Waymo leads today, Tesla is still mostly narrative, Lyft is stuck in between, and by the end of 2027 Uber may sit in the best economic position of all.
Great summary. I think Waymo’s model with Lyft in Nashville will be the future: capture first party riders who are willing to pay more and wait longer directly, and use demand aggregators like Lyft and/or Uber to fill in the gaps.
That said, near to medium term, I’d be a bit concerned about Uber’s partnership with Waymo. Waymo has already announced a dozen-plus markets beyond Austin and Atlanta without Uber, which feels pretty telling.
I previously thought exactly this about Uber. But I am increasingly doubting the "network" thesis. Besides supply becoming more aggregated, I wonder what would stop me from implementing my own agent that looks into, say, 3 different platforms to see which one offers the best price (e.g., Uber or Waymo). I think that in works of Agentic AI, and I guess increasing use of "personal assistant" agents, this is not a hard thing to imagine. Not sure if you (or others) have thought about this, but very keen to hear your thoughts.
Great summary. I think Waymo’s model with Lyft in Nashville will be the future: capture first party riders who are willing to pay more and wait longer directly, and use demand aggregators like Lyft and/or Uber to fill in the gaps.
That said, near to medium term, I’d be a bit concerned about Uber’s partnership with Waymo. Waymo has already announced a dozen-plus markets beyond Austin and Atlanta without Uber, which feels pretty telling.
Thanks for this... Just a quick Q if I may...
What do you think Uber's blended take rate on AV rides looks like vs. human-driver rides, and is it durable as partners scale?
I previously thought exactly this about Uber. But I am increasingly doubting the "network" thesis. Besides supply becoming more aggregated, I wonder what would stop me from implementing my own agent that looks into, say, 3 different platforms to see which one offers the best price (e.g., Uber or Waymo). I think that in works of Agentic AI, and I guess increasing use of "personal assistant" agents, this is not a hard thing to imagine. Not sure if you (or others) have thought about this, but very keen to hear your thoughts.
Great article...