Re: valuation:

http://www.bloomberg.com/news/articl...erating-losses

8% 7 year convertible bonds. Not exactly free money.

That would be relevant if there was some global scale (like Google, who have to turn petabytes of data into instant response to widely distributed queries, and are basically untouchable via their scaling and algorithms). The taxi service business is a fragmented/localized market; there's no real advantage here.

A single server rack running off-the-shelf routing software tailored for that area can handle a city easily. Taxis do it many cities with pickup times and occupancy rates similar to Uber. Then you have the brains of drivers, world-beating supercomputers with deep local routing knowledge. Then you have the simple fact that 95% of optimal routing is as simple as the nearest driver saying "yes" to the fare of the nearest passenger that comes up, or the passenger where they'll be in three minutes.

You are way, way overvaluing software here. It's basically worthless, and there is no moat here.

All that matters is how many drivers and how many customers they can get in how many areas, before:

- Their massive advantages from underpaying drivers and flouting regulations/labor laws means that the price differential disappears, and
- More taxi medallions/deregulation means that the artificial under supply disappears (which will bring down both prices and wait times).

That's it really. You tech guys are getting hard-ons about stuff that's mostly irrelevant, and missing the big picture here.

So good.

One strange thing is that you can get to 95% of Google the same way you'd get 95% of Uber.

Why real time update? Save yourself hundreds of millions batch scripting.
Why optimize for milliseconds? No need to develop proprietary coding languages/compilers. (Actually most of these are open sourced after development.)
Why create multiple proprietary iterations of a DB? Just use Oracle. Billions saved.

God bless the Luddites. Thousands of engineers, billions of dollars and we have an app worth 50-500k and an antique auto driving car. Oyah. And no money in logistics, since Uber isn't even gonna bother with that.

Software isn't Uber's real moat. Their user base is and name recognition are.

Uber is already a verb. Think about that for a second.

Obviously. Google's moat isn't software either. It is necessary to stay on top of your software so you don't end up like Myspace or Yahoo, but network effects for those businesses and the route density for Uber are their actual moats.

There are at least two macro trends* I think will unfold over the next 3-10 years, and put Uber on the leading edge of autonomous cars. I was only able to write up the first of them today.

TREND #1: There is a Storm Coming to the Car Mobility Value Chain


The standard model
To establish a frame of reference, let's simplify any given value chain into three parts:
***************************
Supplier - Distributor - Customer
Supplier - Distributor - Customer
Supplier - Distributor - Customer
***************************
So to build a great business, you either [1] get a horizontal monopoly in one part of the chain (i.e., own supply/own distribution/own the customer relationship); or [2] integrate two parts of the chain---usually within a vertical---for a differentiated value prop.

Before the Internet, a fundamental fact about the (Distributor - Customer) relationship was that transactions are costly. Physical channels are expensive. To double the size of your channel, you will pay about double the fixed costs; and to double your transaction rate, it'll likely cost you about 2x the SG&A.

In a world where transactions are costly, it's super expensive for a distributor to win by owning the customer relationship at scale. You'll do better by integrating back into supply. The Internet turned that dynamic on its head, and I think mobile proliferation will do much the same, in way that has huge consequences for the auto value chain.

I'll start by highlighting the pattern in an illustrative example that contrasts Wal-Mart and Amazon.

Classical retail strategy
Wal-Mart is the canonical example of pre-Internet retail strategy. They invented vendor-managed inventory, and won suppliers by giving them unprecedented visibility into demand signals (while cutting costs in a million other ways.) Wal-Mart warped the value chain:
***************************
Vendor \ ____ / Shopper
Vendor - WMT- Shopper
Vendor / ____ \ Shopper
***************************
Of course, this was a Faustian bargain for the suppliers, since Wal-Mart now saw a strategic opportunity to own more customers; and that opportunity was the suppliers' margins.

Internet-era retail strategy
Speaking of Amazon. Think about how their strategy in books was fundamentally different---and impossible before the Internet. The bookselling value chain looked like:
***************************
Publisher - Chain - Customer
Publisher - Chain - Customer
Publisher - Indep - Customer
***************************

Amazon came out of right field to integrate directly with customers, at scale.
***************************
Publisher - Chain - Customer \
Publisher - Chain - Customer - AMZN
Publisher - Indep - Customer /
***************************
They COULD do this because on the Internet, transactions are cheap. And they had a reason to do it, because they were creating a truly superior customer experience.

Convenience, selection, lower costs, yes; but also a shopping experience customized for you by machine intelligence, and social-proofed by the crowd. (Very importantly, both the last two are characterized by virtuous cycles. The more users they have, the better they become, the better they attract users...)

Better. Amazon shifted the battlefield to the user experience, where they had the permanent upper hand. And when you own the customers, the suppliers will come---upon which you can take a sledgehammer to their margins, in classical John Henry fashion.

Importantly, Amazon did this starting with zero physical assets. No distribution centers, no cutting-edge robots or logistic networks, nothing but the club of user experience and a violent predisposition.

Mobile-era car-for-hire strategy
We move next to a market the Internet alone couldn't truly disrupt, because adding a channel in every living room wasn't good enough---you need a channel in every pocket/purse!

So, the traditional car-for-hire value chain was approximately:
***************************
Taxi Driver - Taxi Co. - Customer
Taxi Driver - Taxi Co. - Customer
Taxi Driver - Taxi Co. - Customer
***************************
The big winners were the taxi companies able to exploit an artificially high price of medallions to extract huge concessions from the local supply of drivers.

Uber et al. have not so much disrupted this value chain as they have created a parallel value chain where the end product wins on customer experience.

• Faster pick-up times, often less than 5 min.
• Superior coverage density that includes areas where dedicated taxi services are simply not viable.
• Payment simplicity.
• Civility encouraged by the dual-rating system. (Drivers rate passengers, passengers rate drivers.)
• Safety because of the auditability of rides.

Better. Again the suppliers are arguably getting exploited, but likely no more than they would be as traditional taxi drivers. (A labor market most Uber drivers are legally barred from in any case.)

The Uber et al. channel required a proliferation of mobile devices; but since they're here now, transactions are cheap AND omnipresent, and the battleground again shifts to the user experience. The #1 differentiator is coverage density. But there are also analogies to the Amazon.com differentiators: machine intelligence to optimize routing (I reject the premise that there's no room for significant improvement here), and the ratings database adds more value the bigger it gets.

Traditional car-for-hire mobility is dead. Long live car-for-hire mobility.

Mobile-era car mobility strategy
Stepping back, we can think of the car-for-hire market as one piece of a much larger car mobility market.
***************************
Car Manufacturer - Uber/Taxi - Customer
-----------------------------------------------
Car Manufacturer - Rental net - Customer
==========================
Car Manufacturer - Dealership - Customer
Car Manufacturer - Dealership - Customer
Car Manufacturer - Dealership - Customer
***************************

The customers are all buying car mobility. Rental networks and taxis have always competed to an extent; while dealerships have been so differentiated there was little use for such an abstract view. But the mobile-era punchline is now obvious.

***************************
Car Manufacturer - Uber/Taxi - Customer
-----------------------------------------------
Car Manufacturer - Rental net - Customer \
========================== _\
Car Manufacturer - Dealership - Customer -- Uber
Car Manufacturer - Dealership - Customer _/
Car Manufacturer - Dealership - Customer /
***************************
Uber can do this because transactions are now cheap and omnipresent; and it will do this anywhere and everywhere that it can go head-to-head with car ownership and win on the user experience. (Which begins and ends with TCO.)

With each new customer relationship, Uber gets more leverage; and eventually they'll have enough to follow Amazon's path back into the supply chain, looking for more leaks to plug in the user experience. They'll find many of these leaks in their physical fleet. And just like Amazon, Uber will move from the world of bits into the world of atoms---with considerable momentum.

At first, maybe its just a matter of tweaking the fleet composition. Perhaps user preference---for higher safety ratings, or lower emissions, or more luxury, or whatever---is being underserved in certain locales. Faced with the weakening demand at dealerships (amplified by the damage that lower EV service revenues inflict on the dealers' fragile cost structure), car manufacturers will at some point begin to negotiate large fleet deals directly with Uber.

Then Uber will really start to flex, pushing for its manufacturers to build cars that are optimized for fleet membership. This will include a range of hardware redesigns. But much more importantly, it will mean software optimization, via machine learning, that can ONLY be done by direct integration with Uber's vast data assets.



The full consequences of this are tied into my next assumed trend, which I'll probably write up tomorrow.

* H/t to Ben Thompson for a lot of these general ideas.

Lucky that there's only ten cities then. Gentlemen, can you just give me numbers for:

1. How many companies can run development+operations+data mining for the current-scale Uber app?
2. How many companies can run development+operations+data mining for the target-scale Uber app?

It's tempting to read you as saying:

1. All of them.
2. All of them.

hoping someone can answer one very generic question i have about UBER:

is UBER taking market share away from taxis in almost all instances? basically because taxi's are too expensive (taxi business is highly regulated and very profitable for someone who owns the actual tax licence).

OR,

is UBER basically filling in at times when it's almost impossible to get taxis? i.e. big cities when weather is bad, business rush hours, bars closing etc...

i assumed it was the first i.e. that cabs are just too expensive (especially for longer rides) but someone tells me it's the 2nd (times when cabs are hard to get)....

of course it could be both.

thanks in advance

In SF, it seems like they're mostly taking away market share from taxis:

http://time.com/money/3397919/uber-taxis-san-francisco/

Speaking just for myself and friends in the SF Bay Area, we use Uber as a substitute for taxis, public transportation and (less often) walking.

suckerpunch, thanks.....

i looked up UBER's fares and was surprised they weren't as cheap as i thought they'd be..

http://www.businessinsider.com/uber-...y-city-2014-10

basically taxis are 0.9 to 1.7 times expensive as UBER (looks like 1.0 to 1.5x for a trimmed mean).

so you tip a taxi driver 20% on a $10 to $20 fare but a UBER driver nothing. is tipping not permitted?

as much as it would be nice to get a 30% cost savings on a $13 fare, my interest is much more situations where taxis seem ridiculous, like on $80 airport fares.

not sure this completely related to overpriced taxi service but i paid a $30 fare from toronto airport to an airport hotel..... the issue was just as much calling X an airport hotel (but they called themselves that and the taxi commission and priceline both had them designated airport hotels).

Tipping is permitted but most people I know don't tip.

Whether it's 20% or 30% less than taxis- I'd guess it's closer to 20%- price is only one of the considerations. Dirty, smelly taxis that you have to hail or call and wait for, which usually come with a neutral to negative interaction with the driver, pale in comparison to Uber or Lyft.

Also, I got tired of typing, but the vertically-integrated new entrants to the car mobility market (Apple, Tesla, Google?) will also accelerate erosion of demand for offerings delivered through the traditional value chain. This will drive the manufacturers into Uber's arms even faster.

Especially if Apple executes well, with their cult-like ownership of the customer relationship, they will hit the traditional value chain like a cluster bomb. (See also: mobile service providers after the iPhone.)

They are also "adding" rides that would have never happened in areas where you'd have little to no hope of catching a taxi and the customers would be driving themselves if it weren't for Uber/Lyft.

Sub,

Are you really even having to ask the prior question regarding the idiocy that has been posted? TS/Brian/etc have proven time and time again they are some of the dumbest posters on technical topics and you ask THAT? I'd reread their posts and seriously think for a while about what you're doing.

Brian,

That is the first comment you have made that isn't totally ******ed! Good work!

sucker,

The commodity they offer is a safe, clean transport from A - B at varying levels of commodization. (Some derp earlier claimed it was no a commodity. He is stupid.) The lack of hassle from drivers scamming you for change. Arguing about tips and generally harassing women along with being dickheads is why people don't like cabs.

That is all common knowledge and I think I'm preaching to the choir why a ride sharing service is a premium to a cab, yet still a commodity. (In the same way clean drinking water is a commodity to whatever you get out of a stream.)

Well, the best BFI posters (especially TS) have a default tone that is extremely [1] contrarian and [2] reductive. I don't think this is an accident.
---
First, our brains have a limited ability to generate a variety of interesting ideas. Even getting to a dichotomy is often hard! (Like F. Scott said, "The test of a first-rate intelligence is the ability to hold two opposed ideas in mind at the same time and still retain the ability to function.")

Our main strategy for getting to at least a dichotomy is to say, "No".
An aggressive, emotional "no" is the most reliable way to generate counterfactuals and learn about the true universe of possibilities.

So I think that either consciously or unconsciously, both TS and BTM2 have emphasized their contrarian instincts as a strategy for generating richer pictures of the adjacent possible.
---
Second, people looking for investment theses have to be very careful what they pay attention to; there's so much noise that small signals won't generate a meaningful edge.

In this instance, I think TS has a prior belief that scale advantages in IT are only meaningful when the service isn't logically segmented across users. Because Uber's service can be segmented per-city, he discounts its scale advantage as a significant moat. (Similarly with SalesForce, etc.)

You are making the mistake of calling the race preseason. A disruptive new business idea is great for picking losers, but useless for picking winners.

Also, your narrative is that Uber will move toward being an extremely low gross margin, capital intensive company?

You're an idiot. Google has a shark infested moat made of lava- it's search algorithm. Keeping that algo years ahead of the competition is why Google has systematically acquired/hired every human capable of doing search algo work they could find for the last 15 years.

Google's brand is ironically almost meaningless. They have a brand because their product is insanely valuable... They could rename the companies search offering ****search(TM) and still have exactly the same market share they have today.

Sure, I understand the psychological basis for some of their behaviors. I also don't think he has any concept of how difficult the last 5% are. (Take Google for example. Is ranking the 22nd-87th top results in real time for "babarian dragon king" really what makes or breaks their model?)

I don't think that Uber is going to stay logically segmented per city. Transport networks have already started to engage intercity travel, but this is where he doesn't understand the fundamentals of the business and leaves him with a really stupid thesis for the entire business.

Part of the reason I drone on about the specifics of what they are trying to do is that it is impossible to analyze the business without understanding what they are actually attempting to do, how they will do it, and how they will sustain that moat. (In some cases it will be Apple making people think iPhones are cool, and can support an obscene gross margin. In others it will be a network effect.)

Don't you guys think uber can make a big improvement in inner-city transport, what seems to be a future proof business, because more and more city's around the world would/are busy with trying to keep the big trucks out of the centre/city, for supplying supermarkets/ warehouses/stores.

And here in holland they are doing tests with electric trucks/boats for supplying retailers/supermarkets/warehouses.

And the environment is getting more and more important.

TS' point I believe is that last 5% doesn't matter. He's wrong, of course, because he fails to see how the network effects produce a virtuous cycle that compounds the lower cost and better service over time.

Technology in general enables massive scale, thereby creating winner-take-all scenarios in a variety of markets. Running any business by the Pareto principle as TS prescribes is a surefire way to lose out to a superior competitor.

You, sir, are the idiot. Surely you realize that Google's ad auction algorithms were far more innovative, and are just as important as their search algorithms, right?

Plus with search ad revenue growth slowing, and Google's wide reach via android, chrome, gmail, maps, etc., it's asinine to claim that their moat is simply the search algorithm. It's their ability to collect and process data in a much broader sense. Which clearly benefits greatly from network effects, as Mihkel stated.

Almost no one switches anything, on any device, even when presented with a vastly superior alternative. Google's moat is a vertically integrated platform. This has fueled their expansion all the way up the ladder from website -> browser -> OS ->device.

You honestly don't even know the ****ing business model of one of the largest companies in the world. That is amazing.

Pretty much. You can afford to get behind a touch, and create a structural moat where switching rarely occurs (google). But this is when someone just comes in with a 10x improvement (facebook over myspace) and takes your lunch money and your shoes.

Perhaps this is why Google bought Waze and is using it for their ride-share app.

https://en.m.wikipedia.org/wiki/Waze

Of course, Google is far too cash and IT resource poor to be capable of coming in and crushing. And customers are loyal! Just like Apple will never win because Blackberry.

So you're admitting that you were disingenuous when suggesting that any old taxi company with a dispatch app could compete with them?