Moments before 2025 came to a close the hottest trend of the year had one more chance to make headlines. Last week, Meta acquired Manus AI for what some claim to be more than $2B. That's amazing for a company that launched in March.
Manus isn't an LLM but rather an AI agent that's more autonomous than the chatbots many call agents today, and it's grown rapidly after its official launch in March.
So, why did Meta splurge on something it can probably build? Many say it's the tech, but I think Meta bought the one competitive moat money can't replicate - usage data.
Manus AI shipped its mobile apps in late March, a few weeks after the platform's web launch. The invite-only launch drove an explosive 677K downloads in April, peaking at 1M in May before stabilizing at 250K monthly since September, according to Appfigures Intelligence.
Our estimates show Manus AI was downloaded 4.5M times from the App Store and Google Play by the year's end, with even distribution between platforms - a sign of genuine demand rather than hype.
The majority of downloads aren't coming from the US - and neither is the revenue.
Brazil is Manus AI's biggest market by downloads, driving 1.4M downloads in 2025, nearly triple the US's 471K. Egypt, India, and Saudi Arabia round out the top five. That's not a common mix for AI apps, and it signals untapped potential in the US, the world's highest-spending AI market.
We see many apps that get downloads but can't monetize them. That's not the case for Manus. At all.
Manus generated $740K in its first month, doubled to $1.5M by August, and hit $2M in December - growing every month except September.
Appfigures Intelligence shows Manus earned a little over $13M in before-fee revenue in 2025. For context, we estimate that Microsoft Copilot earned just $5.4M in 2025.
Based on these downloads and revenue, Manus AI sees roughly $2.92 per download. That's an impressive number that sits right between ChatGPT's $4.88 and Grok's $1.22. I know these aren't identical, but it's interesting to see it converting better than other more recognized AI apps.
Meta didn't just buy technology or revenue - they bought the one thing money alone can't build: real-world usage patterns from millions of paying users. That's the competitive moat that justifies a $2B price tag.
It's not the first time Meta has done that.
I expect to see more 3rd party agents springing up in Manus' wake this year, and more acquisitions chasing the same unreplicable asset.
Less than 3 years after launching its mobile apps, ChatGPT has reached $3B in consumer spending. Here's how OpenAI turned curiosity into commerce.
Mobile game revenue fell $100M in November as top earners took a hit, but the long tail grew $45M. See which games dominated earnings and what's driving the shift.