AI Companies Struggle to Profit: How to Break Through in 2024?
-
If asked which industry was the most popular and hottest in 2023, the answer would undoubtedly be AI. After the uncertainty of 2021 and the sluggishness of 2022, the emergence of ChatGPT led to another major boom in artificial intelligence in 2023.
However, amid such a booming market, some have raised concerns: 'Where will AI profits actually come from in 2024?'
Since the birth of the AI industry, the difficulty of profitability has always hung like a 'Sword of Damocles' over AI companies, threatening their survival and development. From a commercial perspective, AI companies tend to focus excessively on the technology itself while overlooking market development and consumer demand exploration. The AI industry follows the business principle of "the fast eat the slow," leading companies to prioritize "speedy R&D first, commercial applications later." This results in persistently high R&D costs, and even when companies become profitable, the payback period remains exceptionally long.
For example, in 2023, ChatGPT's commercialization brought OpenAI over $1.6 billion in revenue. However, compared to its tens of billions in R&D investment, this still falls short of covering costs. Moreover, OpenAI has stated it will continue increasing its investments.
Looking at China's AI industry as a whole, losses have become the norm. Take SenseTime as an example: statistics show that from 2018 to 2022, the company accumulated staggering losses of 43.8 billion yuan. Despite such "consistent" losses, SenseTime has seen little progress in developing new business lines. Data reveals that in 2022, its primary revenue still came from two main sectors—smart business and smart cities—generating only 1.464 billion yuan and 1.096 billion yuan, respectively, representing year-on-year declines of 25.2% and 48.8%. The other "three dragons" in the industry face similar challenges. In business development, securing government contracts and large corporate orders is often the fastest and most effective revenue source. This has led most AI companies to concentrate their market development efforts on To-G (government) and To-B (business) sectors. The "Four Dragons" (leading AI firms) all started by undertaking government projects. However, this results in a singular business model for AI companies, making them vulnerable to crises during economic downturns or market saturation.
In fact, not just the "Four Dragons," but the vast majority of AI companies in China are currently caught in a dilemma between practical application and profitability, surviving mainly on financing. The recent trend of AI companies collectively listing in Hong Kong exposes the underlying issue: AI firms struggle to turn a profit and remain in a prolonged state of losses.
Industry experts have expressed concerns: If Chinese AI companies cannot expand into the mass consumer market and go global like computer and internet giants, their technology will remain a high-end tool for the few rather than a transformative force that benefits the general public. Currently, the domestic AI industry's inability to achieve profitability stems from both internal and external factors. Internally, the costs of developing new technologies and products are exorbitant and continuously rising, resembling a bottomless pit, yet few ultimately achieve commercialization and profitability. Externally, most revenue-generating businesses are To B and To G oriented, with very few To C ventures.
However, the B-end and G-end consumer markets for domestic AI products remain sluggish. The G-end market, in particular, suffers from prolonged decision-making cycles, with businesses like smart cities offering even smaller residual opportunities. The B-end market is no better, as years of pandemic-induced corporate layoffs have left companies with no budget to spare. In stark contrast, the C-end consumer market remains vastly underdeveloped, with certain sectors neglected for extended periods.
Taking the 'Four Dragons' as an example, statistics show their revenues primarily come from To G security businesses. Among them, SenseTime, CloudWalk, and Megvii derive over 70% of their income from this sector. Although Yitu began expanding into B-end businesses like finance, healthcare, and internet services as early as 2015, these contribute less than 5% to its revenue, with To G operations still forming the backbone of its income. At its core, the problem stems from the "Four Little Dragons" that started with To G (government) business, inherently lacking To C (consumer) genes. Their long-term reliance on government business has trapped them in a "government resource trap," leading to a decline in market competitiveness. When the B2B and G2G markets are impacted, their revenue becomes unsustainable.
Of course, some insightful individuals in China's AI industry have recognized the importance of To C business for AI companies. Liu Qingfeng, Chairman of iFlytek, mentioned at the "Yangtze River Delta Business Summit" that To C business is a key direction for the company's development. At a meeting on elderly care robots organized by the National Development and Reform Commission in December last year, Chen Haibo, founder of DeepBlue Technology, also proposed that "AI companies should develop towards the C-end."
In fact, some domestic AI companies do possess To C genes. For example, iFlytek's AI Translator is an AI product aimed at the general consumer market. Upon its release, it caused quite a stir in the domestic and international intelligent voice product markets and has since undergone multiple iterations. The DeepBlue wheel-legged home AI robot "Lanbao", which debuted on CBN's "Brainstorm" program late last year, represents a cutting-edge digital AI product for home elderly care targeting C-end users. Combining the flexibility of legs with the mobility of wheels, it features diverse movement patterns and strong adaptability to complex terrains, receiving unanimous praise from guests.
Additionally, on January 3rd of this year, SenseTime launched its C-end consumer hardware product - the Yuanluobo Light Wing Lamp. The introduction of the Yuanluobo product series signals SenseTime's initial attempts to expand into the C-end market.
From a market scale perspective, G-end (government) business represents a stock market with stable but limited sources - each project completed reduces available opportunities while facing strict government financial audits, resulting in low efficiency and minimal profits. Although B-end (enterprise) business has a significantly larger market scale than G-end, it remains highly unstable due to macroeconomic conditions and clients' operational performance. In contrast, C-end (consumer) business belongs to an incremental market with vast customer base, large market scale, diverse and stable revenue streams, and minimal macroeconomic impact. Consumer purchases are often impulsive decisions made within 20 seconds, driven by emotional rather than rational considerations. Clearly, relying solely on B2B and government markets is not a sustainable strategy for AI companies. The only way to overcome challenges is to deeply engage with the consumer market, providing tangible hardware products that meet the needs of the general public.
On January 9, the "2024 Consumer Electronics Show" (CES) in Las Vegas, USA, opened, attracting over 4,000 companies and institutions from more than 150 countries and regions. AI emerged as the top theme of this year's CES, signaling that AI will be a major trend and hotspot in the tech industry this year. The good news is that the overseas AI consumer market has recently entered a phase of rapid development.
In fact, over the past two years, several domestic AI and robotics companies have been thriving in overseas markets. According to media reports, in 2023, the renowned domestic AI robotics manufacturer DeepBlue Technology signed multiple export orders for intelligent indoor cleaning robots with distributors from Norway and Spain. So far, its various robots have been exported to 12 countries including Italy, Russia, Japan, and Australia.
Ecovacs Robotics' 2023 financial report shows that its overseas revenue in the third quarter increased by 47% year-on-year, while its overseas business revenue in the first half of the year grew by 26.5% year-on-year, accounting for 34.3% of its total revenue.
Similarly, Geek+, Dreame Technology, and Gaussian Robotics have also intensified their efforts in overseas markets. A Dreame Technology executive revealed in an interview with Yicai that the company's overseas business revenue doubled in the first half of 2023, with a year-on-year growth of 110%. Li Tong, founder and CEO of Keenon Robotics, also mentioned in an interview with Securities Daily that overseas orders in 2023 increased by over 80% year-on-year. It is evident that overseas markets have become a significant source of profitability for some Chinese AI and robotics enterprises. The vast overseas market is undoubtedly set to become the 'goldmine' in the eyes of major AI companies in 2024.
From an economic perspective, profitability is key to a company's survival. However, domestic government-sector (G-end) business is limited in quantity and offers low margins, serving primarily as a credibility booster. Meanwhile, business-sector (B-end) demand is constrained, providing only a safety net. Thus, it appears that Chinese AI companies can achieve self-sustaining 'hematopoiesis' only by tapping into overseas consumer markets (C-end) for their products.
Examining the current trajectory of the global AI industry, venturing into the C-end market presents considerable challenges for Chinese AI firms. Whether they can carve out a viable path in this increasingly competitive global AI consumer market with their To C products remains uncertain. Nevertheless, the strategy of integrating products into C-end user consumption scenarios is undeniably the right direction. Some may still doubt the purchasing power of foreign markets, but how would you explain the recent viral videos showing 'Chinese three-wheelers being snapped up by Americans'? Indeed, it's not that foreigners can't afford products from other countries, but rather that Chinese products offer better value for money.
As Chen Haibo, founder of DeepBlue Technology, stated at the National Development and Reform Commission's elderly care robot symposium: 'Only by boldly going global, exploring overseas markets, and facing global consumers can AI companies grow stronger.'