The Bargain With Your Killer

Based on the research ofMaoxia Zeng and Zhuoxin (Allen) Li, "Request Taxi Service with Uber? An Empirical Study of Traditional Businesses Partnering with Their Platform Rivals," International Conference on Information Systems (ICIS), 2025

In March 2022, Uber announced it would list New York City's yellow cabs on its own app, letting riders hail a medallion taxi through the same interface Uber had used to bleed the taxi industry of half its business over the previous eight years. Taxi drivers, dispatched through the Curb and Arro apps, would now pick up Uber-originated fares alongside their own. Bhairavi Desai, head of the New York Taxi Workers Alliance, greeted the news with open skepticism: "After its business model has shown the failures to protect drivers from ridership downturns and rising gas prices, Uber is returning to its roots: yellow cabs." An industry analyst covering Uber for Argus Research called the arrangement "pure genius" in the same breath. Both were right, and a new empirical study of exactly what happened next shows why the instinct to distrust the deal and the instinct to take it were both defensible, the outcome hinges on terms nobody puts in a press release.

The Rivalry That Became a Roommate Situation

Standard platform strategy has a simple rule: don't feed the company trying to eat you. Multihoming research treats a complementor's decision to plug into a rival platform as fraught precisely because it hands the rival data, leverage, and a permanent foothold inside your customer relationship. Taxi fleets had every reason to follow that rule. Uber's entrance into New York had already helped push incumbent taxi income down and driven the value of a taxi medallion, once a six-figure retirement asset, toward zero, a collapse severe enough that the city eventually built a $468 million medallion debt-relief program just to keep drivers afloat.

And yet the taxi industry did the opposite of digging in. New York was not a one-off. Uber has run similar taxi integrations in Madrid, Málaga, Valencia, and Barcelona, and partnered with the fleet operator TaxExpress in Colombia, where that single relationship generates roughly half of all Uber taxi trips across Latin America, plus arrangements with taxi and fleet operators in Austria, Germany, Turkey, South Korea, and Hong Kong. Most recently, in November 2025, Lyft struck its own deal with the taxi-dispatch platform Curb, folding taxi drivers into the Lyft app starting in Los Angeles after an earlier pilot in St. Louis, and tapping into Curb's network of more than 100,000 drivers across 65-plus cities. The pattern is not incumbents losing a fight and surrendering. It is incumbents choosing, city after city, to route demand through the platform that a decade earlier was trying to make them obsolete.

What Actually Happens When You List on the Disruptor

Zeng and Li's study gives the clearest answer yet to whether that choice pays off, using the New York integration as a natural experiment. The mechanism runs in three parts. First, listing on Uber increases the e-hail demand taxis receive directly, and that demand spills over into more street hails too, riders who found a taxi through Uber once become more likely to flag one down later. Second, a meaningful share of the fare increase taxis capture comes from Uber's own surge pricing being applied to trips that used to run at flat, regulated taxi rates. Third, Uber itself loses ride volume, both because it is actively referring riders to taxis and because some of that referred demand leaks to rival platforms altogether, but Uber recoups the loss and then some, because the fare increases its market power enables outweigh the lost trips, and those higher fares spill over into what Lyft charges as well.

The net numbers are the whole story compressed into two figures: yellow taxis' net revenue rose 3.8%, and Uber's net revenue rose 10.7%, from the same integration. Both sides made money. One side made roughly three times as much.

Both sides gained from the deal. One side gained about three times more, because it still set the price.

The Ledger Uber Still Wins

The asymmetry is not an accident of scale, and it is not proof the taxis were fooled. It is the direct consequence of who controls the pricing algorithm. Taxis got a real, measurable demand and revenue lift from plugging into Uber's app, the paper is explicit that both traditional firms and the platform gain. But Uber's fares were the ones setting the new price level, Uber's surge logic was what got applied on top of previously flat taxi rates, and Uber's market power is what let those higher prices stick across the market, dragging Lyft's fares up too. The taxis benefited from a demand channel they did not build and could not fully price. The platform benefited from a supply pool it did not have to own, at a price it still got to set. That is the coopetition trade in miniature: the weaker party gets growth, the stronger party gets growth plus continued control of the lever that produces it, and the paper's own framing makes the cost explicit, consumer welfare falls as prices climb, even as both firms report gains.

This is also, per the paper and the New York experience since, not a static outcome but a bargaining position that can be renegotiated after the fact. Drivers who route their work through a platform they don't control are exposed to decisions made unilaterally on the other side of the deal. New York's Taxi and Limousine Commission had to step in directly with Uber and Lyft in 2024 to roll back "lockouts", access restrictions that had been cutting into for-hire driver earnings since that May, a reminder that whoever runs the matching algorithm can throttle who gets work, taxi-integrated or not, until a regulator or a union forces a renegotiation. The demand channel is real. So is the fact that someone else holds the tap.

This Is Becoming the Default Playbook, Not an Exception

What makes the Uber-taxi case worth generalizing from is that it has already generalized. Lyft folding taxi drivers into its own app through Curb in late 2025 is the same coopetition bet playing out from the other side of the historic rivalry, on a five-year lag from Uber's first taxi deals and with the same structural question sitting underneath it: who sets the price once the two networks are merged into one interface. Any traditional business facing a platform disruptor now has a template that did not exist a decade ago, not "resist or die," but "join on terms you can actually defend." The taxi industry's answer was not naive capitulation. It was a bet, made with open eyes about who the partner was, that a demand channel worth 3.8% in new revenue was worth the risk of a partner that would keep 10.7% for itself. Whether that bet was correctly priced is not a question the taxis get to answer once. It is a question the terms of the deal answer every time the algorithm updates.