Uber had no upside, Part DeuxDec 11th, 2021
My previous post, Uber had no upside, quickly became the top-ranked post by page views in one day. I am much grateful for your appreciation1. I am equally inspired by the hard questions some have raised2. In this follow-up piece, I would like to address the major points surfaced in the discussions. Please bear in mind this is not investment advice though.
A review of the discourse
Broadly speaking, there were three categories of discussions:
- Uber has a deep moat in economy of scale through its network effects. It would be uneconomic for a new upstart to compete.
- Uber could use its low-margin businesses to fund growth projects which could be high-margin, just like Amazon did with retail and AWS.
- I was contradicting myself by saying Uber had no upside while still holding a meaningful position in $UBER stock.
Let me try to address each of them in the following sections.
How to compete with Uber economically
Reader Martin Kess wrote that Uber’s high utilization and reliability are a moat. This brings down cost-per-ride, and makes it difficult for a new upstart to compete. Let’s say, for a average $10 ride, the cost per ride is $8 for Uber but $12 for the upstart. The upstart could choose to subsidize $8 and bring price down to $4, but Uber would only need to burn $4 to compete. The upstart would obviously be in a very inefficient position. It would be unwise for any upstart to enter the market and start a price war even in a single metro area.
Or is it?
Contrary to intuition, it is Uber that would be in a very bad economic position if there were a price war. The problem is exactly Uber’s high market share. Let’s say a new upstart kicks off a price war in one city where Uber commands 100% of ride-hailing market, which is 1 million rides per day. The upstart sets a price of $4/ride (hence losing $8/ride) and quickly grows to 50k rides per day, or 5% of the market measured by the number of rides. Uber lowers its own price to $4/ride (hence losing $4/ride) to avoid further losing share (hence doing 950k rides/day). At this point, the upstart is burning $400k per day, but Uber is lighting $3.8M on fire per day, 9.5x the amount the upstart is spending!
The fundamental problem here is that Uber cannot selectively subsidizes rides that would be taken away by the upstart. Instead, Uber has to subsidize all rides in the city. Uber’s dominant position becomes its Achilles’ heel in a price war.
To make matters worse, the economics tip more into the upstart’s favor in a number of ways:
- The more the upstart subsidizes, the higher the pressure Uber will feel. If the upstart sets the price at $2/ride, Uber will need to burn 11.4x as much in the above example3.
- Because of #1, the upstart has pricing power despite being the smaller player. Uber has to be reactive and match the price even if it will feel more pain at lower price points.
- There is no floor in the price war. $0 rides are plausible for the upstart, and the price can even go negative. That negative number could come in the form of rider promotions or driver incentives. This is even worse news for Uber because the math is not in its favor when the price drops even more into the negative territory due to #1.
- The upstart’s efficiency will improve as it gets more market share, meaning that it could burn less money per ride. On the other hand, Uber’s efficiency will worsen as it loses market share, causing it to burn more money per ride. Ironically, the relative spending difference between the two will converge despite the efficiency change in opposite directions. Again, the important variable here is the market share for each player. Both efficiency and cash burn are functions of the volume.
- Up until this point we have assumed the market to be static at 1 million rides per day. In practice, the market will expand due to lower prices. This is arguably bad news for Uber again, because it has to burn even more money comparatively as long as it has vastly higher market share.
- Another huge assumption we made was that Uber had economy of scale, i.e. it could burn less per ride. That may be true, but the degree is probably low. Uber obviously incurs a large variable cost per ride - about 80% of its gross bookings goes to the driver4. Even within the 20%5 that Uber takes home with, a large portion is used to pay variable costs like insurance. The fixed costs of paying the corporate employees and developing the software are where the economy of scale comes from, but that portion is small, only accounting for 4% of gross bookings according to Philo’s most generous estimation6. The new upstart does not need to spend billions or years on engineering to get an okay mobile app with passable user experience, and not a single venture capitalist will blink their eye to fund the fixed cost part of software development.
This was largely the playbook that Uber used to compete with Didi in China, coming in as the smaller player. Upon realizing its own disadvantageous position by being the dominant player in China, Didi pulled a masterful trick. It funneled investment into Lyft which staged a price war with Uber in the US from the advantageous market position of being the smaller player. In fact, Didi formed the anti-Uber alliance globally to finance local Uber competitors around the world. Uber was forced to leave China because it could not sustain the cash burn outside of China7. Didi could thank Sun Tzu for writing Besiege Wèi to rescue Zhào in the 2,500-year-old The Art of War.
Uber knew the precarious position it was in. That was why Uber kept raising billions after billions of dollars. Contrary to popular belief, the colossal funding rounds indicated not strength, but weakness. It was a last resort because Uber had no way to differentiate its product.
Uber’s high-margin dream
Several readers, in public8 or in private, has expressed that Uber could use its low-margin businesses to fund growth projects, some of which will hopefully be high margin. After all, this was how Amazon became a trillion dollar company, discovering AWS after spending many years in red ink for the retail business.
I actually gave one concrete example in my previous post. Meituan successfully built a high-margin ads business on the back of a low-margin food delivery business. However, I also explained why this is unlikely to happen for Uber, at least in the US market.
In more general terms though, the comparison between Uber and Amazon is a deceptively dangerous trap. Yes, both companies started with low-margin businesses involving moving atoms in the physical world. And yes, both companies spent their respective first decades losing money. However, there is at least one fundamental difference between the two from the financial perspective. Amazon was operating cash-flow positive9 but Uber was not10. To put things into perspective:
- Amazon was founded in July 1994. It was operating cash-flow positive in its first full year in 1995, and it has been that way ever since for every single year, with the exceptions of the years 1999, 2000 and 2001, when Amazon was aggressively investing for growth at the peak of and after the burst of the dotcom bubble. Amazon did amazingly well in the 2008 recession, growing its operating cash-flow by 20% from 2007 to 2008, and then growing another 95% the following year!
- Uber has never been operating cash-flow positive in any year since 2016, the earliest year for which there is publicly available financial data. From 2016 to 2020, Uber burnt through almost $13B cash in operating activities in the 5-year span. For context, Uber raised a total of $15B in cash from its founding in 2009 to mid-2016 when it reached a $68B valuation.
In other words, Amazon losing money was an accounting gimmick. It was reinvesting every penny of profit into future growth. One can read the 2004 shareholder letter from Jeff Bezos11 to understand why Amazon singularly focused on cash flows instead of profit & loss12. On the other hand, Uber did not have much extra money to invest into future growth, because it was burning though piles of cash in its core business for an extended period of time.
Amazon raised a total of $108M13 in outside capital in its entire history, and built a trillion dollar company by aggressively reinvesting operating cash flows. Uber has raised a staggering $25.2B14 to date, and is still operating cash-flow negative15. The margin for error was just too small for Uber.
My opportunity cost
Now on to the am-I-contradicting-myself question. The short answer is yes, but the full answer requires more nuance.
My primary consideration back in late 2017 was the opportunity cost of my career. It was important to work for an employer with (hopefully massive) growth prospects. I got more than half of my yearly compensation in the form of illiquid equity grants, so I had to think hard into the future and project the risk-adjusted payoff. I was given a meaningful 4-year RSU grant when I joined Uber in October 2015. I was awarded another sizable 4-year RSU grant in April 2017 due to good performance. I was walking away from more than half of my total equity grant in December of that year, because I saw no upside. I could get a much better deal elsewhere even if the new offer was lower on paper at that moment, because I would place much higher premium in the upside of the new company. From that angle, I would argue that my actions were consistent with my analysis.
There is also a real opportunity cost for my current $UBER holdings. As Murat U. and Hersh Mehta pointed out, I could have earned a higher risk-free return by selling and reinvesting into broad-market index funds immediately16. To be fair, my current holding is less than half of the total I acquired via my two-year employment. Nevertheless, I have to admit that I have nothing but emotional attachment by holding onto the remaining shares. I guess the takeaway is that even rational people with great analysis can make irrational decisions, which could be considered as my tiny contribution to disprove the Efficient Market Hypothesis :)
Please use this as an exercise to validate my math. ↩
Technically driver earnings is not a GAAP expense in Uber’s accounting, but that is irrelevant to our discussion here. ↩
The speculation for the reason behind Uber’s China exit was a logical inference. I had no material non-public information at the time of the decision, nor do I have any today. The decision was way above my pay grade as a fungible software engineer. ↩
Bezos noted that, in his 2004 shareholder letter, Amazon does “not focus first and foremost, as many do, on earnings, earnings per share or earnings growth”. He also mentioned “that a focus on EBITDA … would lead to the same faulty conclusion about the health of the business”. ↩
According to its Q3 2021 report, Uber’s 2021 year-to-date operating cash flow at the end of September was negative $338M. It was still negative, but an order of magnitude lower than the past few years. ↩
This post is not investment advice. I may transact in the securities mentioned in this post without providing any updates here. ↩