The EV revolution can be one of your Good News sanctuaries, along with the Pootie diaries and GNR. There’s simply so much good news happening in the EV space that it’s hard to keep up, and sometimes it’s even hard to understand. This diary will look at the two models under which EV market share is expected to grow over the next ten years, and discuss the case for each.
The range of forecasts for EV sales as a percentage of new car sales in the year 2030 begins on the low end with 10-20% by Wards Auto and BNEF, and reaches a high of 60% by Tony Seba, author of Clean Disruption. Notably, however, Mr. Seba expects those 60% of vehicles to drive 95% of the passenger vehicle miles (PVM) covered. The other 40% of vehicles sold will spend most of their time parked, same as most cars do today.
To summarize the reasoning of the different lines of thinking, here’s a simple breakdown:
EV Growth Models
Growth Model |
Slow
(Replacement model)
|
Fast
(Robotaxi fleets)
|
Buyers |
Individuals replacing their old car |
Fleets providing Robotaxi services |
Data Basis |
Historic sales and market conditions |
Converging technology cost curves |
Projection |
Auto experts: Linear growth |
Technology experts: Exponential growth |
At a 1996 presentation on “Valuing technology for licensing negotiations” I attended in Chicago, the speaker asserted that to correctly value intellectual property required a thorough understanding of the market below the Technology S-curve, as traced by the yellow line on the graph below. (Not to be confused with the technology adoption life cycle, the blue line below, which is the market penetration Bell curve from Rogers’ Diffusion of Innovation theory.) And no matter whether the market is small or large, the progression of a technology through that market traces an arc similar to, but often steeper than, the yellow line depicted below.
Technologies progress along a series of interlocking S-curves, with each generation of technology completely supplanting the last. Look at the three curves below and, starting from the bottom, say Records, Tapes, CDs as you climb. Or say horses, cars, robotaxis.
So there’s a fundamental disagreement over whether the growth curve will be linear or exponential, which is startling given that none of the analysts predicting linear growth have explained why the S-curve doesn’t apply. It does apply, and they should know it. Do they? Let’s look at some forecasts.
EV sales forecasts 2030
Bloomberg New Energy Finance (BNEF): about 20% EV by 2030, hybrid’s included. Linear growth.
BNEF continued, here quoted saying 54% EVs by 2040, including hybrids.
Ward’s Auto - Looks like 10-15% BEV in the year 2030, with all other manner of “electrified vehicles” capturing another 35% of new car sales. Linear growth.
Let’s review some facts about EV’s before we continue, because there are some massive assumptions in the Wards Auto projection above. Specifically they believe that 48 volt mild-hybrid vehicles will capture about 25% of the market by 2030. These systems keep the gas engine but add a battery to help out at the low end of the rev-range, so the vehicle performs better off the line while using less gas. So if the driver of EV adoption was gas mileage, this technology would help. But the EV adoption curve will be driven by the total cost of ownership per vehicle mile, no matter what the power source is. Unfortunately for ICE vehicles, (internal combustion engine), their fuel and maintenance costs guarantee that they cannot compete with EV’s on a cost per mile basis. EV’s are taking over because they’re far, far, far more reliable and easy to maintain than gas engines, and those lower costs means they operate at a lower cost per mile. Full stop.
Ironically, Seba doesn’t use an exponential growth curve to show EV’s taking over sales, instead choosing to show the market effect of converging cost curves. In the next slide below he depicts the shift in PVM from individually owned vehicles to app-based mobility services, which he calls Transportation as a Service (TaaS).
By forecasting the convergence of price curves, Seba projected that robotaxis would become technically feasible in 2021, and that in ‘23-’24 they would begin rapidly replacing self driving until accounting for 95% of PVM driven in 2030. He projects that 40% of current car owners will still own their cars, but they will be used less and less, and have no resale value. The collapse of used car values is an important factor in which EV growth model prevails. Gas vehicles will be sunken costs by 2024, having residual value of scrap. This will disincentivize purchasing new ICE vehicles. Death spiral, activated!
He backs this up by saying that robotaxis are a 10x cost improvement (per mile) over individual ownership, because cars spend most of their time parked. When cars drive themselves, your car can earn money while you’re sitting in your office, or sleeping, or while you’re at the game, (where you didn’t have to pay for parking. Win!!!), so over the interval of your car loan those dollars are spread among 10x as many miles. And with EV’s, that doesn’t require a significant increase in maintenance expense, because there are only 1% of the moving parts as in gas engines, so there’s very little powertrain maintenance, and the batteries are proving to last over 500,000 miles. Consequently the incremental cost for car owners to allow their car to deploy as a robotaxi, is basically electricity, and should operate profitably even deployed inefficiently (i.e. 50% utilization).
Tesla predicts their cars can earn $30,000/year profit in robotaxi mode driving 16 hours per day, averaging 16 mph and with only 50% utilization rate. If they’re right, once robotaxi services are legalized, a Tesla would be able to pay for itself in under two years, or simply trickle a few hundred dollars into the owners account, above and beyond the cost of the loan, insurance and charging. If it can operate for 500,000 miles on it’s battery pack, that represents lifetime income of over $150,000, or over $300,000 if Tesla’s million-mile powertrain becomes a reality, even at just 50% utilization rate. So if Tesla wins the race to autonomy, there’s a clear business case for owners to loan them to the Tesla Network robotaxi fleet. Most other robotaxi scenarios are purpose-built AEV shuttles owned by the app provider and never available for sale to individual owners.
But what about the demand side?
In some areas TaaS already accounts for over 20% of PVM through mobility services such as Uber and Lyft. The app-car already serves far more PVM than the entire US taxi industry combined, and it isn’t even efficient yet! It cannot achieve sufficient network effect to achieve greater than 100% utilization rates with their current penetration rates. But autonomy is already happening in limited scale scenarios, so where TaaS ends up is clear: robotaxis.
The reason why TaaS and the replacement model are completely incompatible is because once TaaS is approved in an area, that area will become saturated with robotaxis as quickly as they can be manufactured. And robotaxis will be manufactured in volumes that will make Ford F-150 sales look quaint. Robotaxis will make driving yourself a more expensive and less convenient option than just pushing a button on your phone and being dropped off at the front door. Because owning a car today means dealing with insurance companies and parking tickets and speed traps and shamefully, all the risks of Driving While Black, there is a lot to like about not owning a car. An app on your phone has none of the downside of car ownership, and will save people money, too. For some people more money means more places to go and more time to get there since they’ll sleep on the way. These can be life changing benefits, which will quickly become clear to a rapidly growing audience.
Finally, let’s keep in mind that over 30% of U.S. households are overdue on debt, so if they can save money by getting rid of their POS car, they’ll jump at it. This suggests that there’s nearly a 30% market share locked-in for whoever provides comprehensive robotaxi services first.
But what if robotaxis are still ten years off? Then the question becomes, will EV’s disrupt the transportation industry even before robotaxis become viable? To answer that question let’s explore the below image of the cumulative sales volume, by model, of “electrified vehicles” over the past decade.
The Volt, Bolt, Leaf and Prius, depicted above with the grey lines achieving the greatest height on the graph are exemplars of linear growth. But the red line is the Tesla Model 3, and illustrates what happens when the technology cost curve reaches relative parity with the technology being replaced. Explosive growth follows. Not linear, exponential.
The Tesla Model 3 achieved near price parity and almost immediately became not just the best selling EV, but the best selling midsize luxury vehicle in the world. This year in the US, Tesla will sell more midsize luxury cars than Mercedes and BMW combined. Same as they did in 2019. Which underscores the fact that EV ownership is already a superior experience to ICE ownership, so when EVs achieve price parity in a segment, they are taking over that segment. For Tesla the Model Y and the Cybertruck will be the next two vehicles to prove this point, both being simpler to manufacture and holding more reservations for sale than the Model 3 enjoyed. So both models are likely to see a stepper curve than the red line above. The reality is that EV’s are taking over even without autonomy.
To wrap up:
- There are two models for EV growth, fast (robotaxis) and slow (replacement model)
- Replacement model relies on retail car buyers replacing old cars with new cars
- TaaS model assumes robotaxis will become the low-cost provider of all transportation services
The replacement model thinks the Model 3, Model Y and Cybertruck will be a nice lineup of cars which might/could/should lead their respective segments. The TaaS model assumes that’s right, but off by a couple orders of magnitude, because the 3, Y, Cybertruck (and X and S ) are all going to be OTA software updated to be robotaxis, at which time production will scale to MAX3.
Where robotaxi supply meets demand Tesla will drop their prices and seek to exceed 100% utilization. The existing base of TaaS users will jump to something substantially larger. Logically, the 30% of American households under financial strain today will adopt TaaS as soon as it has the price and reliability that meets their needs. Robotaxis will offer exactly that, so there’s likely going to be a huge correlations between early robotaxi adoption and meaningful financial relief. That’s healthy!
My next diary will go deeper into the issue of the robotaxi timeline, but suffice it to say that there’s a reason I keep drawing attention to Tony Seba’s predictions. The EV revolution isn’t just coming, it’s arriving.
- TaaS is already here as seen among a wide selection of phone apps.
- Robotaxis are just data critical-mass away from becoming real.
- It’s a safe bet that Tesla, having just built their millionth car, is rapidly approaching data critical-mass. If that’s correct, their fleet of vehicles is a software update away from being AEVs, and your first robotaxi.
So looking ahead, do you want to ride in robotaxis, or own one? If you think Tesla knows what it’s doing you can place your reservation here, if you dare!