Some companies build their way into a new market. Uber just paid $2.3 billion, all cash, to skip that step.
On October 6, Uber agreed to acquire ezCater, the largest online catering marketplace in the US. The deal still needs regulatory approval and is expected to close in the coming months. Most of the coverage frames it as Uber Eats "adding catering". I read it differently: Uber didn't buy a catering company. It bought a customer base, a supply network and a set of operating habits that would have taken years to build, and that is the more interesting lesson for anyone running a delivery platform.
What Uber is actually buying
ezCater is not a startup experiment. Founded in 2007 and based in Boston, it lets companies order food for meetings, events and recurring workplace meals.
| Metric | Figure |
| Deal value | $2.3B, all cash |
| Restaurant partners | 140,000+ across the US |
| Gross bookings (trailing 12 months) | $2.5B+ |
| Growth | High teens, year over year |
| Average order value | $400+ |
| Profitability | Profitable on non-GAAP operating income |
The number that matters most is the last-but-one. A typical food delivery basket is a meal for one or two people; an ezCater order feeds a room. That changes almost everything downstream — margins, operations, and what the customer expects when something goes wrong.
Why buy instead of build
Uber already had every technical ingredient: a restaurant network through Uber Eats, a delivery fleet, payments, and Uber for Business, which more than 300,000 companies use to manage employee rides and meals. On paper, a catering feature is a few quarters of engineering.
In practice, the hard part of B2B catering was never the code. It is trust built order by order with office managers who will not risk a failed lunch for 40 people, and restaurants that have learned to prepare, pack and hand over large orders reliably. Neither can be shipped in a sprint. Buying ezCater gets Uber both on day one — and keeps a profitable, growing business out of a competitor's hands.
This is the same playbook I wrote about in my piece on Uber's service portfolio: find the product the market needs every day, make it the cash engine, then cross-sell. Uber for Business is the cross-sell channel; ezCater gives it something much larger to sell.
Why B2B orders change the economics
Consumer delivery is a volume game with thin margins: the cost of a rider, support and marketing is spread across small baskets, and every promotion eats into the unit economics. Catering flips several of those variables at once.
- Basket size. One $400 order costs nowhere near twenty times a regular order to deliver, so the margin per trip is far higher. Uber expects the deal to be margin accretive for exactly this reason.
- Predictability. Corporate orders are placed hours or days ahead and often recur weekly. Planned demand is cheaper to serve than a dinner-time spike.
- Retention. A company that runs its office lunches through one platform does not churn because a competitor offers a discount code. Switching means retraining staff and changing expense workflows.
- Weekday fill. Office catering peaks at lunch on working days, which uses kitchen and rider capacity in hours that consumer demand often leaves underused.
The trade-off is that the cost of failure rises with the order. A late single meal is a refund. A late lunch for a client workshop is a lost account.
What catering demands from the platform
From an engineering seat, catering is not "delivery with a bigger cart". It stresses different parts of the system, and a platform built for on-demand meals has to change in at least five places.
- Scheduling, not dispatch. Orders arrive days ahead with a delivery window. The system has to reserve kitchen capacity and assign riders or vehicles in advance, then re-plan when something changes.
- Capacity-aware supply. A restaurant that can cook 200 single meals an evening may not handle three 60-person orders at noon. Caterer capacity becomes a first-class data model, not a toggle.
- Accounts with roles. One corporate customer has an admin, several ordering staff, approvers and a finance team. Budgets, approval flows and cost-centre tagging replace the single-user checkout.
- Invoicing over cards. Companies want monthly invoices, tax-ready receipts and expense integrations. Payments become a ledger-and-billing problem, not just a gateway call.
- Reliability you can promise. On-time rates, proactive alerts and a human on the line become part of the product. Observability has to reach the order level, so support knows an order is at risk before the customer does.
None of these is exotic. But each one touches a different service, which is why bolting catering onto a consumer stack usually turns into a quiet re-architecture.
What this means for platforms in South Asia
In markets like Bangladesh and India, office meals are still largely handled offline: a call to a known restaurant, cash or a bank transfer, and an office assistant chasing the order. That looks like an opportunity, and in part it is. But the ezCater story suggests three things to get right before chasing it.
- Win reliability before scale. Corporate buyers forgive a higher price far more easily than a failed delivery. The first 50 accounts served flawlessly are worth more than 500 served unevenly.
- Build the boring parts properly. Invoicing, VAT-compliant receipts, approvals and account management decide whether a finance team signs off. These are the features consumer platforms typically skip.
- Use the consumer network as the moat. A delivery platform already has restaurants, riders and payment rails. The edge over a standalone catering startup is cross-selling into businesses that already use the app at home — exactly the bet Uber is making with Uber for Business.
The question for every regional platform is not whether B2B food will grow. It is whether you build it deliberately now, or end up buying it later at a premium.
How I'm reading it from Foodi's seat in Dhaka
I read this deal the way I'd read it for Foodi. Dhaka's office belt — Gulshan, Banani, Motijheel, Tejgaon — runs on meetings, client visits, training days and celebrations, and most of that food is still ordered the old way: a phone call to a known restaurant, a bank transfer or cash, and someone from admin chasing the rider. The demand is real. What's missing is the system around it.
Three things make Bangladesh different from the US case, and each one is an engineering problem before it is a business one:
- Seasonal peaks that dwarf the average. Corporate iftars in Ramadan, year-end parties, Pohela Boishakh and Puja events concentrate large orders into a few weeks. I saw the same shape at Shohoz during Eid ticketing: you build for the peak, not the average, or the peak breaks you.
- Delivery windows in Dhaka traffic. A promise of "12:30 sharp" for 50 lunches means planning around traffic that can double a trip's time. Scheduling, buffer time and vehicle choice matter more here than in almost any market ezCater serves.
- Finance teams that need paper. Corporate buyers expect VAT-compliant invoices, purchase orders and, often, credit terms. A platform that only takes card or mobile wallet at checkout will not get past procurement.
Where a platform like Foodi has a natural edge is breadth. Foodi already delivers food, groceries, flowers and more, so an office account is not just about lunch: it can cover pantry restocking, event flowers and meeting snacks through one relationship and one invoice. That is the same logic Uber is applying with ezCater and Uber for Business, adapted to how offices in Dhaka actually buy.
The takeaway
Uber paid $2.3 billion for something software alone cannot produce: earned trust with buyers and a supply side trained for large orders. For CTOs, the lesson is that a new vertical is rarely a feature. It is a different customer, a different failure cost and a different set of services underneath.
If you are building or running B2B ordering on a delivery platform in South Asia, I would like to compare notes — especially on scheduling and corporate invoicing, where most of the hard problems sit.
