Meta's agent Muse hits 700,000 daily active users in ten days, with service degradation already occurring due to insufficient compute
The daily active user count of Meta's AI agent Muse grew 10x over the past 11 days, reaching about 700,000, but load issues have already emerged, with the status monitoring platform SaaSHub listing its status as service degraded. A researcher stress-tested Muse by asking it to generate 120 sub-agents at once, and only 33 were successfully created. Meta Chief AI Officer Alexander Wang revealed on September 9 that Muse's early usage far exceeded internal expectations, at roughly 10x the level during internal testing, prompting Meta to reset token usage and add limits. The bottleneck may stem from its compute-intensive architecture; Meta promises each user a dedicated virtual machine equipped with 2 virtual CPUs, 8GB of memory and a 100GB SSD. Currently Muse is only launched in the United States and Canada, and registration is not yet open to WhatsApp and Instagram users.
Why read this
AI score 75
Muse's users grew 10x in 11 days but service has already been degraded; readers can use this to observe the scaling bottlenecks of compute-intensive agent architectures.
Topics
Body · AI translation
①Data shows that the number of daily active users of Meta's AI agent Muse grew tenfold over the past 11 days, reaching about 700,000; ②Muse's current user scale is relatively small, but load problems have already emerged, and the status monitoring platform SaaSHub has listed Muse's status as "service degraded"; ③An important reason for the bottleneck may lie in Muse's compute-intensive architecture. Cailian Press, September 25 (Editor: Ma Lan) — Meta's AI agent Muse achieved great success in less than 20 days after release; however, the agent has encountered a key bottleneck in large-scale deployment.
According to Similarweb statistics, Muse's daily active users grew tenfold over the past 11 days, reaching about 700,000. This is an astonishing growth rate, but the overall scale is still relatively small in the AI application field. Even so, Muse has frequently shown signs of being unable to cope.
Increasing evidence shows that Meta is already struggling to meet the needs of Muse's growing user base. The status monitoring platform SaaSHub has listed Muse's status as "service degraded," noting that its monitoring system can still access Muse but continues to report problems, such as frequent replies saying it cannot search.
In addition, a researcher conducted a stress test on Muse, asking it to generate 120 sub-agents at once, but Muse successfully created only 33, and all other attempts failed.
These scalability problems appear especially severe relative to Muse's user scale of less than 1 million. Currently, Meta has launched Muse's service only in the United States and Canada and has not opened registration channels for WhatsApp and Instagram users. If Meta wants to capture further market share, the current computing power bottleneck will become its biggest constraint.
Meta's Chief AI Officer Alexander Wang revealed on September 9 that Muse's early usage far exceeded internal expectations, with user usage about 10 times that during beta testing, prompting Meta to reset token usage and add limits.
An important reason for the bottleneck may lie in Muse's compute-intensive architecture. Meta promises each Muse user a dedicated virtual machine equipped with 2 virtual CPUs, 8GB of memory, and a 100GB solid-state drive.
Previous industry analysis pointed out that assuming Muse's user count reaches 100 million within a year and there is no resource sharing among users, Meta would need at least 200 million CPU cores, equivalent to about 1.58 million 126-core AMD Ryzen EPYC CPUs. In addition, it would need 800PB of memory and 10,000PB of solid-state drives.
A user scale of 100 million is not difficult for AI applications that rank at the top of app stores; the difficulty is how Meta can obtain sufficient hardware support.
On the other hand, this kind of hardware investment may also be unable to balance with Muse's current pricing. Meta currently can give each user up to 100 million tokens free per week, and paid subscription plans start at about $20 per month and go up to $100 per month.