In a stark illustration of the escalating costs associated with the global artificial intelligence race, Chinese short-video giant Kuaishou has announced a massive 26 billion yuan (approximately $3.8 billion) investment plan for its Kling AI large model in 2026. This aggressive capital expenditure, reported by BigGo Finance on Monday, April 27, 2026, represents an 11 billion yuan increase over its 2025 base investment and notably exceeds the company’s entire net profit for the previous year. The announcement underscores Kuaishou’s determination to secure a leading position in the rapidly evolving AI video generation market, even at the expense of short-term profitability.
A High-Stakes Financial Gamble
The scale of Kuaishou’s planned investment caught the market by surprise. Analysts had anticipated an AI expenditure of around 18 billion yuan, making the actual 26 billion yuan figure a significant upward revision. The immediate financial implications of this strategy are substantial. Huatai Securities forecasts that this massive AI investment will drive Kuaishou’s 2026 adjusted net profit down by 15% year-over-year to 17.6 billion yuan (approximately $2.6 billion). The market reaction was swift and severe, with Kuaishou’s stock plunging 14% in a single day following the announcement, reflecting investor anxiety over the company’s willingness to trade margins for future technological dominance.
Kuaishou’s financial gamble is driven by the intense competition within the Chinese AI video sector. Since January 2026, nearly ten video generation models have been released worldwide, with the majority originating in China. Kuaishou’s Kling model, which received a major update on February 5, is currently one of the leading contenders. The company reported that Kling generated more than 1 billion RMB ($147 million) in revenue in 2025, attracting tens of millions of users and thousands of enterprise clients. Kuaishou expects this revenue to double in 2026, providing some justification for the massive upfront investment.
However, the path to profitability in AI video generation remains uncertain. During a March earnings call, Kuaishou CEO Cheng Yixiao acknowledged the challenges, stating, “Video generation models are highly complex… At this stage, we believe video generation technologies and products are still far from maturity.” This candid assessment highlights the inherent risks of Kuaishou’s strategy: the company is pouring billions into a technology that is still evolving and has yet to prove its long-term commercial viability.
The Broader Industry Context
Kuaishou is not alone in its willingness to sacrifice short-term profits for AI advancement. Its primary rival, ByteDance, is also investing heavily in the sector. According to BigGo Finance, ByteDance’s 2025 net profit plunged by more than 70% year over year, driven in part by AI capital expenditures of more than 150 billion yuan (approximately $22.0 billion). Of this total, approximately 90 billion yuan ($13.2 billion) was directed to AI computing infrastructure, chip procurement, and large-scale model research and development. ByteDance’s daily computing costs from token consumption alone are estimated at 50-100 million yuan, highlighting the staggering operational expenses of running advanced AI models.
The intense competition between Kuaishou and ByteDance is driving rapid innovation in the AI video space. ByteDance released its Seedance 2.0 model just days after Kuaishou’s Kling update, and the two companies are fiercely competing for users and enterprise clients. Other major players, including Alibaba, Tencent, and Baidu, are also aggressively developing their own AI video capabilities, creating a highly fragmented and competitive market.
(Related: iQiyi Says AI Will Make Most of Its Films and Shows Within Five Years)
The Race to Monetization
As the costs of AI development continue to escalate, the focus is increasingly shifting towards monetization. While some companies, like Alibaba with its HappyHorse 1.0 model, have opted for an open-source and free-to-use approach, others are implementing aggressive pricing strategies. ByteDance’s Seedance 2.0, for example, is priced at 13.8 RMB for an approximately 15-second clip, a rate that is higher than many of its competitors, including Kling.
An Alibaba Cloud executive recently told Caixin that “Free access is more of a marketing strategy, but it’s unlikely to be sustainable in the AI video space.” This sentiment reflects the growing realization that the massive investments required to develop and operate these models must eventually be recouped through sustainable revenue streams. Douyin estimates that the global enterprise AI video market could reach $36 billion by 2030, providing a massive potential prize for the companies that can successfully navigate the current period of intense competition and high capital expenditure.
Kuaishou’s 26 billion yuan bet on Kling is a clear indication that the company believes the long-term rewards of AI dominance outweigh the short-term financial pain. However, the success of this strategy will depend on its ability to rapidly mature its technology, attract a large and loyal user base, and develop sustainable monetization models in an increasingly crowded and competitive market. The coming years will be critical in determining whether Kuaishou’s massive investment will pay off, or whether the company has simply engaged in an unsustainable arms race.
Regulatory and Legal Hurdles
Beyond the financial and technological challenges, the AI video generation sector in China is also facing increasing regulatory and legal scrutiny. As the volume of AI-generated content explodes—the China Netcasting Services Association reported a 14-fold increase to over 2 billion clips in 2025—platforms are struggling to manage the influx. Six major platforms, including Douyin and Kuaishou, recently removed more than 37,000 problematic AI-generated clips, highlighting the ongoing battle against misinformation and inappropriate content.
Furthermore, the industry is grappling with complex issues involving intellectual property and personality rights. A notable case in March at the Beijing Internet Court involved actress Dilraba Dilmurat, who sued over the production of a synthetic face resembling her own. This lawsuit underscores the legal risks of AI-generated video and the need for clear guidelines on the use of likenesses and copyrighted material. As Kuaishou and its competitors continue to pour billions into this technology, they must also invest heavily in content moderation and legal compliance to ensure the long-term viability of their platforms.