Uber Eats: A $100B Exaptation.
How Uber's ride-hailing network grew a second, multibillion-dollar function — and what it actually took to find it. EGS Case Study No. 1.
By Shripal Shah, DBA — Founder, Exaptation AI
Most people know the outline. Uber built a ride-hailing app. A few years later it launched Uber Eats. Today food delivery is one of the company's two main engines. What is less well known is how deliberate that second business was, how close it came to being shut down, and what Uber's leaders had to unlearn to make it work.
That is why Uber Eats is the first full case in this series on Exaptation Growth Strategy (EGS). In biology, an exaptation is a trait that evolved for one purpose and later took on a different one. In business, it is the moment an asset a company already owns starts doing a valuable job it was never designed to do. Uber's network of drivers was built to move people. Then it started moving meals.
Case Context: "What Else Is Possible With the Architecture We've Built?"
Uber launched in 2010 as UberCab, a way to summon a black car in San Francisco from your phone. Underneath the app, the company was building something new: a local network of drivers, logged in and moving around a city, matched in real time with people who needed a ride. As more riders joined, more drivers signed up; as more drivers signed up, wait times fell and more riders joined. Economists call this a network effect — a product that becomes more valuable to every user as more users join.
By the end of 2014, four years in, the numbers were staggering: roughly 10 million riders, more than 160,000 drivers, over 200 cities, and nearly $3 billion in venture funding. That capital came with an expectation: growth at a pace ride-hailing alone could not sustain. Uber needed another market as large as the one it had just created.
There were clues about where to look. When Travis Kalanick and Garrett Camp pitched Uber to investors in 2009, urban delivery was already on their list of future markets. By 2013, Grubhub, Postmates, and Instacart were live and Amazon was pushing into last-mile delivery. The market was real. But Uber had no delivery expertise, no restaurant relationships, and no logistics operation.
What it did have was the network. In 2014 Kalanick created a division called Uber Everything — a startup inside the startup — and hired Jason Droege, a former co-founder from his UCLA days, to run it. The mandate was specific: find a service that could become as big as ride-hailing. Kalanick's framing in a 2016 interview captures the mindset — you solve today's problem with an architecture, and "you build a machine to solve the problems that are like it later."
Notice where the search started. Not with a customer segment, and not with a competitor's product. It started with an asset the company already owned and a question about what else it could do.
Case Process: Walk the City, Run the Pilot, Question the Beliefs
Uber Everything did not start from scratch. Uber already ran hundreds of experiments a month and had a routine for testing ideas: deep "jam sessions" on a problem, data from millions of trips, build, repeat. Early bets included Uber Movers (campus moving services, which needed an outside partner), UberRUSH (bicycle couriers, which meant building a whole new supply of messengers), and Corner Store (same-day delivery of household items from a fixed list). None came close to the scale of Rides. RUSH, notably, built a new resource instead of reusing one.
So Droege changed the search. Alongside the data, he spent months walking San Francisco's streets looking for patterns: where people were, what they carried, what they waited for. Many walks produced nothing. Some produced concepts, including UberFRESH, a lunch-delivery idea.
In August 2014 the team ran a 10-day pilot in Santa Monica: a fixed lunch menu (sandwich, salad, soup, pasta), a flat $5 delivery fee, ordered through the existing Uber app and delivered by existing Uber drivers. Nothing new was built. It caught on, and by 2015 the service was in Los Angeles, New York, and Toronto.
Then it stalled. In 2015 Eats was not producing meaningful orders, was losing money, and offered only a narrow set of restaurants. At least one board member asked leadership to consider shutting it down.
The problem was a belief carried over from Rides. Ride-hailing runs on speed: how fast a car arrives, how fast the trip ends, how fast the driver is free again. The Eats team assumed food worked the same way — a burrito at your door in five minutes. It did not. Customers would wait 30 minutes if they could order what they actually wanted.
"We were totally wrong that speed was more important than selection." — Jason Droege, Invest Like the Best, 2022
His broader point was about belief: what a team believes shapes what it prioritizes and, ultimately, what it builds. Eats was not Rides plus food. It was a marketplace with its own rules.
Droege made the case to Kalanick to onboard far more restaurants with full menus, betting that more selection would draw more orders, which would draw more restaurants. Kalanick backed it, reportedly authorizing roughly $300 million in 2015 — even as some investors saw echoes of Webvan and Kozmo, the dot-com delivery failures. The rebuild was code-named Project Agora, Greek for marketplace. By 2016 Uber Eats was a standalone app, sharing Uber's back-end data and driver network with Rides.
Case Results: A Second Engine, Not a Side Project
By the end of 2016 Uber Eats had roughly five million users, and Uber's leadership began to suspect the delivery business might one day be larger than rides. Miami and Atlanta turned profitable in 2017, proof the model could work.
Then the environment shifted hard. In 2020 the pandemic gutted ride demand and pushed households toward ordering in. Eats absorbed the shock. In 2021, Uber's Delivery revenue exceeded its Mobility revenue for the first time. A business that a board member had wanted to close six years earlier was, for a moment, the larger half of the company.
Rides recovered, and both grew. For full-year 2025, Uber reported about $29.7 billion in Mobility revenue and about $17.3 billion in Delivery revenue, on roughly $193 billion of gross bookings across the platform. Just as important, the two functions feed each other: Uber reports that customers who use both Mobility and Delivery generate about three times the bookings of single-product users, and that more than half of first-time Eats customers in late 2025 were new to Uber altogether. The second function is not cannibalizing the first. It is recruiting for it.
Step back and the pattern is clear. Transportation and food delivery are different industries with different customers, competitors, and regulators. But both rest on the same underlying action — a person in a car moving through a city — organized by the same dispatch system. That is the asset. The ride-hailing app was only its first expression. Uber has since extended the same base into freight, grocery, and advertising.
| Uber Eats: the case at a glance | |
|---|---|
| The asset | Uber's platform: the app, real-time dispatch, and a local network of drivers and their cars |
| First function | Moving people around cities (Rides), from 2010 |
| The pressure | Investors expecting another market as big as ride-hailing; delivery start-ups proving that demand existed |
| The search | Uber Everything division (2014); failed bets (Movers, RUSH, Corner Store); months of street-level walkabouts; a 10-day Santa Monica pilot |
| The hard lesson | Selection beats speed. The second function needed its own operating rules, not Rides' rules |
| The protection | Kalanick's reported ~$300M commitment (2015) while a board member pushed to shut Eats down |
| Second function | Delivering meals (Eats); standalone app by 2016; back end shared with Rides |
| The result | Delivery revenue passed Mobility revenue in 2021; ~$17.3B Delivery revenue in 2025 and still growing faster than Rides |
What the Uber Eats Case Demonstrates
1. The asset was bigger than the product. Uber's product was a ride. Its asset was a real-time network of people and vehicles moving through cities. Products are built for one job; assets carry latent jobs — unused and not yet visible. EGS calls the original job the first function and the discovered one the second function, and exists to find second functions on purpose rather than by accident.
2. The search started with the asset, not the customer. "What else is possible with the architecture we've built?" is a different question from "What do our customers want next?" Both are legitimate. Only the first one leads to Uber Eats.
3. Search is a discipline, with failures built in. Movers, RUSH, and Corner Store were not wasted effort; they were the price of discovery. The walkabouts and the 10-day pilot were cheap tests of whether the existing network could do a new job before serious capital was committed.
4. The second function needs its own rules. The speed assumption cost Uber a year. The fix was not a better algorithm but a management team willing to write down what it believed and ask whether it was still true in the new business.
5. Discovery needs protection. Between the pilot and the payoff sat a money-losing year, a skeptical board, and a $300 million decision. Second functions rarely survive without an executive willing to cover them.
6. It was visible, and nobody copied it. Lyft ran a comparable driver network and never built a consumer food-delivery business. DoorDash and Grubhub never moved into moving people. Seeing a second function and executing on it are different capabilities.
One clarification, because it is the most common misreading: this is not an efficiency story. Uber did not squeeze more out of a fixed network. It gave the network a second job, and the network grew into it. That is the difference between doing more with less and discovering you already have more than you thought.
This is what Exaptation Growth Strategy makes repeatable: a structured search of a company's existing assets for their second function, cheap tests of the candidates, and protection for the ones that work. It is also what we are building Exaptation AI to do at scale. Growth, on this view, is a search problem — and Uber's search took about two years, a lot of walking, and one very expensive change of mind.
Next in the series: the Google Search case.
Sources
Carson, B. (2019). Uber's Secret Gold Mine: How Uber Eats Is Turning Into a Billion-Dollar Business to Rival Grubhub. Forbes.
Efrati, A. (2018). How Uber Eats Became a Hit Business. The Information.
Helft, M. (2016). How Travis Kalanick Is Building the Ultimate Transportation Machine. Forbes.
O'Shaughnessy, P. (2022). Jason Droege — Building a Second Act. Invest Like the Best (podcast).
Uber Technologies (2026). Uber Announces Results for Fourth Quarter and Full Year 2025. Uber Investor Relations.
Shah, S. (2024). Exaptation Growth Strategy. Doctor of Business Administration dissertation, Pepperdine Graziadio Business School. EGS Case 3: Uber — Rides to Eats.