Meta launches enterprise AI unit and hires MongoDB CEO CJ Desai to lead it
By EnkiEdited by VK, Editor
Published
Reporting from TechCrunch, The Decoder
Meta Enterprise Platform will sell the Muse agent, Muse API, Muse Code and Meta Business Agent to companies under former MongoDB chief CJ Desai. Meta has not published enterprise pricing, though its developer API already lists token rates.
What it means for founders
- Another low-cost model vendor is courting business. At $1.25 and $4.25 per million tokens, Muse Spark sits at the cheaper end of current frontier APIs, so teams paying heavily for inference should benchmark it against their current provider on their own tasks.
- Read the contributor tier carefully. The 10 cent rate is tempting for prototypes, but it grants Meta training rights over your prompts, which most customer contracts will not allow for production data.
- Agent startups face a new incumbent. Meta says it already works with hundreds of millions of businesses through its ads and apps, and Muse for Small Business connects straight into them. Founders selling agents to small merchants now compete with a free default.
- Watch for Meta's first enterprise pricing sheet and named launch customers, which will show whether Desai is chasing large IT contracts or bundling AI into existing ad relationships.
The story
Meta has set up a new division to sell its AI to companies and has hired MongoDB chief executive Chirantan "CJ" Desai to run it. In a post announcing Meta Enterprise Platform, Mark Zuckerberg called the unit the next major pillar of the business. Desai takes the title of Chief Enterprise Platform Officer and reports to Zuckerberg directly.
What Meta says it will sell
The opening lineup is made of things Meta already runs: its Muse agent, the Muse API, the Muse Code coding tool and a Business Agent for companies, with more promised later. Muse arrived earlier this month as a consumer agent that handles tasks such as email and travel bookings, and Meta says its enterprise versions will carry the same security and privacy protections from day one.
What Meta did not publish is an enterprise price list, contract terms or a launch date for packaged business plans. The developer side already has public rates. Meta's model API charges $1.25 per million input tokens and $4.25 per million output tokens for Muse Spark 1.3, with nothing used for training. A contributor tier drops that to 10 cents and 20 cents in exchange for letting Meta train on prompts and completions. A day after the enterprise news, Meta also added business skills and connectors to Muse for small firms, keeping the core app free with paid subscriptions on top.
Why Desai, and why now
Before MongoDB, Desai headed engineering and product at Cloudflare, and before that he worked close to eight years at ServiceNow, rising to president and chief operating officer. That is a résumé built on selling infrastructure and workflow software to large IT buyers, a customer Meta has rarely courted outside advertising. The Decoder points to Meta's AI infrastructure budget, above $100 billion for 2026, and to investors who want revenue to show for it. MongoDB shares fell more than 17 percent on the departure, and former chief executive Dev Ittycheria is back as interim leader while the board searches for a successor.
What we don't know yet
Meta has not said how enterprise Muse will be priced per seat or per task, whether large customers get data residency or private deployment, or which markets beyond the US and Canada it will serve. It is also unclear whether the enterprise tier will keep the cheap contributor rates or restrict them for business data.
Sources
Primary sources
Reporting
Enki Daily
Get stories like this every weekday morning.
The day's AI stories for founders, each with what it means for your company. Free.
More in Funding & Business
- AMD agrees to buy Fei-Fei Li's World Labs in $8.2 billion stock deal
For founders: Check your dependency on Marble or Atlas. If you build on World Labs' tools for 3D content, simulation or robot training data, plan for pricing, licensing or…
TechCrunch · 15h ago - Insurers say hospitals' AI billing tools added $942 million in costs
For founders: Revenue share pricing sells, and invites scrutiny. Paying a vendor a cut of recovered revenue makes the purchase easy for a hospital, but it is also the…
TechCrunch · 2d ago - $100 million-plus Series A rounds hit a multiyear high in 2026, led by AI
For founders: The bar for a standout round has moved. When more than 100 companies raise $100 million at Series A, investors benchmark leaders against that scale, which can…
Crunchbase News · 6d ago