Meta's 'Muse' Reaches No. 1 on US App Store within Ten Days of Launch, Surpassing ChatGPT

Meta's personal AI agent 'Muse' is attracting market attention with an explosive increase in users immediately after its launch. Muse, which Meta first unveiled in the United States on September 8, rose to number one on the US App Store's popular apps list within ten days of its release, and its daily active users (DAU) also recorded approximately 630,000, significantly outpacing ChatGPT's 230,000. According to the market research firm Apptopia, Muse's iPhone download count in the United States and Canada reached 1.8 million, showing faster growth than ChatGPT's 1.3 million.

Independent Assistant Function Based on Dedicated Virtual Computer (VM) per User

Muse goes beyond a simple conversational chatbot and has a strong character as an 'agent' that performs tasks on behalf of the user. Operating on 'Muse Spark 1.3' developed by Meta's Superintelligence Research Lab, this service performs email writing, travel booking, and browser searches, as well as price negotiations with customer centers. As a real-life example, it was confirmed that when a user in the United States requested a reduction in internet service fees, Muse directly called the customer center and, after speaking with an agent, saved $1,920 (approximately 2.6 million won) based on a 2-year contract. Additionally, it features functions that provide practical economic benefits, such as finding $730 in Amazon rewards forgotten for over three years and a $272 refund from a telecommunications company.

The reason these individualized services are possible is due to a structure called 'Muse CQ VM'. While existing chatbots generate answers on shared servers, Muse allocates one independent virtual computer (VM) within the cloud for each user. Since the agent and the user's data are contained within this virtual computer, it can perform the user's work while maintaining independent memory and context, separated from other users.

Commerce Expansion and Revenue Structure through Shopify Integration

Meta is aiming for the full-scale 'agent commerce' market through Muse. Muse has formed partnerships not only with large US retailers such as Walmart, Best Buy, and Sephora, but also with Instacart (grocery shopping) and Expedia (travel). In particular, the integration with Shopify, which supports online stores worldwide, is key. Stores on Shopify are designed so that products can be searched and purchased in Muse without a separate application. When a user requests a product, Muse suggests candidates and completes the order using saved payment methods upon receiving the user's approval.

There are two main revenue models. The basic service is free, but the company operates monthly subscription plans of $20 and $100 for users who desire higher usage. At the same time, Meta is aiming to generate revenue through transaction fees. This is a structure where fees generated when Muse sells goods or handles reservations are received from the participating sellers, showing a flow similar to Meta's past method of providing free services to users and generating revenue through advertisers.

Changes in Hardware and Semiconductor Industries Following the Spread of AI Agents

The emergence of Muse predicts changes in demand across the AI industry in the fields of memory, CPU, and AI devices. First, since a dedicated virtual computer that operates 24 hours a day must be allocated to each user, demand for memory is expected to increase sharply compared to chatbots. According to research data, it was found that approximately 8GB of memory and 100GB of storage space are required per virtual computer.

Furthermore, as the agent performs the 'orchestration' role of breaking down tasks and selecting tools, the importance of the CPU is also growing. This leads to observations that it could present new opportunities for CPU manufacturers such as Intel and AMD, in addition to GPUs. Finally, to break away from smartphone platforms, Meta unveiled a dedicated device, 'Muse Charm'. This device, equipped with a 2-inch screen and a fingerprint sensor, provides a touchpoint where users can communicate with the agent immediately without opening smartphone apps, and it is likely to lead to an AI hardware war with competitors such as OpenAI in the future.

Source: original video (YouTube)