Long-term, I think that self-driving taxi operators could largely capture times of peak demand because the depreciation can just be spread out out over the 10-year life of the car rather than only going for full utilization and hitting 300k miles after just 2 years.
But I agree that the the extra capital cost of self-driving taxis is a key limiting factor right now. For this reason I think adoption might only be about one third as fast as the initial adoption of rideshare in the 2010s.
Great charts and generally agree. Though even in the long run, I think the capital and opportunity costs will pressure any fleet operator to aim for maximum utilization possible, even if their vehicles are cheaper that they are today.
That doesn't mean self driving can't get cheap enough that they can handle peak demand in a price competitive way to Uber but obviously will take a while
Thank you. This is a very useful post. Short and sweet. Makes me bullish UBER. I have no position though. I mean I can buy both GOOG and UBER. :)
I got a question: How about Waymo going all in once costs are down enough? They can develop their own marketplace (most people have Gmail / Google Play accounts + add google maps, etc). They can decide to license google maps for commercial purposes and earn royalty, etc. This helps offset some of the costs. I compare this with YouTube (it took a long time). A standalone YouTube company was less likely to pull this feat. What do you think?
Reference 2: "Waymo could decide they will just have much higher ETAs at peak demand (and lower prices) than Uber. But the additional challenge here is that riders have shown they are relatively unwilling to trade off time for prices in ridesharing historically. That is, they probably prefer $20 and 5 min eta over $15 and 20 min eta. This makes the variable demand problem even harder and means Waymo doesn’t have a lot of room to work with."
Eventually, costs will fall enough that their own marketplace with 100% AV is competitive, but by then Uber will likely have other AVs on their platform too (ex: from OEMs other than Tesla). So it's a bit of a race to see how quickly Uber can get competitor AVs on their platform.
Super interesting… I took my first Waymo in December and certainly arrived at the same assumption on AV… hadn’t considered the variable demand piece which is a super fair / interesting argument
But what I would challenge:..their argument assumes that 1) the status quo rider demand dynamics don’t change in line with availability of AVs 2) the AVs cant perform some value function in off-peak hours to ensure supply isn’t left useless (and incur cost of storage all the while)
1) is pretty unpredictable and could take a long time to evolve but 2) feels like a controllable for Waymo/tesla etc
On 2) functionally what would happen is that Waymo is oversupplied in off hours so you have very low ETAs for anyone who rides and lots of cars roaming the streets available. So not necessarily parked somewhere.
But I think they still can’t avoid the math that if you have enough cars to meet the peak demand, you’ll naturally have low utilization of your fleet on average.
Great analysis!
Long-term, I think that self-driving taxi operators could largely capture times of peak demand because the depreciation can just be spread out out over the 10-year life of the car rather than only going for full utilization and hitting 300k miles after just 2 years.
But I agree that the the extra capital cost of self-driving taxis is a key limiting factor right now. For this reason I think adoption might only be about one third as fast as the initial adoption of rideshare in the 2010s.
I built a chart you might appreciate here: https://www.2120insights.com/i/152869747/medium-term-is-the-price-right
Great charts and generally agree. Though even in the long run, I think the capital and opportunity costs will pressure any fleet operator to aim for maximum utilization possible, even if their vehicles are cheaper that they are today.
That doesn't mean self driving can't get cheap enough that they can handle peak demand in a price competitive way to Uber but obviously will take a while
Thank you. This is a very useful post. Short and sweet. Makes me bullish UBER. I have no position though. I mean I can buy both GOOG and UBER. :)
I got a question: How about Waymo going all in once costs are down enough? They can develop their own marketplace (most people have Gmail / Google Play accounts + add google maps, etc). They can decide to license google maps for commercial purposes and earn royalty, etc. This helps offset some of the costs. I compare this with YouTube (it took a long time). A standalone YouTube company was less likely to pull this feat. What do you think?
Reference 2: "Waymo could decide they will just have much higher ETAs at peak demand (and lower prices) than Uber. But the additional challenge here is that riders have shown they are relatively unwilling to trade off time for prices in ridesharing historically. That is, they probably prefer $20 and 5 min eta over $15 and 20 min eta. This makes the variable demand problem even harder and means Waymo doesn’t have a lot of room to work with."
Eventually, costs will fall enough that their own marketplace with 100% AV is competitive, but by then Uber will likely have other AVs on their platform too (ex: from OEMs other than Tesla). So it's a bit of a race to see how quickly Uber can get competitor AVs on their platform.
Thank you.
Wow, what a great article. Even as an AV enthusiast, I had never thought about those variables before, it really opened my mind
Super interesting… I took my first Waymo in December and certainly arrived at the same assumption on AV… hadn’t considered the variable demand piece which is a super fair / interesting argument
But what I would challenge:..their argument assumes that 1) the status quo rider demand dynamics don’t change in line with availability of AVs 2) the AVs cant perform some value function in off-peak hours to ensure supply isn’t left useless (and incur cost of storage all the while)
1) is pretty unpredictable and could take a long time to evolve but 2) feels like a controllable for Waymo/tesla etc
On 2) functionally what would happen is that Waymo is oversupplied in off hours so you have very low ETAs for anyone who rides and lots of cars roaming the streets available. So not necessarily parked somewhere.
But I think they still can’t avoid the math that if you have enough cars to meet the peak demand, you’ll naturally have low utilization of your fleet on average.
And on 1) can you say more of what you mean?