Pjt A

The Acceleration of AI Technology: A Look at Marketing Strategies and Industry Leaders

The advancement of artificial intelligence (AI) technology is rapidly accelerating. In November 2022, OpenAI's conversational AI, ChatGPT, made a significant impact. Just five days after its launch on December 5, 2022, it surpassed 1 million users. This innovative product highlighted the enhanced performance of AI in society, leading to an explosion of interest and investment in AI technology. Subsequently, the value of GPU chip manufacturer NVIDIA, which powers large language models (LLMs), skyrocketed. NVIDIA is a fabless company capable of exclusively producing high-performance parallel computing GPU chips. To become the winner in the LLM market, companies are competing fiercely to purchase AI chips, build data centers, and develop better models.

In the advanced technology market, a 'winner-takes-all' phenomenon is prominent. AI chips are expensive, data is scarce, and technical skills are limited. AI, which learns and improves from data, was initially thought to favor companies with large data reserves like Google and Meta. However, OpenAI's GPT demonstrated that not only data acquisition but also advancements in model architecture, learning algorithms, and computing power could rapidly attract many users, accelerating the investment and development pace of LLMs. This put countries like Korea at a disadvantage compared to the US and China in LLM technology.

ChatGPT's Market Impact and Competitive Landscape

ChatGPT quickly gained paid subscribers due to its powerful service. As the 'brain' of services, LLM models like GPT can create various applications. For instance, GPT can be integrated into mobile phones, digital cameras, and other applications to provide AI conversational interfaces, content recommendations, and optimization services. As user numbers increase, so does the data, enabling advanced services and further accelerating technological interdependencies and the winner-takes-all phenomenon. For example, OpenAI's investor Microsoft has enhanced its services by integrating ChatGPT into its Azure cloud, browser, and Office suite.

Competitors' Strategies: Google and Meta

Google and Meta are striving to catch up with OpenAI. After the release of ChatGPT, Google introduced 'Bard' using its LLM 'LaMDA,' but it did not meet market expectations. Google then developed the 'PaLM' model and launched the conversational AI service 'Gemini' in September 2023. As a leading AI company and the originator of Transformer technology, Google is strategically closing the technology gap. ChatGPT's premium service, costing $20 per month, offers GPT-4 with no data limits and multimodal capabilities, processing text and images simultaneously. In contrast, Google's Gemini has a high-end model priced at $19.99 per month but also provides multimodal capabilities in its free model, allowing image-based conversations to attract users. Meta, on the other hand, is focusing on building an open-source AI ecosystem. Unlike OpenAI and Google, Meta has released its LLM 'LLaMA' to expand its ecosystem, a strategic move given Meta's current position as a latecomer with less advanced technology. LLaMA's performance is known to be at the level of GPT-3.5, and if Meta can attract many users with this strategy, its market share could increase significantly, leveraging network effects.

Rising Competitor: Anthropic

Anthropic, an AI startup founded by former OpenAI researchers in November 2022, has also been performing well. Within a year of launching its chatbot, Anthropic's 'Claude' emerged as a competitor to ChatGPT, absorbing users with improved performance and speed, although it does not yet offer image generation capabilities.

The Future of AI Technology Competition

The competition for AI technology advancement is expected to intensify. AI is shifting from LLM training to inference, focusing on actual task performance in practical applications. This change will determine whether the advanced technology market's winner-takes-all phenomenon accelerates or if lower technical entry barriers in each field create opportunities for countries like Korea. The interaction between these ecosystems will be closely watched.

Sources:

• Inside language models (from GPT-4 to PaLM) – Dr. Alan D. Thompson – Life Architect
• Korea Economic Daily, June 2024 - AI Semiconductor Frontline Shifts from Training to Inference, Opportunities for NVIDIA Competitors
• ZDNet Korea, May 2023 - Google Unveils Next-Generation LLM 'PaLM 2'
• AI Times, July 2023 - Meta Unveils Commercial Open-Source LLM 'LLaMA 2'
• Electronics Newspaper, March 2024 - Anthropic Launches AI Chatbot 'Claude 3', Claimed Stronger than GPT-4