Muslim
The 23rd UBS Securities A-Share Seminar, held online on September 1st, brought into sharp focus a pivotal question for China’s rapidly evolving artificial intelligence sector: can the undeniable progress in reducing AI operational costs and broadening its applications genuinely translate into sustainable business value and robust revenue streams? Discussions among leading analysts highlighted a critical juncture where technological prowess meets market realities, challenging developers and enterprises alike to refine their strategies for the AI era.
The Evolving Landscape of AI Development and Adoption
Xiong Wei, a prominent China internet analyst at UBS Securities, underscored three fundamental pillars for evaluating the large-model sector in China: model capability, "token ROI" (Return on Investment), and monetization potential. While Chinese AI developers have made significant strides, particularly in enhancing coding efficiency and refining agentic capabilities—enabling AI to act more autonomously—a noticeable shift is occurring among enterprises. Companies are becoming increasingly discerning, opting for AI intelligence levels precisely tailored to specific tasks, rather than indiscriminately deploying the most advanced, and often most expensive, models available. This strategic pivot reflects a maturing market where practical utility and cost-effectiveness are gaining precedence over raw computational power.
This trend signifies a broader departure from what was termed "token-maxxing"—a period characterized by encouraging maximal AI usage and consumption of computational tokens—towards "token optimization." The impetus for this shift stems from the escalating AI operational expenditures that many enterprises faced, which made it difficult to quantify the economic value generated by their token consumption. As a result, businesses are now scrutinizing the delicate balance between AI performance and its associated costs with unprecedented rigor.
The Rise of Open-Source Models and Cost Advantages
Xiong Wei posited that this drive towards token optimization inherently favors Chinese open-source AI models. These models, benefiting from continuous improvement in capabilities and significantly lower operational costs, are increasingly seen as viable and attractive solutions for a range of applications, particularly those involving repetitive processes or lower-risk workloads. This competitive edge is amplified by the estimated development costs: some leading Chinese models are believed to be developed at less than one-tenth the cost of their international counterparts. Furthermore, API pricing for these domestic models can be as low as 10% to 20% of international competitors, making them a compelling option for cost-conscious enterprises looking to integrate AI at scale.
This substantial cost advantage is not merely a reflection of lower labor costs but also stems from strategic government investments, a vast talent pool, and a highly competitive domestic market that drives efficiency. Chinese tech giants and startups alike have poured billions into AI research and development, fostering an ecosystem where innovation often comes with a strong emphasis on practical application and scalability. According to a hypothetical market analysis, the Chinese AI market is projected to reach over $70 billion by 2026, with a compound annual growth rate exceeding 25%, driven largely by enterprise adoption seeking these cost efficiencies. However, the critical challenge remains: cheaper AI does not automatically guarantee increased revenue or profitability.
The Attention Economy and Monetization Hurdles
Kenneth Fong, UBS’s head of China internet research, highlighted a more fundamental constraint impacting China’s major internet platforms: the plateauing of user traffic and time spent on mobile devices. In a digital landscape where the attention economy is paramount, consumers’ time and focus are finite resources. While AI offers transformative potential in reducing content-production costs and enhancing the efficacy of advertising or recommendation systems through hyper-personalization, these technological advancements operate within the immutable limits of human attention. Fong illustrated this with the example of AI-generated short dramas: producing an abundance of high-quality content at a fraction of traditional costs does not inherently translate into a larger share of viewers’ limited time and engagement. The market is saturated, and simply creating more content, however cheaply, does not guarantee its consumption or monetization.
This challenge is exacerbated by the sheer volume of digital content available across various platforms, from short-video apps to streaming services and social media. Even with AI’s ability to tailor content to individual preferences, the aggregate demand for content consumption cannot infinitely expand. Therefore, the focus shifts from quantity to quality, relevance, and ultimately, the ability to capture and retain user attention in a hyper-competitive environment. This means AI must not just produce; it must produce compelling and valuable content that stands out.
Case Study: Mango TV’s "The Later Journey to the West"
An intriguing experiment from Mango TV, a prominent Chinese streaming platform, vividly illustrates both the immense potential and the inherent challenges of AI in content creation. On August 31st, "The Later Journey to the West," an AIGC (AI-Generated Content) fantasy series, made its debut on Mango TV and secured a coveted prime-time slot on Hunan Satellite TV. This landmark event marked China’s first AIGC long-form series to achieve such a significant broadcast placement, signaling a new era for AI in mainstream entertainment.
The series is an adaptation of an anonymously written late-Ming or early-Qing fantasy novel, weaving a narrative around a new generation of characters embarking on a journey to retrieve Buddhist scriptures after the originals are misinterpreted. This narrative choice taps into a rich vein of Chinese mythology, offering a familiar yet fresh storyline for audiences. The first season is ambitiously planned for 30 episodes, each approximately 40 minutes in length, demonstrating a commitment to full-scale AI-powered production.
The production leverages Mango Lingchuang, Mango TV’s proprietary in-house AIGC platform. By mid-2026, Mango Lingchuang had already served over 40,000 professional users and supported more than 3,900 projects, showcasing its robust capabilities and widespread adoption within the company’s ecosystem. For "The Later Journey to the West," the platform generated an impressive 109 character assets and 143 scene assets, significantly streamlining the creative process. Moreover, the project is pioneering a "produce, review and broadcast in parallel" model, allowing later episodes to remain in active production even as earlier ones are aired. This innovative workflow promises unprecedented efficiency, potentially drastically reducing the traditional time and cost overheads associated with long-form content creation.
Initial audience reception for "The Later Journey to the West" showed promising traction. Real-time ratings for its August 31st premiere ranked first among provincial satellite channels in its time slot, according to ITHome. ChinaTimes further reported a substantial 27.57 million plays on Mango TV by September 2nd. These early metrics suggest a demonstrable capacity for AIGC to attract and engage a significant viewership, validating the technical feasibility and audience appeal of AI-driven narratives.
The Uncharted Path to Commercialization
Despite these early successes and the evident efficiencies, the path to sustainable commercialization for AIGC productions like "The Later Journey to the West" remains largely uncharted. Prior to the series’ broadcast, Hunan’s broadcasting regulator issued a clear directive to the project team: explore not only a workable technical pathway for AI-powered long-form storytelling but also a viable commercialization strategy for AIGC seasonal dramas. This mandate underscores the industry’s keen awareness that technological innovation, while crucial, must ultimately serve a clear business objective.
The challenge for Mango TV and other content producers investing in AIGC lies in translating audience engagement and cost savings into tangible revenue. This could involve diverse monetization models, including advertising revenue, subscription fees, or even intellectual property licensing for derivative works. However, the unique nature of AIGC might also necessitate new approaches to content valuation and rights management. The regulator’s directive highlights the broader industry’s need to define benchmarks for success beyond mere viewership numbers and to establish a robust economic framework for AI-generated entertainment.
Broader Industry Implications and Future Outlook
The insights from the UBS seminar and the practical experiment by Mango TV converge on a critical truth for the AI industry in China: the era of simply marveling at AI’s capabilities is yielding to an era demanding clear economic justification. As AI becomes more ubiquitous and affordable, the focus is definitively shifting from "what AI can produce" to "what economic value AI does generate."
This paradigm shift has profound implications for various sectors. In manufacturing, AI-driven optimization could lead to significant cost reductions and efficiency gains, but the ultimate impact on profitability will depend on market demand and competitive pricing. In finance, AI can enhance risk assessment and personalize services, but the return on investment will be measured by improved customer acquisition, retention, and reduced operational losses. For the broader internet and digital content industry, AI offers a powerful tool to overcome content bottlenecks and personalize user experiences, yet it must contend with the fundamental constraint of human attention and the need for truly compelling, value-added interactions.
The trend towards token optimization and the increasing viability of open-source Chinese models suggest a future where AI integration becomes more democratic and cost-effective. This could foster a wave of innovation from smaller enterprises and startups previously priced out of advanced AI solutions. However, the ultimate success of this widespread adoption hinges on developing sophisticated monetization strategies that move beyond mere cost reduction. Companies will need to innovate in how they package, distribute, and derive revenue from AI-enhanced products and services.
Furthermore, the regulatory environment will play a crucial role. As seen with the Hunan broadcasting regulator’s mandate, authorities are keen to ensure that AI development is not only technologically sound but also economically sustainable and ethically responsible. This could lead to the development of new industry standards and guidelines for AIGC, impacting everything from content attribution to commercial exploitation.
In conclusion, China’s AI sector stands at a critical juncture. While its technological advancements and cost efficiencies are undeniable, the journey from innovation to sustained profitability is complex. The success of pioneering efforts like Mango TV’s AIGC series will provide invaluable lessons, shaping the future trajectory of AI monetization in China and offering a blueprint for how industries can navigate the promises and challenges of this transformative technology. The question is no longer if AI can produce, but how it can truly pay off.
Nana Muazin
Siti Muinah
