After Meta (META.US) Muse exploded, it received many hits on Wall Street! Cantor FitzGerald raises target price to $860 trillion AI smart device market

The Zhitong Finance App learned that with Muse, a personal AI agent owned by Meta Platforms (META.US), quickly became popular after its launch, Cantor Fitzgerald believes that if AI agents become the “third S-shaped growth curve” of artificial intelligence development, Meta is expected to become one of the main beneficiaries of this trend. The bank reaffirmed Meta's “bullish” rating and raised the target price sharply from $680 to $860.

Cantor analyst Deepak Mathivanan said that Meta has a complete competitive advantage in the field of AI agents, including models dedicated to intelligent task training, Muse's differentiated user experience, and huge computing power infrastructure. These advantages are expected to help Meta quickly promote personal AI agents to its huge user base before competitors enter on a large scale.

Muse quickly rose to the top of the Apple App Store after launching earlier this month. Its rapid popularity has not only raised the market's attention to Meta's AI business, but has also raised investors' concerns that AI agents may disrupt some traditional business models. Recently, stocks in various related industries have been sold off as a result.

Mathivanan believes that the potential market size for personal AI agents could reach up to 1 trillion US dollars. He said that if calculated according to a relatively narrow business model, considering only subscription revenue and e-commerce transaction commissions, the long-term revenue potential of individual AI agents may reach more than 250 billion US dollars; if more extensive estimates are made based on the time saved by AI for users and their corresponding value, the market opportunities may exceed 1 trillion US dollars.

Using the smartphone market as a reference, Mathivanan points out that smartphones have now generated more than 600 billion US dollars in advertising, commissions, and subscription revenue by improving the Internet usage experience. He believes that if personal AI agents are popularized on a large scale in the future, their long-term market size may even exceed the related market created by smartphones. He pointed out that if AI agents become the “third S-shaped growth curve” of the artificial intelligence industry, Meta is already in a favorable position.

He also said that Meta has the “right mix” of model architecture, agent operation framework, and user interface needed to further popularize Muse. Among them, Muse's product design is particularly noteworthy. Mathivanan pointed out that Meta's accumulated product experience in feed information flow and recommendation systems is more compatible with AI agent applications, and can simultaneously carry various user interaction methods such as chat, information flow, and creativity in Muse, thereby increasing usage frequency and user stickiness.

At the same time, Meta's huge computing power infrastructure means that once the number of Muse users continues to grow, the company has the foundation to rapidly expand AI reasoning capabilities and promote products to the existing user base. This also allows Meta to have a different expansion path compared to AI startups that lack a huge user base and computing power resources.

However, Muse is still in the investment phase. Mathivanan estimates that in the early days of Muse, Meta's cost for each AI task was about $0.25, which is equivalent to about $11 per user per month based on current average usage levels.

He believes that currently Muse's unit economic benefits are actually still being subsidized by Meta, but as the number of users expands and the AI model continues to advance, there is room for significant improvement in inference costs and profit margins.

Mathivanan predicts that in the next 18 to 24 months, Meta will have multiple ways to reduce Muse's cost per task by more than 80%. If this goal is achieved, Muse is expected to achieve profits through the “free+paid value-added” Freemium model while operating independently.

In terms of commercialization, Mathivanan believes Meta has multiple paths, including charging subscription fees for high-frequency or advanced feature users and receiving commissions from commercial transactions facilitated by AI agents. He said that although Meta may pay more attention to Muse's user growth and usage scale at this stage, the continued reduction in costs will create more room for future commercialization.