How much does it cost
to build an app like Uber?

A launch version in one city costs ₹30–80 lakh; a multi-city product built to compete costs ₹1–2 crore. The difference is not more screens. It is better dispatch, pricing, safety and the operations tooling that lets a small team run thousands of trips a day.

These ranges sit inside the on-demand band of our app development cost guide, which prices a first on-demand app at ₹25–80 lakh. Ride-hailing lands at the upper end because it is the most real-time of all on-demand apps: a food order can wait two minutes for a rider; a passenger standing on a roadside cannot.

Launch version in one city

₹30–80 lakh

Rider, driver and operations

3 apps

To a store-approved launch

5–7 months

Typical platform commission per ride

20–30%

The rider app is what everyone imagines. It is the smallest of the three.

The remaining 25–35% is the backend that ties them together — and that is where the real engineering lives.

Here is how a ₹50 lakh launch budget typically splits. Use it to check any quote: if one line is missing, ask why.

Notice what is not on the list: a custom map. Nobody builds their own maps at this stage. You license them — and that is a running cost, not a build cost.

Dispatch is the heart of an app like Uber, and the reason the build costs what it does. When a rider taps "Book", the backend has to find nearby available drivers, rank them by distance and estimated time, offer the trip to one (or a few) with a timeout, handle a decline, and move on — all in a few seconds, while every driver's phone is streaming its location.

Uber built its own grid for this. It divides cities into hexagons with a system called H3, which it uses to optimise pricing and dispatch and later open-sourced³. You can use H3 yourself for free; it makes "which drivers are near this pickup?" and "where is demand outstripping supply?" cheap questions to answer.

What your launch version needs from the core:

• Location updates from online drivers every few seconds, stored in memory, not just a database
• A trip state machine: requested → offered → accepted → arrived → started → completed (or cancelled at any step, with rules)
• Matching by estimated time to pickup, not straight-line distance
• Offer timeouts and fallbacks so a request never sits unanswered
• Fare calculation by distance, time and zone, with a clear estimate before booking
• Masked calling or in-app chat so phone numbers are never exposed
• An SOS flow that reaches your operations team and the rider's emergency contacts

The build is a one-time bill. Every ride after launch carries its own small costs, and at scale they decide whether your commission covers you.

Maps. Since 1 March 2025, Google Maps Platform gives each billing account 10,000 free requests a month per Essentials product, then charges per thousand calls, with volume discounts at higher tiers⁴. A single trip can use several calls — address autocomplete, geocoding, a route and an ETA — so 50,000 trips a month can mean a few lakh calls. Ola Maps advertises 100,000 free calls per API with India-focused data⁵, which is why many Indian mobility start-ups prototype on it. Whichever you choose, keep the map layer behind your own interface so you can switch.

Messages and notifications. OTP SMS for login and pickup, plus push notifications for every trip state. Use app-based OTPs or WhatsApp where users accept them; SMS is the expensive fallback.

Payment fees. Gateways charge a percentage of each online payment. Cash rides avoid it but create collection and reconciliation work.

Servers. Real-time location is chatty. Expect cloud costs to grow roughly with active drivers, not riders.

Traditional ride-hailing platforms take a commission of roughly 20–30% of each fare from drivers⁶. Namma Yatri, which launched in Bengaluru, took a different route: drivers pay a small subscription and keep the whole fare⁷, and the app runs on the open Beckn protocol behind India's ONDC network, with its code open-sourced⁶.

The model you pick changes what you build:

Neither is cheaper to build by default. The subscription model removes split payouts but adds billing and makes driver retention your revenue line.

1. Launch in one city with one vehicle type. Autos or bike taxis in one city is a real business. Multi-city, multi-vehicle on day one is how budgets double.
2. Talk to drivers before riders. Supply is the hard side of the market. Ten driver interviews will reshape your driver app more than any design review.
3. Use cross-platform for the rider app. One Flutter or React Native codebase for Android and iOS saves 25–40%. Keep the driver app Android-first: that is what most drivers carry.
4. License what is not your edge. Maps, payments, OTPs, chat and push are solved problems. Spend your budget on matching, pricing and the driver experience.
5. Build the operations console early. A good console lets three people run a city. A missing one means hiring ten.
6. Instrument the funnel. Track every step from app open to completed ride. The drop-off between "fare shown" and "booked" will tell you more about your pricing than any survey.

Clone scripts promise a ready-made "Uber clone" for a few lakh. They are tempting and occasionally useful as a demo for investors. As a business, they come with problems:

If you go this way, have an independent developer review the code and the licence before you pay.

Plan a quiet pilot with 50–100 drivers in one area before you announce anything. It is the cheapest way to find the bugs only real traffic reveals.

The ranges are Science of Apps editorial estimates, consistent with our app development cost guide: the features each version needs, sized across research, design, development, testing and launch, priced at the hourly bands Indian teams publish, before 18% GST. Map pricing comes from Google's own billing documentation⁴ and Ola Maps' published pricing⁵; Uber's architecture figures come from its engineering blog¹²³. We review these ranges every six months.

Dispatch — the system that matches a ride request to a driver. ETA — estimated time of arrival. Surge pricing — higher fares when demand outstrips supply in an area. Geofence — a virtual boundary used for zones, airports and pricing. Payout — the transfer of a driver's earnings to their bank account. H3 — Uber's open-source hexagonal grid for dividing a map into cells.