Manus, which sells an AI agent that can research, code and carry work across websites and business tools, is nearing a $500 million financing at a $4 billion valuation, Bloomberg reported in a story published by Moneycontrol. The round is expected to close soon, according to people familiar with the talks, but the negotiations remain at an early stage and the terms could change. Prospective investors weren’t identified, and Manus and its existing backers didn’t comment.
For business teams, the product turns an assignment into a sequence of actions: researching information, analyzing data, writing and deploying code, or producing finished slides, websites and charts. Manus says it connects with Google Calendar, GitHub, Notion and Slack, while shared workspaces and administrative controls let colleagues contribute to the same work.
The business pairs a per-seat Team subscription with usage credits pooled across a customer’s staff. Active tasks consume credits according to the model tokens, cloud computers and third-party services they require, so complex or lengthy jobs use more credits than short ones. Manus said in December 2025 that it had passed $100 million in annual recurring revenue eight months after launch; it also reported more than $125 million in total revenue run rate, including usage-based and other revenue.
The financing would put a new price on a company that only recently returned to independent operation. Meta agreed to buy Manus for $2 billion, but Beijing ordered the transaction unwound. Bloomberg reported that the founders and existing investors later repurchased Meta’s shares at that same valuation. Manus said Sept. 1 that it had resumed independent operations under its founding team, so the proposed round would double the valuation used in the reported repurchase.
The central tension is that Manus doesn’t train its own foundation models. Its application instead combines model calls with browser automation, code execution and cloud computers to finish a job. Bloomberg reported that general-purpose models such as Anthropic’s Claude and Moonshot’s Kimi are increasingly capable of the same desktop tasks, keeping pressure on agent makers’ prices even as investors discuss valuing Manus at twice its post-Meta level.