According to Fortune, DoorDash, founded in 2013 by Tony Xu and three Stanford classmates, now controls roughly 60% of the U.S. food-delivery market—more than double Uber Eats. The company could surpass $13 billion in revenue in 2025 and sits at No. 394 on the Fortune 500. CEO Xu showed off proprietary mapping tech that guides Dashers on where to park and which building entrance to use, data he claims doesn’t exist on Google Maps or ChatGPT. Xu, who is also a director on Meta’s board and is admired by Mark Zuckerberg, argues that in tech, “if you are not making improvements, you are actually decaying.” CFO Ravi Inukonda, with the company seven years, emphasizes a logistics engine powered by a decade of machine learning.
DoorDash’s real moat isn’t food
Here’s the thing: everyone gets stuck on the delivery wars. Restaurants, fees, Dashers—that’s the battlefield we see. But DoorDash’s long-game strategy is way more interesting. They’re not just delivering pad thai; they’re trying to build what Xu calls “the catalog for the physical world.” Think about that. It’s a wildly ambitious claim. Their mapping data, telling a driver the best spot to park at a specific apartment complex or the side door to use at a hospital, is a hyper-local, logistics-focused layer on top of the basic geography of Google Maps. That’s a proprietary asset that gets more valuable with every single delivery, across any category. It’s what lets them expand into retail, groceries, and beyond. That’s the flywheel. More deliveries mean more data. More data means more efficient deliveries, which wins more customers and merchants. Rinse and repeat.
The AI infrastructure advantage
So, can anyone catch them? A recent analysis by Klover.ai argues they’re uniquely positioned for the AI-driven local commerce future, and it’s hard to disagree. They’ve been running on machine learning for ten years. That’s not just adding a ChatGPT plugin last month; that’s a purpose-built infrastructure designed for rapid, real-time iteration on routing, pricing, and inventory. It’s operational AI, not just conversational AI. When Xu says their data doesn’t exist on ChatGPT, he’s pointing out a crucial difference. LLMs are trained on the public corpus of the internet. DoorDash’s models are trained on a private, real-time stream of millions of physical transactions and movements. That’s a completely different kind of intelligence. And in the race to actually move stuff, not just talk about it, that might be the ultimate edge.
The risks of being on top
But let’s not crown them forever. There’s plenty of risk. Regulatory pressure on gig worker classification isn’t going away. Restaurant and consumer fee fatigue is real. And when you’re the 60% gorilla, everyone is gunning for you. Uber isn’t giving up. New, niche players could chip away at edges. The whole “catalog of the physical world” plan also requires massive, capital-intensive expansion into lower-margin retail categories. It’s a huge bet that their tech advantage in food delivery translates perfectly to delivering electronics or clothes. Still, Xu’s mindset is telling: “Until it’s over all of a sudden.” It’s a recognition that in tech, dominance is fragile. You have to keep building, keep iterating, or you decay. Right now, their data lead and operational tempo suggest they’re still building faster than anyone else. The question is, for how long?
