Conversation
"But Tim, doesn't Waymo need to pay someone to manually label every stop sign and pothole?" Tesla has a technology called fleet scale auto-labelling for this purpose. I bet Waymo has something similar!
Quote
Tesla AI
@Tesla_AI
Replying to @Tesla_AI
All of this is enabled by fleet scale auto-labelling. By using video data from multiple trips in the same location, we can reconstruct the entire scene
1:05
Elon Musk and Alex Hormozi: The Powerhouse Partnership's Hypothetical Venture
Imagine Musk's unparalleled technical innovation being combined with Hormozi's mastery of sales, marketing, and social media.
Pushing the boundaries of what's possible while excelling in capturing ...
Taxi rides are around $2/mile, so I figured 5x a typical taxi ride would be a reasonable guess. But what number would you guess?
If I'm not mistaken Google has these cars with camera's for Streetview, that map roads, wouldn't that be useful? In some countries it is open data; see the Netherlands ~150.000 km of roads in open data. A recent project is to get all road signs correct nationaalwegenbestand.nl/nieuws/informa
Waymo's maps are more detailed than the street view maps but I would expect they could add sensors to the street view cars to get the data.
Also doesn’t Google have all these maps already, via Google Maps and StreetView etc? Or are they not suitable for whatever reason?
Waymo's maps are more detailed, so it's possible the regular Google Maps data isn't good enough. But yes they shoudl be able to add some additional sensors to the Street View fleet and get the data they need.
It has to be less than $10/mile. The hardware for mapping isn't especially expensive not is the car. The data processing for that is essentially free compared to e.g. training some perception model.
They’re at 0.5% of the Uber/Lyft geo right now, so they need to scale that up 200x.
Of course, they only have regulatory capture in four states, which they need because they can’t follow basic highway code rules.
It’s not $10 per mile, based purely on how long it takes.
Quote
John Berry (aniccia.bsky.social)
@aniccia
Waymo announced they are offering ridehail in 37 sq miles of Austin. That brings their total ridehail geo to <500 sq miles across 4 metros, which is
<0.5% of US urbanized area (approx Uber & Lyft US geo)
~0.01 of US
Waymo still excludes fwy despite testing since January. x.com/Waymo/status/1…
Show more
Wall Street ain't ready for this...
Coinbase launched Base ~1 year ago
This Layer 2 blockchain has raked in ~$1.2M every week on average
Now Wall Street is FOMOing into crypto.
Front run them by reading Milk Road. 5 minutes. Every day. For free.
They are scaling! 10x over the last year and they are currently expanding to Atlanta, Austin, and La. I think reason they aren’t growing even faster is mostly banal stuff like car manufacturing, depot construction, hiring support staff.
San Francisco is a dense city with high revenue potential per mile. That's why the system isn't in Montana.
Redo the math for a rural area and waymo is in trouble.
Also, road maps are not a one-time expense. They have to be updated.
The problem is that you need to keep the map up to date. Potholes appear, road blocks, changes, etc... So even if it's 430k, it's each time, all the time...
If it costs a competitor (Tesla) $0 to map a similar area, how is this trivial?
You're also not considering the working capital aspects: Waymo needs to come out of pocket a lot of money to buy the cars and put sensors on them.
Tesla *gets paid* by customers to put
Show more
It's trivial because Waymo's annual revenue in SF will be much much larger than $400k.
$100 is $4 milllion. Still trivial relative to the size of the sf ride hail market.
Elon Musk and Alex Hormozi: The Powerhouse Partnership's Hypothetical Venture
Imagine Musk's unparalleled technical innovation being combined with Hormozi's mastery of sales, marketing, and social media.
Pushing the boundaries of what's possible while excelling in capturing ...
Where'd you get $10 from? If you paid *me* to map each and every mile of the SF metro area and keep that map continually updated, I'd charge way more than that.
Once the service is running the commercial fleet can collect the data automatically
I'm not a fsd doubter, but $10 seems like a self serving number to pick. It's undoubtedly higher than that.
They are working on it. Scaling other stuff (cars, depots) takes longer.
I’m truly only asking you to chip in $1 today to boost our fundraising numbers so that when everyone gets a look at our campaign, they see what I see:
A strong, far-reaching grassroots movement made up of thousands of supporters. Can I count on your $1 today?
What percentage of US lane miles is the SF metro area? Trivial relative to the size of the US.
My strong hunch is that you are off by multiple orders of magnitude. may be able to answer more precision, but if you look at the many billions Waymo has spent versus the number of miles Waymos are operating on, your math is highly suspect. They have likely spent
Show more
Dude I’m literally in the city where Waymo has been operating (well!) for a dozen years and there are large portions of it where it can’t operate.
The question is how often you have to map everything? It's more often than once, but is it annually? Weekly?
It depends on how robust the cars are to changes in the underlying conditions.
My understanding is that waymo costs the same as an Uber. If that's true even with widespread expansion it won't fundamentally change anything about transportation. It will be cool but won't be a large effect on the economy.
I’m truly only asking you to chip in $1 today to boost our fundraising numbers so that when everyone gets a look at our campaign, they see what I see:
A strong, far-reaching grassroots movement made up of thousands of supporters. Can I count on your $1 today?
“The SF metro area has 43,000 lane-miles of roads”
Source for this? Because it seems a bit implausible that there is nearly double the roads in SF than the circumference of the Earth
i think that per-mile rate might me off by an order of magnitude but i think the conclusion still seems valid
If autonomous driving was a map issue, the problem would have been solved years if not decades ago.
Well their speed to scale is definitely impacted by the fact that each new market they enter requires human drivers to collect the map data and then they run simulations on how to drive FOR THAT SPECIFIC CAR FOR THAT SPECIFIC MAP. Whenever they change platforms or change
Show more
Save with seamless supply chains in Southern Alberta. Access to markets is unparalleled, with major highways, air, and railways offering ample opportunity to transport goods to a global market.
saiti.ca
#Transportation #RuralInfrastructure #ConnectingRegions
Are you invested in Lidar or Waymo, or just really don’t understand they cannot scale to ever be competitive.
Even if they needed a Waymo car to drive down all ~4M miles of roads in America each day w each car going ~20mph & costing 70¢/mi & needed a full time googler per active car for labeling at $150k/yr pay, that’d be <$340/yr/mi & ~$1.4B/yr nationwide. A pittance vs the market size.
I rode in a Waymo in Scottsdale a few months ago. I first saw a Waymo about 5-6 years ago there and talked to the guy driving. They were mapping. Driving the routes again, and again and again. The cost per mile has to be very high and the mapping needs to be continually updated.
Waymo isn’t scalable yet because of the cost of vehicles/hardware/insurance/regulatory compliance costs…in addition to having to get utilization.
Maps is hilariously bad take as you point out, parent company owns the best maps solution ever created.
they 3d-scan each street 3 times with a manual driver before they let the automatic ones loose. It costs more than $10, but it also creates Street View
Wanting to expand to western Canada?
Consider one of the most resource-rich, business-friendly places in the west. Southern Alberta is strategically located for efficient delivery to over 50m consumers.
saiti.ca
#ConnectingRegions #Transportation #Infrastructure
Do you know how they mapped the roads? How much it actually costs? I have no idea.
streetview should also have shown that it's very doable to scan literally every single street
They are also owned by Google which literally owns Google Maps & Google Earth? I’m sure that brings some of the cost down at least.
Nobody is saying they can’t be profitable in their niches, they just can’t ever be a ubiquitous service.
And they have to remap basically the whole city every year so it’s not like it’s a 1 time cost.
You have no idea what you’re talking about in terms of cost.
Trying working in releasing products in that industry for a decade and then share your thoughts with the world.
You also need license to operate in every city, and every city needs 2 years of safety-driver operated data. Tesla isn't even at step 0 of all these things. It's just a bunch of clueless idiots who take Elon words as gospel
Which cities are you talking about? I've never heard of this licensing regime.
Good to see you point this out, the cost of mapping us a non-issue. Not to mention that the data can be continuously updated by the fleet of waymo taxis.
This Post is from an account you muted.
This Post was deleted by the Post author. Learn more
This Post was deleted by the Post author. Learn more
Can’t they detect that automatically from the fleet as it drives passengers around?
Maryland’s U.S. Senate race is currently TIED. This race WILL decide Senate control in November, which means we’re running out of time to help Angela take the lead.
Will you chip in $10 to defeat Larry Hogan and defend our razor-thin Democratic majority?
Discover more
Sourced from across X
When I tell this story, all Americans
and most British
don’t believe me.
Italy
cannot have startups by design. This is due many items, starting with the untranslable “studio di settore”.
The
tax authorities assume that given x company assets you MUST make a profit 1/5
I think what Zuck realized is that the norm of avoiding ostentatious displays of wealth is entirely about intra-elite social status and common people actually love stuff like this