Article reprint source: AI DreamWorks
Original source: Silicon Research Lab
Image source: Generated by Unbounded AI
With the popularity of AI big models, China's technology giants are setting off a storm of "running AI". From top-level strategy to business line reconstruction, AI has undoubtedly become a new story for big companies to bet on the future.
Since the emergence of "Miaoya", Alibaba has sounded the clarion call for comprehensive product upgrades with AI. "Silicon Research Lab" has observed that multiple business lines of Alibaba have connected to or upgraded generative AI applications and tools based on large models, including "Taobao Q&A", digital human video generation tool Live Portait, the latest AI-driven "Future Elf", Quark Scanner King, etc.
Tencent, on the other hand, is investing heavily in "AIGC+social" and "AIGC+music" based on its advantageous pan-entertainment position. Baidu is a large company that announced earlier that all its businesses will be restructured based on AI, and it emphasizes the "AI native" concept internally.
In summary, the main strategies of large companies can be divided into two categories: one is the upgrade and reconstruction of old markets and old businesses, which mainly include three actions: using AI to consolidate entry value, AI access to the entire product family, and combining their own cloud business with AI to better "sell cloud" with AI.
The other is the creation of new markets and new scenarios, which has two main lines. The bright line is that large companies are exploring AI native applications in head scenarios (such as AI painting, digital humans, social and office) and long-tail scenarios through internal competition, while the dark line is that they are expanding their capabilities by betting on potential AI star unicorns, just like overseas technology companies such as Microsoft and Amazon.
An analyst in the field of Internet media previously analyzed to Silicon Labs that the development of the large-scale application layer is faster than expected, whether it is to B or to C. "The scenarios and applications that Chinese technology companies are best at are already a stock game in the mobile Internet era, and they are all patchwork. In the large-scale model ecosystem, there are many giants, and it is still difficult to produce a hit application at present." According to the above analyst's prediction, the industry's boom upswing will begin in 2024.
Although the turning point has not yet arrived, this does not affect the new competition among major manufacturers. The speed of time, the differences in product paths and the differences in AI ecology may all determine their positions in the subsequent competition.
Seize the advantage in the field, product path "integration" and "division"
Over the past six months, there has been a consensus from the model side to the application side that the big model craze initially brought about by chatGPT has gradually passed the exciting and surprising period. As a large number of AI native applications enter the diffusion period and big models flow into mainstream developers, as Sequoia Capital defines in its report, "Generative AI is undergoing a process of transformation from technology-driven to customer-driven."
For China's major technology companies, during the cooling-off period, the shift from competing for technology to competing for customers is reflected in various specific actions.
As we mentioned above, one of the strategies of large companies is to upgrade and restructure old markets and old businesses, but in terms of product paths, there are also subtle differences and similarities.
The similarities are that major companies are all using AI to strengthen the value of their business entry points, while the differences lie in the different paths for realizing that entry value.
The actions of Alibaba, Tencent and ByteDance are mainly aimed at "patching" existing advantageous businesses. For example, in the social and pan-entertainment fields where Tencent has advantages, Tencent Music has launched the AI social product "Weiban" and robot-assisted creation functions, and simultaneously tested "AI Listen Together" and the AI companion "Xiaoqin".
Alibaba has taken the lead in using AI to reshape its business in e-commerce and productivity scenarios. For example, Taobao's AI native application "Taobao Q&A", which is currently in internal testing, essentially uses AI to improve the efficiency of user search behavior and realize the function of AI shopping guide. In the learning and office scenarios, the AI PaaS of DingTalk and the AI of Quark also reflect that Alibaba's multiple business lines are fully integrating AI capabilities.
ByteDance also quietly launched two "AI artifacts" - Xiao Wukong (formerly Wukong Search) which provides AI tools and the AI dialogue product "Doubao", and released two AI video projects on Github in one go, one is MagicAvatar which generates multimodal animations, and the other is MagicEdit which focuses on text-oriented video editing.
Baidu has greater ambitions, and its ideas are similar to those of OpenAI and Microsoft, building an ecosystem through plug-ins and creating a super traffic portal.
Not long ago, Baidu released the Wenxin Yiyan plug-in ecological platform "Lingjing Matrix". He Junjie, senior vice president of Baidu Group and general manager of Baidu Mobile Ecology Group (MEG), defined the relationship between the big model and plug-ins as "brain and hands and feet": "If the big model is a smart brain, then the plug-ins are the hands and feet of the big model. With plug-ins, the big model can not only answer general questions, but also be proficient in professional questions. It is both a generalist and a specialist."
It is not difficult to see that whether it is using AI to consolidate the value of the entrance based on the original business, or creating a super traffic entrance through a large model plug-in, large companies integrate large model capabilities based on the original huge user base, lower the threshold for using AI, and prepare for subsequent large-scale applications.
Another similarity is the integration of AI into the "full range of products" of front-end businesses. For example, Baidu has upgraded its front-end products, including Baidu Search, Baidu Wenku, Baidu Input Method, and Wenxin Yiyan APP. Alibaba has also enabled AI to empower its travel, entertainment, life, office, search and other business line products.
At the same time, large companies are also combining their own cloud businesses with AI to better "sell cloud" with AI. Internet cloud has changed from being an "integrator" in the early days to being "integrated" with each doing its own job, and its own advantageous technical products and role positioning are becoming increasingly clear. With the implementation of large models, cloud vendors can use the MaaS (Model as a Service) model to achieve product standardization in one stop, better implement the industry, and export AI capabilities and AI computing power to the outside world, thereby improving profit performance.
According to incomplete statistics from the Silicon Lab, Alibaba Cloud and Tencent Cloud have won many large orders in the fields of government affairs and finance since August this year, showing a strong position. Among them, Alibaba Cloud won the largest project in the market in August - the Zhejiang Provincial Big Data Development Administration Government Affairs Cloud Resource Leasing-Cloud Service Project, which amounted to 268 million yuan. In October, Alibaba Cloud won the bid for Beijing Energy International's 900 million AI computing power order.
A Baidu Intelligent Cloud person also mentioned in an interview with Caijing Eleven that Baidu pursues the sale of standard products and hopes to integrate them more intelligently. Therefore, it provides necessary integration services to some customers based on the actual needs of the industry and scenarios.
What is certain is that there is no obvious difference in the specific implementation paths of large companies as they rush to AI. The reason is that AI’s reconstruction of current business is not a static process, but requires a certain period of time.
The depth and breadth of AI applications depend on the allocation of computing power and other resources within large companies, business priorities, etc., and there are many uncertain factors. A typical example is Jing Kun, Baidu's vice president and former CEO of Xiaodu, who recently announced his resignation to start his own business. As one of the important figures in the landing of Baidu's AI ecosystem, the departure of the key figure "Father of Xiaodu" also adds more uncertainty to the future direction of this unicorn.
Fighting for AI’s original spark, big companies start “horse racing game”
In addition to integrating AI capabilities faster and better, with the rise of the "AI native" concept, major companies have also started a new round of horse racing games.
The so-called "AI native" applications actually refer to applications that rely entirely on large model capabilities. Kai-Fu Lee, chairman of Sinovation Ventures, cited WeChat, the most successful product in the mobile Internet era, as an example. In his opinion, the reason why WeChat was able to win in the mobile Internet era was that it abandoned the compatibility emphasized in the PC era, bet 100% on new platforms, and focus on the characteristics of mobile Internet. In other words, without mobile Internet, there would be no chance for WeChat to exist.
Back to the current era of AI big models, although major companies are still in the early stages of exploring AI native applications, at present, there are mainly two approaches: internal racing and external alliances.
First, there is internal horse racing, where those who run better will get more resources. Baidu has established a "horse racing mechanism" internally, and all large model-related applications will participate in internal horse racing, and only those who run better can get more resources.
The AI native applications emerging from Alibaba also show the same characteristics. The previously popular "Miaoya" was created by the entrepreneurial team behind Alibaba's big entertainment, and the "Zhennengzao" AI home design product launched in the Taobao zone was backed by the Taobao Jiyoujia technical team.
The "horse racing mechanism" is not uncommon among Internet giants. An Internet product manager told Silicon Labs that, on the one hand, internal horse racing can create samples and encourage innovation. Only when one product is successful can more products appear. On the other hand, it can discover potential talents. "Many teams that win the horse racing mechanism will be seen and get more resources."
Second, form external alliances and bet on potential star unicorns.
The cases of OpenAI, Microsoft and Amazon investing in Anthropic have proven that technology giants and start-ups can gain greater competitive advantages through early alliances. And when we look at the star AI start-ups that have emerged in China, we can see that the big companies are also working together to form alliances.
For example, Zhipu AI, an AI big model startup that recently completed its B-4 round of financing, has Tencent and Alibaba Cloud as its strategic investors. Previously, Zhipu AI's B-2 round was exclusively invested by Meituan. Another star startup, MiniMax, received a $40 million investment from Tencent in its new $250 million round of financing.
Previously, the two star companies that attracted Tencent and Alibaba to gather together were Didi and Xiaohongshu. Now, both have become giants in the industry. This time, the joint efforts of the big companies also announced from a strategic level the importance of alliances between technology giants and star startups in the era of big models.
To make running AI-based, tech giants need to overcome three challenges
In this AI native storm, although major companies have demonstrated their determination to "all in AI" or "all in big models", objectively speaking, there are still three visible hurdles in front of them if they want to win this game and become the final winner.
First, there is the myth of user value at the product level. Yu Jun, a legendary product manager in the Internet era, once proposed a formula for user value: user value = new experience - old experience - replacement cost.
Under the above formula, there are three main ways to increase user value: maximizing new experiences, minimizing old experiences, and reducing replacement costs.
Maximizing new experiences relies on accurate insights into user needs and an open mindset (whether it is fragmented integration of long-tail needs or focusing on head areas such as pan-entertainment and social networking). Minimizing old experiences actually requires breaking the old product design form. For example, in terms of the interactive interface, Baidu founder Robin Li once emphasized that "each AI native application cannot have more than two levels of menus." The replacement cost is the replacement demand and migration cost from old applications to new applications, which depends on the education of user behavior.
From the current AI application layer, there is an obvious tendency towards homogenization. How to innovate the experience requires large companies to think about and innovate new generative interfaces and editing experiences, and use complex proxy systems and other technologies to further improve user satisfaction and retention.
Secondly, rethinking AI native. As Kai-Fu Lee explained above, many of the current large companies’ application layer reconstructions are just under the banner of AI native. How to further develop breakthrough AI native applications based on the understanding, generation, reasoning, and memory capabilities of large language models depends on a more powerful model development stack and a more unique product blueprint.
Finally, it is a commonplace topic that large companies need to reconsider their role positioning and capability boundaries. Whether it is internal horse racing or external alliances, it is essentially "both and both", which is understandable in early development. However, the complex relationship between Microsoft and OpenAI on the other side of the ocean, "sleeping in the same bed but dreaming different dreams", has already confirmed that in the course of technological history, there are no permanent friends.
What is certain is that in the era of big models, the story of alliances and dreams among big companies will be repeated in the future. However, unlike the era of mobile Internet, this round of competition iteration is faster and more brutal. How to make the most of uncertainty, seize the first-mover advantage, lower the threshold for AI use, and achieve large-scale expansion is a new topic for big companies.
References:
1. Founder Park: "What do the top model layers and middle layers in China think about the implementation of large models?"
2. Sequoia Capital: Generative AI’s Act Two
3. Guotai Junan Securities Research: "Microsoft AI Application Development History (In-depth)"
4. Yu Jun: Yu Jun Product Methodology
