The Paris Olympics and the Evolution of Large AI Model Business Models

December 10, 2024

The Paris Olympics and the Evolution of Large AI Model Business Models https://picit.ai/features/ai-image-generator/create https://picit.ai/

This summer, the globally anticipated Paris Olympics provided a stage for athletes to showcase their talents, while also presenting opportunities for the commercialization and industrialization of AI and large model technologies. A French company developed an AI Q&A assistant called AthleteGPT, which interacts with athletes around the clock. Alibaba's Tongyi became the first large-scale AI model technology provider for the Olympics, supporting live broadcast signal distribution, assisting in event commentary, 360-degree live streaming, and bullet-time video generation. Additionally, other technologies like Baidu's AI-assisted diving training system and SenseTime's smart basketball systems, along with various referee assistance and decision-making products, were also showcased. AI large models have become a key force driving innovation across various industries. In April 2024, the International Olympic Committee (IOC) released the “Olympic AI Agenda.” However, challenges such as rising AI chip prices, limited computing power, intensified competition, and stagnant growth in consumer applications have made the commercial pathway for AI large models less straightforward. Even after rapid industry growth, leading AI companies are still in a money-burning phase, requiring various commercial partnerships and product integrations to sustain continuous updates and iterations of large models. With the fleeting nature of many industry applications, how can businesses achieve a closed-loop commercial model focused on the iterative core capabilities of AI large models? A mature industrial model may already offer some answers. From “Monopoly” to “Control” – Large AI Models Gradually Moving Behind the Scenes The advent of ChatGPT marked the beginning of the AI era, with its ability to rapidly attract a large following, leading the market to believe that products based on large models have the potential to become the gateway to the next digital age and a symbol of user entry into the intelligent era. The influx of capital also suggests that AI large models and AI assistants have great potential to become new internet gateways and traffic distribution hubs. From a technical perspective, AI large models possess powerful language understanding and generation capabilities, providing users with more intelligent and precise services. AI assistants can proactively execute tasks based on user intentions, enabling more personalized interactions. In terms of application scenarios, AI large models and AI assistants not only play roles in intelligent customer service and intelligent search but are also gradually permeating education, healthcare, finance, and other industries. For example, in intelligent search, a new internet portal battle has begun, with Perplexity's simple and direct search style already diverting a significant number of Google users, becoming a new avenue for information acquisition. As technology continues to evolve, AI will achieve “monopoly” on two levels in the future digital world. On one hand, AI assistants will become the gateway for users to access the digital world, assisting them in acquiring, processing, and sending information. On the other hand, AI assistants will act as intelligent operating systems, managing and accessing various applications to meet user needs. From the user’s perspective, intelligent services significantly reduce the learning curve for various applications; from the perspective of the digital world's information flow structure, AI large models will quickly centralize the internet world, with human interactions increasingly mediated by AI. The rapid iteration of AI has sparked concerns about “control.” From tech leaders like Elon Musk calling for a halt to the development of more powerful AI, to various associations publicly warning of AI's potential dangers, countries have begun preparing policies to regulate the safe development framework for AI large models. Compared to concerns about global conflict resulting from technological control, a more immediate risk is the systemic imbalance in social equity, data security, and innovation motivation caused by technological monopoly. AI technology is rapidly evolving, yet the promotion of AI large models is becoming increasingly low-key. Relative to the massive investment in building AI capabilities, individual commercial applications and star products struggle to support a sustainable commercial closed-loop. Based on the release of AI-related products both domestically and internationally, AI large models are gradually moving from front-line products to behind-the-scenes roles, becoming integrated modules like “GPT inside.” Beyond avoiding accusations of monopoly, what other commercial blueprints might be behind this shift? Cloud Empowerment as a Proven Success – Helping Clients Generate Revenue In May 2024, the Intelligent Terminal Large Model Alliance was announced, with partners like OPPO, Vivo, Honor, Xiaomi, Samsung, and ASUS gradually integrating their smart assistants and AI office applications with Volcano Engine's large model services. For instance, Samsung's new generation of Galaxy Z series smartphones integrated AI vision and assistant services with Doubao large model, while Honor built vertical product capabilities in mobile office and smart office by integrating various models from Doubao, such as speech recognition and role-playing. In May 2024, PwC signed on as the largest corporate customer of OpenAI’s enterprise version. In June 2024, Apple announced a partnership with OpenAI to integrate ChatGPT's AI capabilities into Apple Assistant Siri, incorporating this technology into its mobile, tablet, and computer operating systems. As early as March 2023, Microsoft integrated ChatGPT into its Copilot and began preliminary integration into its Office software suite. The flourishing AI applications at the Paris Olympics resemble a collection of puzzle pieces, where no single piece alone can support the extensive costs and expenses of large AI models. However, the complete picture formed by these pieces is gradually revealing the pathway to a commercial closed-loop for large AI models. Products and companies directly facing users, like Google and Facebook, face far greater antitrust risks than those providing B2B products and services, such as chip companies and cloud service providers. AI large models will inevitably shift away from a business model focused on creating traffic entry points, instead providing B2B services similar to cloud services, enhancing the competitiveness and intelligence of existing products and services. Tech giants like Apple and Microsoft, as well as leading domestic IT companies, have their own independently developed AI assistants, and they are now integrating more powerful large models. In the face of fierce market competition, only better products can win user recognition. More flexible IT business process integrations are also better suited to respond to the market's demand for instant services. The widespread adoption of cloud services requires companies to overcome the psychological barrier of moving core data to the cloud. The commercial value of cloud computing lies in its ability to provide scalable, pay-as-you-go services, simplifying development processes, and driving the transformation of enterprise IT architectures. AI large models empowering various industries follow a similar logic. As data protection technology matures and the capabilities of large models continue to improve, more enterprises will choose external large model services to provide users with better intelligent experiences at lower costs. The path to a commercial closed-loop for AI large models may likely offer a service similar to “cloud computing,” with the focus of this stage being the creation of an industry ecosystem centered around large models. For B2B companies, adopting advanced third-party AI technologies can improve customer experience, drive traffic to their own brands, and boost product sales. For large model developers, leveraging established channels to promote large models can reduce external antitrust pressure, obtain training data, and expand diverse application scenarios, which is the best soil for model upgrade and iteration. Through collaboration, both parties can maintain a leading position in an increasingly competitive market. Plus, if you are finding a convenient website to edit your photo, you can have a look on Picit.ai, its image generating function is fabulous as well.