Meta hires MongoDB's chief to sell its AI to companies and find a return on its spending
Meta has launched an enterprise platform and recruited CJ Desai from MongoDB to run it. The move is an attempt to give its vast AI budget a second source of revenue beyond advertising.
Meta Platforms on Monday, September 28, announced Meta Enterprise Platform, a new division that will sell its artificial intelligence models, agents and developer tools to businesses, and named Chirantan "CJ" Desai, until that morning the president and chief executive of MongoDB, as its head. Mark Zuckerberg described the unit as "the next major pillar of our business". Mr Desai takes the title of chief enterprise platform officer and reports directly to Mr Zuckerberg.
The announcement matters for two reasons. It is the clearest admission yet from Meta that advertising alone may not be enough to justify the scale of its AI investment. And it puts one of the best financed companies in the world into direct competition with Microsoft, Salesforce, ServiceNow and a large crop of venture-backed start-ups that sell AI agents to corporate buyers.
What Meta actually announced
According to Meta's own post, the platform will at first package the company's existing technology stack for businesses and developers: the Muse agent, Meta Business Agent, the Muse API and Muse Code, a set of coding tools. Muse, a personal agent that can carry out tasks such as booking travel, shopping and sending emails, was launched to US consumers on September 8. Meta Business Agent, aimed at companies that deal with customers over WhatsApp, Messenger and Instagram, followed an earlier release this year.
Mr Desai said the division would focus on turning the company's AI stack into products and services that companies can deploy, with security and privacy built in from the start. Mr Zuckerberg pointed to Meta's "advanced models, leading agents, large-scale infrastructure" as the assets it would draw on.
What the announcement did not contain is just as telling. Meta disclosed no pricing, no enterprise launch date, no named customers and no revenue target. For now, Meta Enterprise Platform is a statement of intent backed by a senior hire rather than a business with a measurable footprint.
The arithmetic behind the pivot
The motive becomes clearer in Meta's second quarter results, published on July 29. Revenue rose 28 per cent to $60.80bn, but advertising supplied $59.36bn of it, or about 97.6 per cent. Reality Labs, the hardware and virtual reality unit that was once meant to be the second leg of the business, brought in $431m and lost $4.62bn in the same three months.
Meanwhile the cost of the AI build-out has climbed steeply. Capital expenditure, including principal payments on finance leases, reached $31.08bn in the quarter, and Meta guided to between $130bn and $145bn for the full year, having lifted the bottom of that range from $125bn. Free cash flow in the quarter fell to $784m. The company ended June with $90.26bn of cash and marketable securities, against long-term debt of $83.66bn.
Those figures explain why investors have been asking what Meta gets for its spending beyond better ad targeting. Microsoft, Amazon and Alphabet can point to cloud divisions that rent the same infrastructure to outside customers. Meta has had no such outlet. An enterprise business, even a modest one at first, would let it present its data centres as a platform that earns revenue directly rather than a cost centre that supports the feed.
A difficult record in the enterprise
Meta does not arrive with a clean slate. Workplace, its corporate collaboration product launched in 2016, was wound down this year after Meta announced its closure in 2024. Horizon Workrooms, a virtual reality meeting app, was retired in February. Corporate technology buyers sign multi-year contracts and expect vendors to support products for a decade; they will notice that Meta has walked away from two enterprise efforts within the past year.
Trust is the other obstacle. Selling agents that read company email, touch payment systems and act on behalf of employees requires a reputation for handling data carefully. Meta's history on privacy is long and well documented, and early coverage of Muse has already raised security questions. The decision to stress security and privacy in the launch messaging suggests the company knows where the objections will come from.
Its advantage lies elsewhere. Meta runs messaging channels used by billions of people and already counts millions of businesses as advertisers. For companies that sell to consumers, an agent that answers customers on WhatsApp and is billed through the same relationship as their ad account is a plausible product. The harder market is the back office of large corporations, where Microsoft and Salesforce are entrenched.
The cost to MongoDB
The hire also carried a price for the company Mr Desai left. MongoDB shares fell about 18 per cent in morning trading on Monday. Mr Desai had run the database company only since November 2025, when he replaced long-serving chief executive Dev Ittycheria. Mr Ittycheria, who had moved to the board, returns as interim president and chief executive while MongoDB searches for a permanent successor. Analysts at Piper Sandler argued that MongoDB "remains well-positioned strategically" and called the reaction "overdone".
The episode shows how fierce the competition for executives who can sell AI to corporations has become. A chief executive leaving a listed company of MongoDB's size after less than a year to take a divisional role at a larger group is unusual. It suggests Meta was prepared to pay heavily, and that Mr Desai judged a newly created unit at Meta a larger opportunity than running an independent public company.
What it means for start-ups
For venture-backed companies building AI agents for business, the arrival of Meta is a mixed signal. On one side, a new rival with effectively unlimited compute and distribution through its messaging apps could compress prices, particularly in customer service and marketing automation, where Meta's channels are strongest. Start-ups whose product is essentially a wrapper around a large model and a messaging integration look most exposed.
On the other side, Meta's move confirms that enterprise agents are a market the largest platforms consider worth fighting for, which supports the valuations investors have been paying. It may also create acquisition demand: a company that has to build an enterprise sales organisation, compliance processes and industry-specific products from scratch often finds it faster to buy them. Start-ups with deep integrations into corporate systems, security credentials and real customer contracts have assets Meta currently lacks.
Our view
Meta's enterprise push is rational and overdue. A company spending more than $130bn a year on capital projects cannot rely on a single revenue stream indefinitely, and the hire of a proven enterprise operator is a serious signal rather than a public relations gesture. But announcements are cheap, and this one came without prices, customers or dates. The first test is whether Meta publishes a credible commercial offer within months rather than years, and whether it commits publicly to long support periods that buyers can hold it to.
For founders, the lesson is to compete where Meta is weak: regulated industries, deep system integration and data governance. Building on top of Meta's channels is reasonable, but depending on them alone is risky. For investors, the useful metric will be any disclosure of enterprise revenue in Meta's coming results. Until that appears, the platform is an option on future revenue, not a reason to change how Meta's spending is judged. MongoDB's sharp fall is also a reminder that key-person risk at software companies remains real and is rarely priced in advance.
Sources
- Meta: Launching Meta Enterprise Platform
- The Star (Reuters): Meta taps MongoDB CEO Desai to drive enterprise AI push
- The Stack: Meta makes its enterprise ambitions official, poaches MongoDB CEO Desai
- Forkast: Meta bets its future on an enterprise platform that does not exist yet
- Runtime Wire: Meta puts Muse models, agents and coding tools under a new enterprise platform
- SEC: Meta Platforms Form 8-K, Exhibit 99.1, second quarter 2026 results
Filed under Big Tech. Written by InvestTimes Editorial under our editorial standards. Commentary and analysis for information only, not investment advice.
