In the rapidly evolving world of artificial intelligence (AI), China has emerged as a major player. Fueled by the success of OpenAI’s ChatGPT and driven by tech giants like Tencent, Baidu, Alibaba, and Huawei, China has witnessed a surge in AI startups and product announcements. However, beneath the buzz, there are signs of an impending shakeout in the Chinese AI scene. Investors and analysts have raised concerns about the sustainability of these AI ventures due to various challenges.
The “War of a Hundred Models
The AI frenzy in China, sparked by ChatGPT’s success almost a year ago, has led to what a senior Tencent executive referred to as the “war of a hundred models.” China currently boasts approximately 130 large language models (LLMs), accounting for 40 percent of the global total, trailing only slightly behind the United States’ 50 percent share. Moreover, numerous companies have unveiled “industry-specific LLMs” linked to their core models.
Challenges on the horizon
Despite the rapid growth of AI in China, investors and analysts have identified several challenges. Many AI projects have struggled to find viable business models, resulting in a crowded and competitive market with similar offerings. Rising costs are another concern, further exacerbated by tensions between Beijing and Washington. These tensions have led to reduced investment from US dollar funds in early-stage AI projects and difficulties in obtaining AI chips from companies like Nvidia.
Survival of the fittest
Esme Pau, head of China internet and digital asset research at Macquarie Group, predicts a period of consolidation and price wars among AI players as they vie for users. Only those with strong capabilities are expected to weather the storm. Several leading companies have already signaled their intent to compete on price, a strategy previously employed by cloud services giants like Alibaba and Tencent. Pau anticipates that in the coming months, LLMs with lower capacities may be phased out due to chip limitations, high costs, and intensified competition.
Differing opinions on market dominance
Opinions on which firms will ultimately dominate the market vary widely. Yuan Hongwei, chair of venture capital firm Z&Y Capital, believes that only two to three general-purpose LLMs will emerge as market leaders. Z&Y Capital, known for its investments in drone maker DJI and autonomous driving startup Pony.ai, has chosen to back Baichuan Intelligence. This five-month-old startup, led by Wang Xiaochuan, the founder of China’s second-largest internet search engine, Sogou Inc., aims to develop an open-source AI model to compete with Meta Platform’s Llama 2. Baichuan Intelligence secured approval from Beijing to release a public chatbot in August and is set to close a funding round valuing the company at $1 billion.
Backing experienced founders
Yuan Hongwei’s decision to support Baichuan Intelligence underscores the importance of experienced founders in the AI space. Other prominent entrepreneurs and tech executives, such as Kai Fu-lee, former chief of Google China, and Yan Juejie, a former vice-president of SenseTime, have also ventured into the world of Chinese AI startups. Their expertise and industry recognition make them attractive prospects for investors.
Incumbents hold the advantage
Some industry experts believe that China’s tech giants—Alibaba, Tencent, and Baidu—hold a significant advantage due to their extensive user bases and diverse services. These giants can easily integrate generative AI services into their existing ecosystems. This advantage stems from their ability to offer AI solutions as plug-ins to their cloud users.
The quest for unique solutions
Tony Tung, managing director at Gobi Partners GBA, points out that many investors jumped into the LLM frenzy earlier this year, only to discover that numerous startups were tackling similar problems with minor innovations in data processing techniques or model architecture. As a result, investors are now seeking unique and robust business cases, with some startups exploring partnerships with tech giants or the possibility of being acquired.
The initial excitement over LLMs in China has given way to a more sobering reality. While the AI landscape is undeniably dynamic and promising, investors and industry players are navigating challenges related to market saturation, escalating costs, and the need for differentiation. The Chinese AI scene is at a crossroads, where only the most capable and innovative players are likely to emerge as winners in the “war of a hundred models.”
