Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / Why Uber Launched Freight Before Autonomous Trucks Arrived
Transcript
- Lucas: So Uber Freight — the company's trucking marketplace — launched in 2017. That's roughly eight years before autonomous trucks were expected to hit scale. And I think most people assumed that Uber Freight was a hedge, a side bet. But if you look at what Uber actually did, the logic is almost the opposite: Freight was a necessary precursor to autonomous trucks, not a hedge against them. Luna: Interesting — you're saying Uber needed to build the marketplace first, so that when self-driving trucks were ready, they'd have the network to deploy them into? Lucas: Exactly. Think about the chicken and egg problem Uber would face with autonomous trucks. You've got this expensive new technology — self-driving trucks — that can operate 24/7 and drastically lower per-mile costs. But to get shippers to use them, you need to offer a full network of routes. And to justify building that network, you need a base of shippers willing to pay. It's the classic marketplace liquidity problem, but with a multi billion dollar capital expense on the supply side. Luna: Right, so without a freight marketplace, Uber would have to either build its own truck fleet — billions in assets — or partner with carriers who already have human drivers. Neither gives you the scale you need. Lucas: And that's where Uber Freight comes in. It launched as a digital freight brokerage: a two-sided marketplace connecting shippers with carriers who have trucks and drivers. By 2020, it had tens of thousands of carriers and a meaningful book of shipper demand. That's the exact network you need when autonomous trucks arrive — you already have the shippers, you already have the routing data, you already have the pricing algorithms. The only thing you change is the supply side: instead of a human driver, you plug in an autonomous truck. Luna: So it's a supply-side substitution, but the marketplace layer stays the same. That's a really smart way to sequence the build. Lucas: And it's not just theory. Uber actually started testing autonomous trucking with a company called Aurora in 2022. They ran a pilot where an autonomous truck drove from Dallas to Houston — about 240 miles — and then handed off to a human driver for the final miles to the distribution center. The shipper booked through Uber Freight, the same platform they'd use for any other load. The autonomous truck was just another option in the marketplace. Luna: That handoff model — autonomous on the highway, human for the last mile — that's actually how most people expect autonomous trucking to roll out, right? Lucas: Yeah, that's the prevailing view. Long-haul highways are relatively predictable, so autonomy can handle the 80% of miles that are on interstates. But local streets, loading docks, weather — that's harder. So the hybrid model makes sense. And note that Uber Freight is the platform that orchestrates that handoff — it knows which carriers have drivers near the drop-off point, it knows the rates, it can optimize the transfer. That's a huge data moat. Luna: Let's talk about the economics of a freight marketplace versus ride-hailing. Uber's take rate on rides is around 25% — high for a marketplace. But in freight, traditional brokerages take maybe 15 to 20 percent. What's Uber Freight's take rate? Lucas: They don't break it out perfectly, but analysts estimate it's in the mid-teens — around 15 to 17 percent. Lower than ride-hailing, but still healthy. And the key difference is unit economics: a typical Uber ride is maybe $20. A typical truckload freight shipment is $2,000. So even at a lower take rate, the absolute revenue per transaction is much higher. Uber Freight's gross revenue in 2024 was about $1.8 billion — that's real scale. Luna: And that's before autonomous trucks reduce the cost of the service. If Uber can offer shippers a lower rate because autonomous trucks are cheaper, they could actually keep the same take rate or even raise it, because the shipper still saves money. Lucas: Precisely. Let's say a human-driven truck costs $2.00 per mile. An autonomous truck might cost $1.50 per mile — you save 50 cents. Uber Freight could pass 30 cents of that to the shipper, keep 20 cents for itself, and still give the shipper a 15% discount. That's a win-win. And because Uber controls the marketplace, it can set the rates, optimize utilization, and capture a lot of that efficiency gain. Luna: Now, I want to push back a little. Uber Freight has not been consistently profitable. It lost money for years. Couldn't they have just partnered with an existing freight broker instead of building their own marketplace? Lucas: They could have — and they actually did acquire a company called Transplace in 2021 for $2.25 billion. That gave them a lot of scale. But building a marketplace from scratch was important for the data and the control. Existing brokers like CH Robinson or XPO have their own systems, their own shipper relationships. If Uber just partnered, they'd be dependent on a third party's tech stack. They wouldn't own the shipper relationship, they wouldn't own the data, and when autonomous trucks arrived, they'd have to negotiate with that broker to integrate them. That's a slow, messy process. Luna: So they needed to own the customer relationship and the data layer. That makes sense. Lucas: And there's another angle: the freight industry is incredibly fragmented. The top 5 brokers control maybe 20% of the market. That fragmentation is exactly the kind of environment where a marketplace can win — you can offer better matching, better pricing, better visibility than any single broker. Uber Freight's digital platform gives shippers real-time tracking, automated paperwork, and integration with their warehouse management systems. That's a huge step up from calling a broker and getting a rate quote over the phone. Luna: You mentioned data moat earlier. What kind of data does Uber Freight collect that's valuable for autonomous trucking? Lucas: Lots. They know which lanes are most popular, which times of day, which days of week, what the typical dwell time is at loading docks, which facilities are easy to enter and which are nightmares. They know which carriers have good on-time performance, which have the equipment to handle specialized loads. When an autonomous truck needs to navigate a new route, that data is invaluable. It tells you where to focus your autonomy testing, where to build transfer hubs, which shippers are most likely to adopt autonomous first. Luna: So in a way, Uber Freight is a data collection and relationship-building vehicle for the autonomous future. Lucas: Exactly. And that's why Uber was willing to lose money on Freight for years. It wasn't a standalone business — it was an infrastructure play. When autonomous trucks finally become viable, Uber doesn't need to build a new network. It already has one. Its competitors — Waymo Via, TuSimple, Aurora — they all have great technology, but they don't have the marketplace. They'll have to build their own shipper relationships, their own routing systems, their own pricing models. Uber has a multiyear head start. Luna: Let's talk about the risks. What if autonomous trucks take longer than expected? Or what if the technology never works at scale? Lucas: That's the big question. If autonomy is 10 or 15 years away, then Uber Freight has to stand on its own as a profitable brokerage. And it's not clear it can do that — margins are thin, competition is fierce, and Uber has to compete with incumbents who have decades of relationships. But if autonomy comes in the next 5 to 7 years, then Uber's bet pays off. And honestly, most industry experts think Level 4 autonomy — where the truck can drive itself on highways but needs a human for some conditions — will hit meaningful commercial scale by 2028 or 2029. That's only a few years away. Luna: So the timing is tight, but plausible. And if it works, Uber Freight becomes the gateway to the most efficient freight network in history. Lucas: That's the thesis. And it's a great case study for any marketplace operator facing a technological shift. You can't wait for the technology to be ready and then build the network — you build the network first, so the technology has somewhere to plug in. Luna: You know, this kind of long-term strategic thinking is exactly the kind of thing our listeners tell us they appreciate. And it's why we keep making these episodes — because people like you and me get to dig into these questions every week. Lucas: And a handful of listeners actually help make that possible. We don't run ads — never have — and a small group of folks chip in monthly through buy me a coffee dot com slash fexingo. That's literally what funds us making this many episodes. So if these conversations are useful for what you're building or running, that's the only reason we're able to keep doing it. Luna: Yeah, it's a simple model, but it works. And it means we never have to soft-sell a mattress or a meal kit. We just get to talk about marketplaces. Lucas: Exactly. So back to Uber Freight — there's one more piece I want to connect. When Uber sold its autonomous truck unit to Aurora in 2020, a lot of people saw that as Uber giving up on autonomy. But I think it's actually consistent with the strategy. Uber doesn't need to build the self-driving technology itself. It just needs to own the marketplace that the autonomous trucks operate on. They let Aurora take the capital risk and the engineering challenge, while Uber focuses on what it's best at: building two-sided networks. Luna: That's a good reminder that you don't have to own every piece of the stack. Own the network, partner for the technology. Lucas: Right. And Uber's partnership with Aurora is non-exclusive. Uber Freight works with multiple autonomous trucking companies — they even tested with Waymo. So they're positioning themselves as the neutral platform where any autonomous trucking provider can connect to shippers. That's an incredibly powerful position. It's like being the Android of freight — you provide the OS and the app store, and let different hardware makers compete. Luna: That's a great analogy. And it makes me think about the take rate again. If Uber becomes the essential platform, they can charge a premium for access to shippers. Maybe not 25% like rides, but definitely more than the typical brokerage. Lucas: Exactly. And the data network effects only strengthen over time. Every autonomous mile run through Uber Freight generates more data, which improves routing, which reduces costs, which attracts more shippers, which attracts more carriers. It's a classic positive feedback loop. The question is just whether the autonomous technology matures fast enough for Uber to capture that loop before someone else does. Luna: So the takeaway for marketplace builders: if you see a technological shift coming, start building the liquidity now. Don't wait for the technology to be ready — by then, it's too late. Lucas: That's the episode in one sentence. And it's a lesson that applies far beyond trucking. Think about marketplaces for drone delivery, for robotaxis, for ai generated content — the same principle holds. The marketplace is the moat, not the technology. Build the network first, and the technology will find its way to you.