Meta Platforms is plowing ahead in the AI arms race, snapping up fast-rising startup Manus for more than $2 billion in a lightning deal that delivers a revenue-generating AI agent business — and draws fresh scrutiny over the new acquisition’s Chinese roots.
It was generating more than $125 million in annual revenue and was valued at $500 million in its last funding round before Meta swooped in with an all-cash buyout of existing investors.
Meta, led by CEO Mark Zuckerberg, acquired a Singapore-based AI startup for more than $2 billion. AP
Meta has moved to neutralize those concerns, insisting the acquisition leaves no residual China exposure.
The company told The Post all Chinese investors were fully bought out, Manus will shut down its China-facing products and operations, and employees based in China will be relocated or cut off from sensitive systems.
The company has also said Manus staff joining Meta will not have access to customer data and that its AI models will remain geo-fenced — steps aimed at heading off a potential national security review and keeping regulators at bay.
Manus was founded by Chinese entrepreneurs and originally operated out of Beijing under the startup Butterfly Effect. Instagram / @manusaiofficial
“Meta’s acquisition of Manus AI will enable us to provide the most advanced technology to our users with safeguards in place to eliminate areas of potential risk,” a Meta spokesperson told The Post.
“There will be no continuing Chinese ownership interests in Manus AI following the transaction, and Manus AI will discontinue its services and operations in China.”
Manus was generating more than $125 million in annualized revenue and was valued at $500 million in its last funding round. Instagram / @manusaiofficial
Still, “Scrutiny is almost guaranteed; anything with Chinese roots and ‘AI’ in the headline now triggers Washington’s reflexes,” analyst Jeremy Goldman, senior director at Emarketer, told Reuters.
The scrutiny comes as Meta has been pouring tens of billions of dollars into AI infrastructure and talent, making the Manus deal both a strategic bet on autonomous AI agents and a potential political test case for how far US tech giants can go when foreign roots collide with national security sensitivities.
Meta has said Manus will continue operating its subscription service as a standalone product after the acquisition, even as its engineers are folded into Meta’s broader AI teams to accelerate work on autonomous agents across Facebook, Instagram, WhatsApp and the company’s Meta AI assistant.
The deal was struck in roughly 10 days, people familiar with the matter told Bloomberg News, as Meta raced to lock up a fast-growing AI agent business that executives believe can immediately bolster both its product roadmap and revenue base.
The Manus acquisition is just one piece of a far broader blitz in the AI space.
The company is forecasting $70 billion to $72 billion in capital spending in 2025 alone, with internal expectations that outlays will top $100 billion in 2026.
Zuckerberg has pledged more than $600 billion in US investment by 2028.
In recent months, Meta has dangled compensation packages worth up to $300 million over four years — and in some cases as much as $1.5 billion over six years — to pry elite AI talent away from OpenAI, Google and Apple in an escalating Silicon Valley arms race.
BEIJING, April 2 (Reuters) - The Chinese government supports companies with transnational operations and technology cooperation deals according to needs and the law, commerce ministry spokesperson He Yadong said on Thursday.
He's remarks were in response to a question on what measures China would take regarding Meta's acquisition of Chinese artificial intelligence startup Manus.
China has barred two co-founders of Manus from leaving the country as regulators review whether Meta's $2 billion purchase of the firm violated investment rules, the Financial Times reported in March.
(Reporting by Joe Cash and Beijing newsroom; Editing by Christopher Cushing)
It typically gets you a comradely laugh. Or maybe an obligatory chuckle.
But thanks to AI, we can now pose it as a genuine question to Silicon Valley CEOs.
Case in point: Mark Zuckerberg.
According to The Wall Street Journal, the Meta Platforms Inc. (META) CEO is building his very own AI “CEO agent” to help him with the job. That’s right – AI is moving from supporting employees to taking on CEO-level responsibilities.
The agent is still in development, but Zuckerberg is already using it in limited ways. It helps him retrieve information quickly, a task that would typically require going through multiple teams or layers of people. It’s meant to speed up decision-making and act like a high-level internal assistant.
So, Zuck: Working hard or hardly working?
As Meta’s CEO pushes deeper into AI agents, the implications go far beyond that social media company
In today’s Smart Money, we’ll unpack how this shift is redefining Meta’s broader AI strategy and further examine the widespread deployment of AI agents.
Then, before AI agents enter every C-suite, I’ll reveal how the best course of action is not to invest in the firms developing those agents… but in the companies powering them.
Let’s dive in…
Zuckerberg’s New Right-Hand “Man”
Meta, like many companies worldwide, already uses AI to speed up its operations.
In addition to using AI to complete employee performance reviews (I think everyone’s doing this), it’s encouraging its workers to create their own AI tools to assist with everyday tasks. The tech giant recently acquired Manus, a Singapore-based agentic AI startup, which is already being used by Meta employees to do just that.
But that’s just the beginning. The company’s goal is for every employee and, eventually, every Facebook, Instagram, and WhatsApp user to have a personal AI agent.
The reason for this widespread adoption, up to and including the CEO himself, is simple: to remain competitive with AI startups that rely heavily on automation, have smaller staffs, and move faster as a result.
From Susan Li, Meta’s chief financial officer:
Making sure that we don’t – for a company at the size and scale that we are – that we don’t work any less efficiently than companies that are AI native from the start, that’s something that I think about a lot.
Meta is far from alone in striving for AI-powered efficiency.
In an interview last week with MIT Technology Review, OpenAI Chief Scientist Jakub Pachocki discussed the company’s ongoing project to build a fully automated researcher, which aligns with CEO Sam Altman’s timeline of its models acting as research interns by 2028.
These efforts from OpenAI show that AI is redefining what organizations are capable of producing in the first place, expanding the frontier of what’s possible.
AI is being built to cover responsibilities from intern level to CEO level. It’s only a matter of time before it comes for the “dad jokes.”
Of course, large scale AI implementation has serious consequences, especially for job security. Meta’s AI-related layoffs began in 2022, when it cut around 13% of its workforce. And it looks like that may be ramping back up. The Information reported earlier this week that Meta is chopping a few hundred more jobs today.
But here’s the thing: As the company’s workforce shrinks and its AI use grows, that means so, too, is its capital spending.
The Paradox of AI Efficiency: It Isn’t Cheap
Over the last three years, Meta has spent roughly $140 billion on AI infrastructure and plans to spend up to $135 billion this year alone. By 2028, it expects to have spent $600 billion on AI data centers.
Additionally, just this month, Facebook’s parent company announced its $27 billion acquisition of cloud provider Nebius Group NV (NBIS) to meet its compute needs.
Of course, Zuckerberg isn’t alone in spending more… in the hopes of ultimately spending less.
Last week, multiple outlets reported that Jeff Bezos aims to raise $100 billion for a new fund that would purchase manufacturing companies in the chipmaking, defense, and aerospace sectors, “and seek to use AI technology to accelerate their path to automation.”
AI has become essential for these tech giants to reach their ambitions. But that also means the physical building blocks behind AI – like raw materials, energy, and memory – are just as essential. AI agents in particular have massive need of these components.
But, as we’ve been sounding the alarm on, these resources are increasingly in short supply. And shortages of these key physical components are creating bottlenecks that are rapidly reshaping the entire investment landscape.
I believe Main Street investors have a narrow window to act before it’s too late.
P.S. To show my appreciation to all of you Smart Money readers, I’ve put together a brand-new, free special report: Meet the Mag 7 Killers: 3 Heavy Assets Crushing Al Hyperscalers in 2026.
Meta Platforms testing AI agent to streamline executive decision-making Proactive uses images sourced from Shutterstock
Meta Platforms Inc (NASDAQ:META, XETRA:FB2A, SIX:FB) is developing a personal artificial intelligence tool for CEO Mark Zuckerberg as part of a broader push to integrate AI into internal operations and enhance productivity, according to a Wall Street Journal report.
Per people familiar with the matter and internal documents reviewed by the publication, the agent is designed to allow Zuckerberg to access information more quickly than traditional channels, bypassing layers of staff and reducing the need for multiple emails or meetings.
The tool can query internal systems and team data to provide synthesized answers to questions that would otherwise require coordination across several groups.
The project is still in development, but is operational as an on-demand interface for company knowledge. It is part of a wider ecosystem of internal AI tools at Meta, including “Second Brain,” which indexes and queries project documents, and personal employee agents such as “My Claw,” which can access work files and chat logs, and even interact with colleagues or other AI agents.
Meta is using these tools to accelerate decision-making, reduce hierarchical structures, and enable smaller teams, or even individual employees, to manage projects that previously required larger groups. Internal messaging emphasizes “AI-driven impact” and positions the adoption of AI tools as a factor in performance evaluation.
Zuckerberg has previously said that 2026 will be a year when AI reshapes the way Meta operates, with the potential to increase output, speed product development, and foster “AI-native” ways of working.
He has also spent more time coding recently, and highlighted AI-focused efforts during the company’s January earnings call, stating that the initiative aims to elevate individual contributors and flatten team structures.
Meta Platforms (NASDAQ:META) said Tuesday it has started work on a data center in Tulsa, Oklahoma, a project worth more than $1 billion, as Meta moves to expand AI infrastructure and add computing capacity.
Meta said the Tulsa site is part of a wider buildout as large technology companies race to secure power and hardware needed for artificial intelligence systems. Meta did not give a completion date.
Meta also said it plans to invest more than $25 million in local infrastructure upgrades, including roads and water systems, around the Tulsa campus. At the construction peak, Meta expects more than 1,000 workers to be on site.
When finished, Meta said the facility should support about 100 jobs. The Tulsa project will be Meta's 28th data center in the United States and its 32nd worldwide.
The investment comes as Meta, like other large tech groups, seeks more room for AI workloads that require large amounts of computing power and electricity. The company has been spending on data center capacity across the U.S. to support those needs.
Goldman Sachs AI Stocks: Top 12 Stocks to Buy. On March 22, 2026, the Wall Street Journal reported that Mark Zuckerberg, CEO of Meta Platforms, Inc. (NASDAQ:META), is developing an artificial intelligence assistant to assist with his executive tasks to accelerate decision-making and information access. The tool is still being developed, and it retrieves replies directly, reducing the need for internal communication layers. Zuckerberg is utilizing the project to test broader AI integration across the corporation as it works to incorporate artificial intelligence into its operations.
Meta Platforms, Inc. (NASDAQ:META) uses AI tools to optimize workflows, flatten organizational structures, and improve productivity among its 78,000 employees. The company encourages more dependence on AI-native solutions while maximizing individual contributors and minimizing management layers. Zuckerberg has stepped up his direct involvement in coding as part of these initiatives. During a January earnings call, he underlined that AI-driven tooling will help staff accomplish jobs more effectively, which supports the firm’s plan to compete with smaller, AI-focused startups.
Meta's (META) CEO Mark Zuckerberg is Developing an AI Agent to Help Him be CEO Tesla and SpaceX Plan to Build a New Chip Factory in Texas
Meta Platforms, Inc. (NASDAQ:META) specializes in the creation of social media applications. It develops technology that enables individuals to connect and share, discover communities, and grow businesses.
While we acknowledge the potential of META as an investment, we believe certain AI stocks offer greater upside potential and carry less downside risk. If you’re looking for an extremely undervalued AI stock that also stands to benefit significantly from Trump-era tariffs and the onshoring trend, see our free report on the best short-term AI stock.
