Sheelah Kolhatkar ends her brilliant profile of Dara Khosrowshahi - Uber’s newly appointed CEO - with a quote from the head of Uber’s autonomous driving unit, Eric Meyhofer. “The problem,” Meyhofer says “is, if someone builds [autonomous vehicles] and puts [them] on a ride-sharing network, their cost competitiveness will be stronger than ours. And if someone else does that, and we don’t have it, how long can we survive?” This quote implies a stark truth about Uber - as well as a lot of technology companies out there.
With Uber’s current cost base (their drivers), it is almost impossible for them to make a profit. The company lost $4.5 billion last year and is set to lose more this year. Subsidising rides grows the business but destroys any hope of profitability. Ride subsidies are such an established practice in the marketplace that if Uber was to reduce subsidies then one of their competitors would drop their prices and pick up users. Growth is god, and so the focus must be on acquisition. The self driving bet is really, then, an existential one for Uber. Cut out the cost base by replacing drivers with machines and they will create one of the most profitable companies on the planet. Fail, and they’ll have one of the most cited corporate disaster case studies ever written.
One of the key reasons Uber isn’t profitable is that human drivers cost so much. Classic tech management consultancy would tell us that this only leads to one strategic imperative: replace the drivers with machines as quickly as possible. A company like Facebook, for instance, is immensely profitable, raking in $4.3 billion in 2017. Why did Facebook profit while Uber lost billions? Facebook has minute staff costs: each extra Facebook user requires no new people, whilst each new Uber trip requires one more driver.
So far there’s nothing surprising here: the dream technology business has always been one that earns passive income - a technologist sets something up and it runs itself (perhaps with a core team of a few dozen), and quickly generates money beyond their wildest dreams. Yet in this pursuit of absolute automation something critical is lost. Rather than providing the best service, the company is motivated by providing the best technology. In an ideal world, Facebook’s ads would all be manually reviewed by a panel of well trained moderators. This would have prevented all the fake news and election manipulation scandals we’re now seeing play out. However, it would also mean that Facebook would be eternally loss making. Each extra ad would create an additional human cost, driving Facebook’s unit economics into the ground.
The obsession with technology only services is powering the world we see around us. Why bother to provide a better but more costly service with a dual human/machine combination where you could get 60% of the way (with some unintended consequences) and profit hugely? This battle is being fought fiercely in the education space already. All the evidence shows that smaller class sizes and more time with a human teacher increases attainment, but smaller classes are costly and resource intensive. Supply children with iPads and deliver them personalised, artificial intelligence powered, individual learning paths and you have suddenly slashed your costs.
This cannot be right. Humans enabled and empowered by machines will be the most effective option for almost any use case imaginable, yet tech entrepreneurs will always want to drive us closer to a digital only solution. Sometimes this might be the right option, but more often than not it’s a decision driven by investors, capital and, ultimately, profit.