The True Survival Situation of Companies Behind the AI Craze

In the past couple of years, AI news reports have been everywhere, and AI-related companies have been springing up like mushrooms after a rain. According to Zhang Xueli, Deputy Director of the China Academy of Information and Communications Technology (CAICT), by September 2018, there were 5,159 AI companies worldwide, with China having 1,122 (excluding Hong Kong, Macau, and Taiwan), ranking second globally. Beijing, with 445 companies, became the city with the most AI companies worldwide.

The AI Frenzy

Behind the frenzied news reports, the AI industry is also attracting significant capital. In the first half of 2018, global funding for the AI sector reached $43.5 billion, with China accounting for $31.7 billion, representing more than three-quarters of the global total. This was another impressive victory: China secured three-quarters of global funding with just one-fifth of the companies, demonstrating investors' high expectations for China's AI sector.

The Real Survival Situation of AI Companies

According to a media report on the commercialization of artificial intelligence, Chinese AI startups received over 50 billion yuan in funding in 2017, but the combined revenue of the top 100 AI companies amounted to less than 10 billion yuan. In the entire industry, over 90% of AI companies remain unprofitable, with the majority generating less than 200 million yuan in annual revenue.

Alibaba's former CEO, David Wei, commented on this phenomenon, stating that the current AI bubble is extremely large, driven by media hype and overheated markets. Many companies claim to be "AI companies," but nine out of ten AI companies are "pseudo-artificial intelligence."

The Proliferation of Pseudo-Artificial Intelligence and Poor Usability

The inability to commercialize technology is the primary reason why many AI companies are labeled as "pseudo-artificial intelligence." One reason is excessive promotion of laboratory data, leading to a significant gap in user experience. Another reason is blind imitation without genuine innovation, as exemplified by the oversaturation in facial recognition technology. When product technology becomes homogenized, price wars are bound to erupt.

The Essence of Artificial Intelligence

What is artificial intelligence? The true essence of artificial intelligence is "machine learning." Machines can only successfully complete certain tasks when they have correct data to learn from.

Four major bubbles affect the development of artificial intelligence:

• Technical bubble: AI is an interdisciplinary and application-oriented field with high professional barriers, and few people truly understand it.
• Capital bubble: There is too much capital and too many concepts; everyone can see the "hot spots," but everyone is watching to see how to capitalize on them.
• Commercial bubble: Many AI applications are pseudo-intelligent, with most companies lacking the ability to use AI and data to serve customers and create value.
• Valuation bubble: Achieving billion-dollar valuations is not difficult; the real challenge is finding a sustainable business model. Most AI remains in the technical tools stage, with a long way to go before reaching platform and product stages, and profitability is still far off.

The Path Forward for AI Companies

AI companies struggling with the capital bubble face significant product homogenization, with some companies merely blindly following trends without true innovation. "Deflating the bubble and addressing key issues" is the only path forward for AI companies. In this environment of survival of the fittest, AI companies should focus on solidifying their internal skills, pursuing steady growth, leveraging data and technology, and finding better commercialization pathways to serve users and realize their value.

An industry needs its own ecosystem and rules to grow; throwing money at a problem won't necessarily solve it. It requires funds, time, talent accumulation, and team culture development. Rushing to get results will yield undercooked outcomes, while slow and steady approaches can lead to better long-term success.