Xiao-I Tries To Generate Buzz In Crowded China AI Field

Key Takeaways:

• Xiao-I has introduced its Hua Zang Universal Large Language Model ecosystem, becoming China’s latest company trying to capitalize on demand for conversational AI services
• Commercializing the product will be challenging in an increasingly crowded field with more than 100 similar products in the market

By Warren Yang

It’s no secret that artificial intelligence can do amazing things. But it’s far from guaranteed that the hot technology can do equally amazing things for the growing number of companies jumping on China’s AI bandwagon. With big guns like Baidu BIDU and Alibaba BABA all piling onto the AI train, just how and whether small shops like Xiao-I Corp. AIXI can stand out in the crowd is a question that even the smartest AI can’t easily answer.

Last Thursday, the recently-listed AI developer unveiled its new Hua Zang Universal Large Language Model Ecosystem tool, seeking to attract any eyeballs it could by labeling its newest product as “ground-breaking.” Large language models (LLMs) like ChatGPT that power conversational or generative AI are all the rage these days for their vast potential in a wide range of applications. They “learn” by processing huge volumes of data that allows them to mimic human language skills and generate text via chatbots.

Whereas many chatbots making headlines these days are targeted at a general audience, Xiao-I is focusing specifically on business users to set itself apart from the crowd.

“The Hua Zang Ecosystem is a testament to Xiao-I’s commitment to providing state-of-the-art solutions for businesses and further advancing the field of conversational AI,” Xiao-I said. “By leveraging the latest advancements in AI technology, these services enable organizations to unlock the potential of AI-powered virtual assistants and chatbots to seamlessly facilitate interactions with their customers.”

Xiao-I says Hua Zang is customizable to meet specific business needs, and the company has formed partnerships with more than 100 “renowned” businesses across 20 industries. That may sound impressive, but also may not mean much. Here’s why.

The market for conversational AI in China can be massive for homegrown companies, especially because ChatGPT is banned in the country. But it doesn’t mean that it’s an easily market to exploit. For starters, even though the whole craze about ChatGPT is still quite new, there are already more than 100 LLMs developed by Chinese companies, from big names like Alibaba to older homegrown startups like SenseTime (0020.HK).

Many of their features are all similar, so it’s hard to differentiate one product from another. Also, commercializing them hasn’t been easy, even for big companies like Alibaba, probably because no one is willing to pay big sums to use them, at least not yet.

Many issues may be making businesses, which are typically the most willing to pay for such services, hesitant to use AI chatbots. They include a lack of in-house expertise, insufficient supporting infrastructure, and concerns about data security, just to name a few. Some may also be concerned after reading about recent cases where AI responded to some queries by simply making up fake information.

High Costs

All this means that in addition to their huge product development costs, LLM developers like Xiao-I will also need to spend lavishly to promote their products and probably offer hefty discounts that keep them squarely in the red and further delay any profits.

Chinese developers are also subject to strict domestic regulations due to the sensitive nature of information gathering and publication. In August, the Cyberspace Administration of China introduced rules specifically for generative AI to make sure chatbots don’t produce content the government considers inappropriate. Full compliance can be tricky for LLM developers, as one lapse in coding can result in a chatbot saying things that it shouldn’t, triggering severe consequences.

While compliance is a more abstract challenge, controlling spending on essential R&D is another very real problem all companies are facing. That’s especially true for smaller independent players like Xiao-I, whose R&D spending challenge is quite evident in its latest financial results.

The company’s revenue more than doubled to $26.5 million year-on-year in the first half of this year. That’s still a modest amount, but its gross profit margin increased 6.2 percentage points as it reduced its cost of revenue thanks to greater economies of scale that come with increased sales. All of this is nice – except, its R&D expenditure swelled eightfold to nearly $30 million – more than its total revenue – moving the company into the loss column after reporting a small net profit a year earlier.

Xiao-I appears to be focusing this year on selling its “cloud platform products,” which accounted for more than 85% of its revenue in the first half, compared to just 28% a year earlier when customized technology development services were its biggest income source. The growth in cloud platform services is driven by AI tools that help businesses automate their processes.

The company is also vulnerable due to its heavy reliance on a handful of customers. Its top five accounted for nearly 60% of its revenue at the end of last year, which means that the loss of any would be a big blow.

Xiao-I went public in New York in March, pricing its shares at $6.80 apiece, the bottom end of its indicative range, suggesting lackluster investor interest. It raised nearly $40 million in the listing, which doesn’t look like much but is actually relatively big for Chinese companies listing in New York in the current frosty climate.

Since the listing, Xiao-I has been a company on steroids in trying to wow investors, issuing news releases on a weekly basis touting its latest accomplishments, including product launches, media appearances and new business developments.

Despite that, the stock has lost more than 70% of its value since the listing and now trades at a price-to-sales (P/S) ratios of just about 2. That’s far lower than 11 for SenseTime and 6.8 for the recently listed Fourth Paradigm (6682.HK), two of the most advanced AI companies in China. That valuation gap shows the challenges Xiao-I faces in making a splash in China’s fiercely competitive AI space, no matter how hard it tries to talk up its innovations and accomplishments.