On paper, buses have a surprising number of advantages compared to trains. They’re lighter, so bridges and overpasses cost less to build. They’re mass-produced at a scale trains simply aren’t, which makes each unit significantly cheaper. The infrastructure is just pavement — no tracks, no signaling systems, no electrification. You can open a corridor in stages rather than waiting for the whole line to be finished. And maintenance is surprisingly more manageable: when a single bus breaks down, you take that bus out of service. When a train breaks down, you have a problem. You can even add or remove vehicles dynamically based on demand, something no train operator in the world can do. You can even repurpose existing roads for a test corridor without building anything from scratch.

You almost have to ask yourself “wait, what are we using trains for?” But then reality kicks in and you remember the most important detail: capacity. We simply can’t use buses with dedicated right-of-way everywhere because there is a large ceiling between the number of riders a bus route can carry compared to a train. A standard articulated bus carries around 150 passengers. Even a well-run Bus Rapid Transit corridor tops out at roughly 4,000–5,000 passengers per hour per direction. A light rail line moves 10,000–20,000. A heavy metro like the NYC subway or Tokyo’s network? Up to 80,000 per hour per direction. That’s not a rounding error — that’s an order of magnitude. Now what if I told you that autonomous buses throw that equation out of the window?

The capacity problem with buses is not really about the vehicles themselves. It’s about human reaction times. Drivers need large following distances because humans are slow to react, inconsistent, and unpredictable. That’s what limits how many buses you can push through a corridor per hour. Autonomous systems don’t have that problem. With reliable AI control, buses can platoon — running close together like the cars of a train, with millisecond reaction times and coordinated acceleration and braking. You’re no longer limited by human reflexes. You’re limited by physics. The gap between bus and train capacity doesn’t disappear, but it shrinks dramatically. More importantly, once society adjusts its expectations to the slightly larger stop distances a safe autonomous system requires, the practical capacity ceiling of an autonomous busway corridor becomes close enough to light rail that the comparison stops being embarrassing.

Now here’s where things get dangerous. Public transportation is not just an engineering problem. The politics are arguably even more important. And they’re amazing when it comes to autonomous buses. Rail projects are notoriously brutal to fund and execute. They require enormous upfront capital, years of disruption, and a political consensus that has to survive multiple election cycles. Most of them don’t survive. The ones that do often come in two or three times over budget and a decade late. Autonomous busways flip that equation. You close some lanes, lay dedicated pavement, and you have something running. Yes, the long-term maintenance cost per passenger-mile is higher than rail — pavement wears faster than track under heavy loads, and that’s a real tradeoff. But politically, upfront cost and deployment speed are what matter. A project that gets built beats a project that doesn’t. Every time. Incremental rollout makes it even easier: start with painted lanes and signal priority, add physical separation later. Each stage delivers value. You don’t need to convince anyone to spend ten billion dollars before a single passenger boards.

They’re not as sexy as a train. They’re not as comfortable, reliable, fast, or elegant — all these things are true. But that doesn’t matter. Autonomous buses don’t win because they’re good, they win because they’re good enough. You essentially create the JavaScript of the transportation industry. A tool which isn’t as good at a technical level, but is just convenient enough that it spreads like a virus.

A virus which starts optimizing. Just like in the web world, because autonomous buses become the low-hanging fruit, people start building around them. Stations get redesigned around fast, precise autonomous docking — cutting dwell times dramatically. Off-board fare collection becomes standard, removing the boarding bottleneck entirely. Corridor-specific vehicles emerge: longer, higher-capacity, purpose-built for dedicated right-of-way rather than general city streets. Pavement materials get engineered specifically for heavy, repetitive bus loads. Signal systems get rebuilt around convoy logic rather than individual vehicles. Scheduling software starts treating a stream of buses almost like packet routing — dynamically dispatching capacity to where demand is in real time. Battery swapping or wireless charging gets embedded in station infrastructure. An entire ecosystem grows up around the platform. At a certain point, you get near enough to native performance for 90% of use cases. All for a fraction of the cost.

You don’t even need to replace trains. In those 10% of cases where capacity is a real bottleneck, you would still want autonomous buses. Big rail projects take ten to twenty years from planning to opening. During that entire construction period, the corridor they’re meant to serve is congested, disrupted, and politically toxic. An autonomous busway can be deployed on existing roads in months, absorbing enough demand to keep the city functional while the train gets built — and crucially, keeping the public on side. Construction kills political will. A pressure valve that keeps things moving protects the project from being cancelled halfway through. The busway doesn’t compete with the train. It’s the thing that gets the train built.

Politics beats physics. Rail may be the native code of mass transit — elegant, efficient, optimal. But the best transit system in the world is the one that actually gets built, actually gets funded, and actually keeps improving because enough people use it. Autonomous busways are not the right answer in theory. They’re the right answer because the world runs on what ships.