In 2025, TriMet had 262 million passenger miles at a system cost of $812 million, for a cost of $3.09 per passenger mile.[1] Fares covered 7.8% of their costs. The other 92.2% came from payroll taxes and federal grants.
For comparison, a Lyft or Uber in the same area would cost you $1-2 per mile. Obviously it's not feasible for all 200k daily riders to take Uber/Lyft, and the Uber/Lyft cost doesn't include externalities like extra traffic, but TriMet is very expensive per passenger mile.
This is an implementation problem, not a problem with the underlying concept.
Public transit like buses and lightrail are significantly more efficient per person than personal vehicles. This is because they can transport many more people for the same amount of space and energy. They also typically run on set tracks, which yields more efficiency gains.
The US is really, really bad at doing public transit. It doesn't help that everything is car centric, which makes public transit much harder.
For example, in your comment you're excluding road cost, but you're including the full system cost of transit. That's a car centric side effect, e.g. we take roads for granted. But the cost of cars also includes the cost of roads, the cost of land under those roads, the cost of parking, etc.
The $812 million figure for 2025 did not include the cost to build the rail system. Nor did it include many other expenses. TriMet's expenditures for this year are $1.185 billion.[1]
If you divide passenger miles for TriMet busses (141,726,107) by the number of revenue miles (21,195,016), you get an average of 6.7 passengers per bus, or around 10% of available seats. For MAX (the train) you get an average of 27.4 passengers per train, or around 16% of available seats. In both cases that's seats, not total capacity including standing room. I realize it's important to provision the system for peak demand, but still this seems very wasteful.
And because road wear scales with the fourth power of axle loading, a bus will typically cause 1,000x more road damage than a car.[2] Assuming every car on the road has only one occupant, this means that, on average, a TriMet bus causes 150x more road wear per occupant. The main externality created by cars is traffic.
I agree with you that public transportation can work. It clearly does in many places. But Portland's public transportation is dysfunctional, and I don't see that changing any time soon. That's why substitutes (even partial substitutes like Waymo) are beneficial. The more options people have for getting around, the better off they'll be.
Public transit is only efficient if people use it. If there’s only 1 person on the bus it is less efficient than a car.
Almost everyone drives so the government needs to pay for roads either way. Public transit on the other hand is easy to cut. What matters here is marginal costs of people using ride shares instead of public transit, not the full cost of driving that would need to be paid either way.
> This is an implementation problem, not a problem with the underlying concept.
I agree. The question remains - why do U.S. municipalities universally and repeatedly fail to successfully implement rapid transit at an efficient price point? Buses, trains, and subways in America have ever-growing budgets (both in absolute and per customer mile terms) with ever-declining quality of service. Just asking for more tax revenue again and again is not the solution.
> buses and lightrail are significantly more efficient per person than personal vehicles
Assuming the transit is fully utilized and the car is mostly not. And maybe that's a good way to look at it. But in Portland the light rail is often well under capacity, and in that case a carpool likely wins on efficiency.
> the cost of cars also includes the cost of roads, the cost of land under those roads, the cost of parking
Partially. Those roads will have to exist even if we did not have personal cars.
Train-advocates being against self-driving cars will be recognized as being equivalent to environmentalists being against nuclear power. Fortunately, I don't expect train-advocates as being nearly as successful. Once someone has tried Waymo, there's no going back to the old ways.
> Public transit like buses and lightrail are significantly more efficient per person than personal vehicles.
I would be interested to see a study on that. I see many buses driving around with zero or one passengers on them. If a bus is full, the efficiency would be off the charts. But for a city like Portland, that only happens during commute times. The rest of the time, the buses are driving around empty.
Partially not, as gas taxes cover part of it. I think gas and diesel taxes should cover the full cost of roads, which would help. Still doesn't mean transit should be run so inefficiently.
Oh wow I didn't know Uber solely relied on private roads, had their own DMV, or fleet of millions of cars; truly an innovative company that doesn't rely on public infrastructure!
The cost of providing a bus exceeds the cost of operating a car in many cases, like lower population density neighborhoods. It may save the public money to centralize transit on major corridors and then subsidize trips on Waymo in some areas and at some times.
"Centralize transit on major corridors" is about full buses. But transit agencies spend as much per hour on an empty bus as a full one. Transit agencies run empty buses on routes that are rarely full, and run vans and even microtransit that may just be a waste of money.
> Waymo is an expensive taxi service, not a solution to public transport.
Why not both?
The absolute biggest problems with mass transit in the US are the "first mile" and the "last mile".
If I wanted to take mass transit, I had to show up before 7:00 AM in order to park my car. Every single train after 7:00AM became useless to commuters. That's idiotic.
And then I needed a car at the destination station to drive to my workplace. So, a bunch of us had completely idle cars parked at the commuter station that we used roughly 15 minutes per day but needed parking at both the station AND the workplace--just to use the train. Good lord that is stupid.
Waymo at the right price solves a whole bunch of these issues. Suddenly utilization of your train can go up because you've decoupled train utilization from train station parking. In addition, train utilization isn't so dependent upon close distance to the station. Now, you can build a transit station and allow it to organically fill in instead of getting killed because it's an expensive money sink for 10+ years until housing builds around it. etc.
Sure, you should be able to take a bicycle from the station; that's not how the US is laid out so you have to deal with what you are stuck with today. Sadly, this isn't the old days where everybody works at the mill and dropping a station right there gets you 80% of the population; you have to put that station in and wait a decade while things adjust.
Waymo gets you across the interim while the mass transit convenience transitions from poor to something useful over multiple decades.
In 2025, TriMet had 262 million passenger miles at a system cost of $812 million, for a cost of $3.09 per passenger mile.[1] Fares covered 7.8% of their costs. The other 92.2% came from payroll taxes and federal grants.
For comparison, a Lyft or Uber in the same area would cost you $1-2 per mile. Obviously it's not feasible for all 200k daily riders to take Uber/Lyft, and the Uber/Lyft cost doesn't include externalities like extra traffic, but TriMet is very expensive per passenger mile.
1. https://trimet.org/about/pdf/trimetridership.pdf
This is an implementation problem, not a problem with the underlying concept.
Public transit like buses and lightrail are significantly more efficient per person than personal vehicles. This is because they can transport many more people for the same amount of space and energy. They also typically run on set tracks, which yields more efficiency gains.
The US is really, really bad at doing public transit. It doesn't help that everything is car centric, which makes public transit much harder.
For example, in your comment you're excluding road cost, but you're including the full system cost of transit. That's a car centric side effect, e.g. we take roads for granted. But the cost of cars also includes the cost of roads, the cost of land under those roads, the cost of parking, etc.
The $812 million figure for 2025 did not include the cost to build the rail system. Nor did it include many other expenses. TriMet's expenditures for this year are $1.185 billion.[1]
If you divide passenger miles for TriMet busses (141,726,107) by the number of revenue miles (21,195,016), you get an average of 6.7 passengers per bus, or around 10% of available seats. For MAX (the train) you get an average of 27.4 passengers per train, or around 16% of available seats. In both cases that's seats, not total capacity including standing room. I realize it's important to provision the system for peak demand, but still this seems very wasteful.
And because road wear scales with the fourth power of axle loading, a bus will typically cause 1,000x more road damage than a car.[2] Assuming every car on the road has only one occupant, this means that, on average, a TriMet bus causes 150x more road wear per occupant. The main externality created by cars is traffic.
I agree with you that public transportation can work. It clearly does in many places. But Portland's public transportation is dysfunctional, and I don't see that changing any time soon. That's why substitutes (even partial substitutes like Waymo) are beneficial. The more options people have for getting around, the better off they'll be.
1. https://trimet.org/budget/pdf/2026-adopted-budget.pdf
2. https://www.kgw.com/article/news/verify/yes-bus-more-road-da...
Public transit is only efficient if people use it. If there’s only 1 person on the bus it is less efficient than a car.
Almost everyone drives so the government needs to pay for roads either way. Public transit on the other hand is easy to cut. What matters here is marginal costs of people using ride shares instead of public transit, not the full cost of driving that would need to be paid either way.
> This is an implementation problem, not a problem with the underlying concept.
I agree. The question remains - why do U.S. municipalities universally and repeatedly fail to successfully implement rapid transit at an efficient price point? Buses, trains, and subways in America have ever-growing budgets (both in absolute and per customer mile terms) with ever-declining quality of service. Just asking for more tax revenue again and again is not the solution.
2 replies →
> buses and lightrail are significantly more efficient per person than personal vehicles
Assuming the transit is fully utilized and the car is mostly not. And maybe that's a good way to look at it. But in Portland the light rail is often well under capacity, and in that case a carpool likely wins on efficiency.
> the cost of cars also includes the cost of roads, the cost of land under those roads, the cost of parking
Partially. Those roads will have to exist even if we did not have personal cars.
5 replies →
Train-advocates being against self-driving cars will be recognized as being equivalent to environmentalists being against nuclear power. Fortunately, I don't expect train-advocates as being nearly as successful. Once someone has tried Waymo, there's no going back to the old ways.
2 replies →
> Public transit like buses and lightrail are significantly more efficient per person than personal vehicles.
I would be interested to see a study on that. I see many buses driving around with zero or one passengers on them. If a bus is full, the efficiency would be off the charts. But for a city like Portland, that only happens during commute times. The rest of the time, the buses are driving around empty.
> in your comment you're excluding road cost
Partially not, as gas taxes cover part of it. I think gas and diesel taxes should cover the full cost of roads, which would help. Still doesn't mean transit should be run so inefficiently.
2 replies →
Oh wow I didn't know Uber solely relied on private roads, had their own DMV, or fleet of millions of cars; truly an innovative company that doesn't rely on public infrastructure!
try it again while calculating infrastructure and road costs for 262mm uber/lyft rides
Because roads are a shared resource used by everyone (even non car owners) Uber/Lyft's portion is small and covered by taxes they already pay.
1 reply →
What about infrastructure costs for lyft and uber?
Perhaps it isn’t expensive once you consider the peak load and externalities. How many new roads would you have to build to do that?
The cost of providing a bus exceeds the cost of operating a car in many cases, like lower population density neighborhoods. It may save the public money to centralize transit on major corridors and then subsidize trips on Waymo in some areas and at some times.
> cost of providing a bus exceeds the cost of operating a car in many cases
You can fit 40-50 people in your car?
This only works if the bus is popular. How often do buses in Portland have more than 5 people on them?
how many people can fit in a bus compared to a car?
Doesn't matter if there's only four people willing to ride on a given schedule.
3 replies →
"Centralize transit on major corridors" is about full buses. But transit agencies spend as much per hour on an empty bus as a full one. Transit agencies run empty buses on routes that are rarely full, and run vans and even microtransit that may just be a waste of money.
The OP you're responding suggested using Waymo's to help fill the buses, not get rid of buses.
2 replies →
If it can deliver transit to the public at a reasonable price…
Even five dollars a ride would be twice the price. It's just not comparable.
How many tax dollars go into subsidizing a public transit ride? Varies from place to place but it's not insignificant.
10 replies →
At the margin, it substitutes for some trips.
> Waymo is an expensive taxi service, not a solution to public transport.
Why not both?
The absolute biggest problems with mass transit in the US are the "first mile" and the "last mile".
If I wanted to take mass transit, I had to show up before 7:00 AM in order to park my car. Every single train after 7:00AM became useless to commuters. That's idiotic.
And then I needed a car at the destination station to drive to my workplace. So, a bunch of us had completely idle cars parked at the commuter station that we used roughly 15 minutes per day but needed parking at both the station AND the workplace--just to use the train. Good lord that is stupid.
Waymo at the right price solves a whole bunch of these issues. Suddenly utilization of your train can go up because you've decoupled train utilization from train station parking. In addition, train utilization isn't so dependent upon close distance to the station. Now, you can build a transit station and allow it to organically fill in instead of getting killed because it's an expensive money sink for 10+ years until housing builds around it. etc.
Sure, you should be able to take a bicycle from the station; that's not how the US is laid out so you have to deal with what you are stuck with today. Sadly, this isn't the old days where everybody works at the mill and dropping a station right there gets you 80% of the population; you have to put that station in and wait a decade while things adjust.
Waymo gets you across the interim while the mass transit convenience transitions from poor to something useful over multiple decades.