Meta Is Going After Enterprise AI, and It Hired MongoDB's CEO to Lead the Charge

When Mark Zuckerberg posted on X Monday morning that MongoDB CEO Chirantan "CJ" Desai was joining Meta, Wall Street went straight to the MongoDB ticker. Shares fell more than 26% in early trading, and the day's headlines were mostly about one of the biggest database companies in software losing its chief executive a day before its Investor Day.

The bigger news came first in Zuckerberg's post, though. Meta is launching a new business unit called Meta Enterprise Platform, which he called "the next major pillar of our business" in Meta's announcement. Desai will run it as Chief Enterprise Platform Officer and report directly to Zuckerberg.

Meta makes almost all of its money from advertising. Now it says it wants to sell AI to companies directly, and it went and got one of the most experienced enterprise software executives in the business to make that happen.

What Meta says it's building

Meta hasn't said much yet. There's no pricing and no launch schedule. Zuckerberg's pitch is that Meta has advanced models, agents, huge infrastructure and years of experience working with businesses, and that few competitors have all of those at once.

The first products on the platform, according to Meta, are the Muse agent, Muse API, Muse Code and Meta Business Agent.

Muse is the personal AI agent Meta launched on September 8. Each user's agent runs inside its own virtual machine and can connect to email, calendars, payments and shopping. It has climbed past ChatGPT in Apple's free app rankings. Muse Code is a coding agent Meta released in August. Meta Business Agent went global in June and lets businesses put AI agents on WhatsApp, Instagram and Messenger.

Meta also knows security will be the first thing any CIO asks about. It plans to release something called Muse Confidential VM later this year, which SiliconANGLE reports will wall off customer instances so that even Meta can't see inside them. Customers will get an ongoing audit to check that it works.

Put all of that in one place and you have models, agents, developer tools, APIs and a privacy layer, sold to businesses. That's the shape of a cloud business. Constellation Research made the comparison to AWS and Google Cloud, both of which were dismissed early on as side projects for consumer companies.

Why Desai

A consumer company that wants to sell to large enterprises needs someone who has done that job for years. Desai has done it at three very different companies.

He spent nearly eight years at ServiceNow and became its president and COO. He then ran product and engineering at Cloudflare. In November 2025 he took over MongoDB from Dev Ittycheria, who had led the company for 11 years.

That covers business software, infrastructure, security and developer tools, which is more or less the full list of things Meta now wants to sell. In his post, Zuckerberg pointed to Desai's industry network as something Meta plans to use. Meta already has the models and the data centers. It has never had an enterprise sales organization, and Desai knows how to build one.

What it cost

MongoDB put together a package designed to keep Desai around. According to its 2025 SEC filing, he got a $500,000 base salary and a $2.5 million cash signing bonus paid in two halves, six and twelve months after he started. He also received $32.5 million in stock awards: $15 million in restricted stock vesting over two years and $17.5 million in performance shares tied to MongoDB's share price through 2030. MongoDB's 2026 proxy statement valued his total fiscal 2026 pay at about $52.8 million.

He left most of that behind. TheStreet points out that the second half of the signing bonus was due at his one-year mark in November, and he left about six weeks early. Much of the equity hadn't vested. A source told CNBC that Zuckerberg made the offer to Desai personally and that it was good enough for him to give up both his MongoDB pay and his severance.

Meta hasn't disclosed the terms. For the move to make sense financially, Desai's new package would have to beat what he gave up by a lot. That fits the way Zuckerberg has been hiring. In 2025 Meta paid $14.3 billion for 49% of Scale AI, mostly to bring in Alexandr Wang, whose lab now builds the Muse models.

Can Meta pull this off?

Meta's track record here is poor. It shut down Workplace, its Facebook-style product for companies. It shut down Horizon Workrooms, its VR meeting app. Meta is good at consumer apps and ad auctions. Enterprise customers want procurement processes, security reviews, service agreements and account managers, and Meta has never shown much interest in any of that.

The model question matters too. For a while, Meta's clearest opening with businesses was Llama: a cheap, open model that companies could run and customize themselves. Enterprises were using Llama before Meta reorganized its AI lab. The Muse Spark models behind this new platform are closed. That means Meta is going up against OpenAI, Anthropic, Google and Microsoft on the same terms they compete on, instead of offering the open, low-cost option. If Meta keeps an open model in the lineup and prices aggressively, it has a real angle. If it doesn't, it's one more closed-model vendor in a crowded field.

There are good reasons to think Meta could make it work anyway. It already has relationships with hundreds of millions of businesses through ads, WhatsApp and Instagram, which is a bigger customer list than most enterprise startups ever get. Muse is the first Meta AI product in years that consumers have clearly embraced, and ChatGPT showed how consumer popularity can turn into workplace demand. Meta also needs this. Investors have been pressing for a return on its AI spending, and selling to businesses is the most direct way to earn money from all that infrastructure without relying on ads.

MongoDB takes the hit

The sell-off wiped out roughly $8 billion in MongoDB's market value. The stock had already lost close to 25% this year. Meta's shares slipped about 4%, which suggests investors have questions about the cost of this plan too.

MongoDB is hosting its Investor Day at the Nasdaq MarketSite in New York on Tuesday. It was supposed to showcase Desai's plans. An interim CEO will present them instead. The board brought Ittycheria back as interim president and CEO and has started looking for a permanent replacement.

The business itself looks the same as it did on Friday. In early September MongoDB reported second-quarter revenue of $771.8 million, up 30% from a year earlier, and raised its full-year forecast to between $2.99 billion and $3.03 billion. The company reaffirmed that guidance on Monday. Ittycheria ran MongoDB for over a decade, and companies building AI applications still need somewhere to store their data. Some analysts say the long-term picture hasn't changed. Tuesday's event will be the first chance to see whether investors agree.

What comes next

Pricing is the first thing to watch. A low price, or an open model option, would tell us Meta plans to compete on cost. After that, look for the first named enterprise customers, since Desai's contacts are a big part of what Meta paid for. Muse Confidential VM needs to ship on time and hold up to scrutiny if Meta wants companies to trust it with their data. And on Tuesday, Ittycheria will have to explain MongoDB's plan and give some sense of how long the CEO search will take.

Meta has tried selling to businesses before and given up. This time Zuckerberg is calling it a core part of the company and has put a proven enterprise executive in charge, reporting to him. We'll find out over the next year whether that's enough.