facilitating ecosystem enterprises to quickly seize market opportunities in vertical industries.
As the technical capabilities of large AI models gradually improve, the trend of their integration with multiple industries to create new business models or new products is becoming increasingly evident. For example, SenseTime has released the JiRui Planting Decision Large Model and the JiRui Large Model Smart Farm Solution, demonstrating the role of new productive forces in empowering innovative development in modern agriculture.
The application prospects of 'AI Large Model+' across various industries are self-evident. SenseTime's 'SenseNova' large model system can continuously innovate based on general capabilities for downstream scenarios, creating large models for each industry scenario and generating greater value in niche areas.
This is currently the clearest path for SenseTime to achieve high performance growth and surpass the profitability inflection point. In the long-term development of the enterprise, this strategic organizational restructuring demonstrates the company's determination to accelerate along this path.
Of course, the challenges facing SenseTime in the future cannot be underestimated.
In the future, intelligence will be a global competitive landscape. As reasoning model capabilities improve and costs decline, overseas technology giants are accelerating breakthroughs at both the technological and application levels. For example, OpenAI's o1 series of models released in September are faster, 80% cheaper, and more competitive in coding tasks compared to the previous o1-preview.
The pressure between technology investment and commercial revenue falls on the shoulders of every player in this global competition. As Xu Li stated in an internal letter, 'We are experiencing what may be the largest technological wave in history. This era is full of uncertainties, but the only certainty is that the ultimate outcome of AI will have a profound impact on human society. The path may be tortuous, but we must be more adaptable, resilient, and capable of rapid transformation.'
In the wave of generative AI, SenseTime aims to become 'an organization with sustainable profitability and strong growth potential,' requiring both long-term focus on its core business and full awareness of the challenges it faces.