Alex Levashov – eCommerce Consultant (Melbourne, Australia)

Why Uber has insanely high valuation or unit economics of self-driven taxi

Why Uber has insanely high valuation or unit economics of self-driven taxi

In the wake of the recent legalisation of Uber in Victoria (Australia) and news about self-driven car test that the company conducting in Pittsburgh, I decided to make a short analysis of self-driven taxi industry prospects.

uber-self-driven-volvo

Uber self-driven Volvo, source: Uber via Bloomberg

The questions I was interested are:

  1. How will introduction of autonomous cabs will change profitability of service providers like Uber?
  2. How it affect ride price consumers pay?

The approach

I decided to calculate the profitability of one self-driven vehicle as an asset with a limited life span. So a business buys it, uses for certain period of time and then disposes (sell). This is close to how taxi and rent car company operate. I admit that it is a bit simplified approach, but I was interested to look into unit economics of the business operating fleet of autonomous cabs, taking a vehicle as a unit.

Assumptions and facts

Let’s list assumptions I made in my analysis

  • Assume that the vehicle cost will be $50,000, it will be used for 200,000 km and then sold for $15,000. The number depends for different vehicle, but quite close to reality, at least in Australia
  • Next assumption is to decide empty miles share – number of miles/kilometres that a vehicle runs without passengers, so nobody pays for the ride. Different sources quote different numbers for ride-sharing and taxi industries. Some researchers suggests that with a pool of autonomous cars the efficiency of up to 10% empty miles can be achieved, but I took more conservative number, 25%. Note that empty miles is different from empty time or utilisation, a vehicle can stand waiting, it doesn’t have to move
  • Petrol consumption and cost. I took quite conservative 12 litres/100 km consumption and $1.2 per litre cost
  • Car service cost. I took $4,000 (which is expected planned service cost for Toyota Camry up to 210,000) and used 2x coefficient, assuming that self-driven car may require more attention and for more expensive technicians
  • Parking. Here $5/day is taken
  • Cleaning. One clean per day assumed with cost of $15, which I think can be achieved with enough scale
  • Insurance, duties, registration. Number of $43 per week from RACV fact sheet is taken, for Toyta Camry again. On the one hand it is for private use, but on the other hand, self-driven cars are considered more safe and big car fleet owner will have more bargaining power than an individual
  • Tyres. The assumption is that the tyres should be changed every 60,000 km and the cost of replacement is $600

The cost above can be classified in two groups:

  • Time-dependant, like parking and cleaning. A car has to be cleaned and parked daily, no matter how many rides it takes (the case of zero rides is ignored)
  • Mileage dependant, like petrol, service and tyres. Because I assume that we write-off a vehicle after 200,000 km in fact that expenses are fixed for the life of the car.

In addition to assumptions there are several facts used in calculations

  • I’ve checked recent review Uber statistic, it tells that the average ride for Uber in US is around 10 km
  • I used Uber fare estimator to check the cost of such trip in Melbourne, Victoria, it starts from $19 (Uber X)

That’s it, it is a simple analysis of gross profitability per vehicle, so factors as taxes, admin, marketing and other expenses as well as additional earning opportunities (such as commercial delivery) are ignored.

Utilisation and gross profit with current prices

So how profitable will Uber and other companies who use self-driven taxis will be if they replace human drivers to robots? It is a significant shift of business model – from being a middle-man connecting driver and passenger to becoming an operator of own (in some way) fleet.

Well, as I found from my simple calculation, if the prices stay near the current level, it will be fantastically profitable business

The chart above demonstrates gross profit per vehicle (Y axis) depends on utilisation (X axis). As Utilisation I refer the share of time that the vehicle used to carry passengers. Utilisation of 50% means that 12 hours out of 24 hours in average the vehicle is working.

Even with very modest 15% utilisation, which means that there will be just 7.2 trips per day, each self-driven vehicle brings over $150K. It takes over 5 years in this scenario, but still looks like incredibly good business!

You may say “Hold on. The price will inevitable go down in such scenario!”. I think you are right, let’s check what can happen with profitability with decreasing prices.

Autonomous taxi ride price and profit

Let’s what will be with profitability of our vehicle if the prices go down. In this analysis we kept utilisation ration high, at 50%.

As you may see from the chart above, the business is still profitable if the price per ride is above $6. It is over 3 times less than the current price people pay to Uber in Australia!

With almost 2X decrease of the price to $10 each autonomous cab generates $60K of gross profit.

As a final test let’s look to gross profit with different combination of price and utilisation.

Horizontal axis shows the price, there are 3 scenario displayed: $7, $10 and $15. Blue bar indicates profit for 20% utilisation, red – for 30% and yellow – for 50% utilisation. Again, up even with $7 price with 50% utilisation unit economics of the business looks not bad at all.

What is the secret of such astonishing financials? The answer is simple – remove of a human driver from the equation. Driver salary is a major cost component in this business expenses at the moment. So the future is not very bright for Uber and taxi drivers, in the future a taxi with human driver will be a premium service.

Probably venture capitalists that invests billions in Uber made similar calculations, self-driven cabs look like the longer term game of the company. And while the unit economics looks very good, the operation require sufficient scale to be effective and spread costs like software development, IT infrastructure, administration etc over larger base.

At the moment Uber reportedly has 62 million rides per month (July 2016) in US only. With $10 sale price and $6 per ride cost (based on the calculations that include car depreciation) it will result almost $3B of gross profit in US alone. Given that cheaper ride should result more trips, self-driving cabs industry looks quite promising.

If you are interested to check the calculations in more details – you can find them in this Google Spreadsheet. You can copy it and play with numbers. Comments and corrections are welcomed!