Here we go. Tony Seba declares it’s “boom, over” for the internal combustion engine (ICE).
In this series of posts I’ve tried to provide some metrics to measure whether Tony could be right. Central to my thesis is that when electric vehicles (EVs) match or exceed ICE vehicles on every criteria then “yes” it will be “boom, over”. Let’s revisit the criteria from post number 13 in this series. The sources of utility derived from a car can be thought of as threefold:
- Top speed
- Internal volume
- Off-road and specialist capabilities
- Status signalling
With respect to mobility, EVs are already superior to ICE vehicles in terms of acceleration and top speed. With their low centre of gravity (governed by the battery), their handling characteristics are also very good. Moreover, the small size of EV motors means that ultimately we can have motors front and rear, or, indeed, on each wheel. With each such modification, EV handling will pull away from that of ICE vehicles.
The powertrain of an ICE vehicle is a lot bigger volume-wise than that of an EV. It’s only through taking into account the big battery that the entire drivetrain of an EV matches, or possible exceeds, an ICE in volume. Here again, however, technology favours the EV. Batteries are getting smaller. The internal combustion engine has been around for 100 years. It won’t get much smaller. So, in time, the entire drivetrain of the EV will be smaller than that of an ICE, freeing up space in the rest of the car.
The fact that EV motors are so small and don’t require gear boxes, exhaust systems and other ancillary equipment means that they already favour creative configurations. At the minute, the size of the battery is so big that it limits flexibility provided by the smaller electric motors, even though we could chop up the battery and put it in different places. Distributing the battery around the car though is currently not an efficient thing to do since it adds complexity to the electronic control panel and heating systems regulating the battery. But as the size of individual battery cells come down, the battery can be placed within the required form factor of the overall car just as it is in a lap-tap computer. Ultimately, we will move to a designer-led car rather than an engineer-led car, as with smart phones.
With the independent control of each wheel and a freehand regarding configuration, designers can also dream up the perfect MPV, SUV or pick-up truck. Dropping the tyranny of the big engine, gear box, crankshaft, and exhaust system means that future EVs will be able to cope with challenging terrain or complex pulling and carrying requirements far better than ICE vehicles.
Now we get to aesthetic and status signalling. The rush by high-end German auto makers (such as Mercedes, BMW, Porsche and Audi) to roll out EVs in response to Tesla shows that EVs are soon going to dominate high-end auto sales. Indeed, Jaguar‘s top management is now debating whether to transform the brand into an only-EV one near term (here). The strategy of other top-end makers like Porsche is to offer a full-range of EVs alongside the old ICE vehicles. But the key point here is that all the luxury brands now see no contradiction between a car with an electric drivetrain and a car that conveys high status.
True, for those ICE aficionados, the sound of an ICE engine and the steam punk-type glory of lifting a bonnet to see a V6 will never be eclipsed by an EV. But the new Tesla Roadster will place all of these vehicles in the rear-view mirror. So there will remain an aesthetic and status-signalling rump demand for ICE vehicles, but this will be the same rump demand as exists for Swiss mechanical watches. Demand for Swiss mechanical watches exists but its irrelevant in terms of aggregate revenue for the watch industry.
But now we come to the chink in the EV armour: price. As pulled out in previous posts, affordability is a temporal constraint: it stretches from the present to the future. So the purchase decision is not just bounded by the current available budget available to buy a car now, but also by the budget you will have to run the car through time and, ultimately, replace it. Restated, the purchase decision takes into account the sticker price and future costs captured by fuel, maintenance and depreciation.
EVs are already ahead of ICE vehicles with respect to fuel costs, maintenance and depreciation. And if something costs less in the future, theoretically you should have more money to buy it now (financing and credit facilities allow one to push payments for a car from the present to the future). In other words, if we can lessen future costs, we have more money to buy a more expensive car today.
So to repeat: given EVs are superior to ICE vehicles with respect to mobility functions, and can be at least as good with respect to aesthetic and status-signally, once their price approaches that of ICE vehicles, looked at in terms of the aggregate purchase price and future running costs, the market will tip. That is, sales of EVs will grow exponentially and ICE sales will collapse.
At this juncture, we should emphasise that the car market is not one amorphous mass. It is highly segmented with the vehicles in each space having a particular combination of mobility, aesthetic and status-signallying functions and price points. Thus, we are not really talking about a single tipping point, but a series of rolling tipping points as EVs meet the required sales take-off criteria in each segment one by one. To get a sense of the segment breakdown in the US, 2016 sales by type are given here:
Next question: “Are EVs at the tipping point in any of the above segments?”. Twelve months ago, the answer to that question would have to be ‘no’. But you are lucky to be living at a unique point in history. As of the third quarter of 2018, the answer must be “yes”, and that is down to Elon Musk and his Model 3. The rise of Model 3 sales over the course of 2018 has been nothing short of stupendous despite all the much-reported production issues.
The Model 3 is leaving its rival brands in the dust:
The next segments up for attack by EVs are luxury mid-size and large SUVs and crossovers. Here, the incumbent brands look like they will get to market first since Tesla has its hands full fulfilling its Model 3 order book before it can move on to its crossover Model Y.
At the top end, we have the Jaguar i-Pace, soon to followed by the Audi e-Tron and the Mercedes EQC. The adage “build it and they will come” appears to be holding true again, as the i-Pace has a full order book and Jaguar has hopelessly limited capacity to meet demand. In short, we have evidence that when an EV offering is at a similar price point to incumbent ICE vehicles, you will be able to sell all the EVs that you are able to make.
To repeat: what the sales data are starting to show is that if an EV is offered at a similar price point to an ICE vehicle, the public will buy the EV since an EV is a better transport alternative in terms of performance. Basically, the meme that the pubic don’t want EVs appears total rubbish. Given this state of affairs, what are we to make of these two charts. The first is from a Citibank report on EV penetration rates, and the second from an article by David Roberts of Vox.
In the top chart, the high-end 70% sales penetration rate by 2030 forecast is predicated on a breakthrough in technology (probably a commercial solid state battery). Personally, I don’t think that will happen until a decade later, so can put that forecast aside.
Based on lithium-ion technology, we are really looking at 2030 sales forecasts by every single forecasting entity of less than 50% of total vehicle sales. One of the most bullish forecasts is Bloomberg New Energy Finance (BNEF). This company’s latest Electric Vehicle Outlook 2018, which forecasts forward to 2040, was published in August and can be found here. The headline quote for the report:
And from inside the report a percentages sales penetration number:
So the bullish BNEF suggests we need another 20 years at least before EVs overtake ICE vehicles in total sales. Keeping that in mind, Tony Seba’s belief that the EV sales penetrations will be over 95% in 2030 is not just an outlier compared with other forecasts but on a completely different planet (my graph based on Tony’s statements):
The disconnect is massive. In effect, BNEF, one of the most bullish mainstream forecasters of EV sales, sees a penetration rate a lot less than half of that of Tony Seba in 12 years’ time.
Nonetheless, as one of the first vindications of Tony’s theory, in the small- to mid-sized luxury car segment Tesla alone was around one third of total sales in September 2018!
Moreover, with a slew of new entrants, the luxury crossover and SUV segments is going to experience an identical attack by EVs on ICE vehicles as the cost and performance metrics are basically the same as the luxury car segment. Accordingly, the only way that BNEF and all the other companies and organisations that forecast total penetration rates well below 50% in 2030 can be right if one or both of two conditions are fulfilled:
- EVs can’t get down to cost parity in the mass market car, SUV, crossover and pick-up segments, or
- The demand exists across all segments for a Seba style 90%-plus penetration rate, but not enough batteries can be produced and/or EV production lines put in place to fulfil demand.
Now if you listen through Tony’s presentations, he claims that battery-price falls have accelerated from 14% to 20% per annum. He then projects these trends forward and sees EVs being chapter than ICE vehicles across the board in a few short years.
The head of BNEF‘s team Colin McKerracher has some rude things to say about this. From his Twitter account,
And in response to Seba a short Twitter spat ensued:
Mckerracher’s argument is that as the price of the battery comes down, in percentage terms it will become less dominant in the overall price of the car. I agree with this argument, but I don’t think it is enough to knock Seba’s huge EV penetration number out of the ring.
As I have argued throughout these posts, EV’s are superior to ICE vehicles in a number of domains like performance, running costs and so on. Accordingly, they don’t need to be cheaper than ICE vehicles to replace them. All they need to do is reach price parity. And we are witnessing real time what happens when EVs match ICE vehicle in a particular segment with respect to price: in the small- to mid-size US luxury car market we are seeing the market tip toward EVs in a matter of months.
In the same Twitter exchange, another poster put up a helpful chart to show the price fall dynamics of a 16% battery price decline with the price of the rest of the vehicle’s main components kept static. The battery is very much the swing component.
In my last post, I did some rough, back-of-the-envelope calculations so as to determine what the battery price needed to get down to in order for a mass market EV to sell at parity to an equivalent segment ICE vehicle. Look at the post for details, but the bottom line is that the battery needed to be produced for $2,500 for a 2017 US average sticker price model of $37,500.
From Electric Vehicle Outlook 2018, BNEF sees battery pack prices at $70 per kWh in 2030. Further, given the almost daily advances in battery charging speeds taking place at present (we seem to be on a through train from 100kW to 350kW and beyond), a mass market vehicle should be able to get away with a 50 kWh battery (which with a 350kW charging station will recharge in a little over 8 minutes, assuming that battery chemistry and control will have advanced to a stage that can cope with that speed of charge in 10 years’ time).
Seventy times 50 is $3,500 dollars. So let’s say that this makes the EV $1,000 more expensive than ICE vehicle or 2.7% more. As I mentioned above, the aggregate cost of a car is the price of purchase plus running costs. Given EV running costs are lower, a slightly higher EV sticker price of purchase is completely compatible with my concept of total cost parity.
As a reality check, even today we are seeing the first stirrings of proper EV offerings in the smaller vehicle segments. The Hyundai Kona, a small SUV, boasts a 64kWh battery and a range of around 300 miles. Just launched in the UK, it sells for around £25,000. Add back in the government subsidy of £5,000 and the total comes to £30,000 or about $40,000. By comparison, in a similar segment in the US the Honda CR-V and the Toyota RAV 4 sell for around $30,000.
Hyundai uses the LG Chemical NCM 622, which has a relatively high cobalt content compared with Tesla‘s battery. Consequently, I would guess that the Hyundai Kona battery costs around $200 per kWh, or $12,800 for the 64kWh battery overall. If we could get the kWh cost down to $70 in line with the BNEF forecast, the battery would come in at $4,420, for a saving of $8,320. At that price, the Hyundai Kona would sell for about $1,500-$2.000 more than the RAV 4 or CR-V. Basically parity given that the Kona will be cheaper to run.
True, getting the battery cost well below $100 is a challenge. There are two opposing forces we should consider in the declining battery cost trajectory. BNEF rightly point out that Tony’s per annum price decline forecasts are less useful that an experience curve approach (in economics we talk about economies of scale and learning-by-doing effects). What this means is that price declines are not a function of time but rather how many batteries you produce. Thus, if the volume of batteries produced goes up exponentially, the price of the batteries comes down exponentially.
And if you want to get a sense of how fast battery capacity is being ramped up, then you should follow this link and read the article by Simon Moores of Benchmark Mineral Intelligence. Moores notes that existing and announced battery factories will have the capacity to produce 1.1 TWhs of batteries per year. That is over a 10-fold increase over existing production. The scale of this application of capital and innovation should reap huge cost-cutting rewards.
The major impediment preventing batteries costs coming down well-below BNEF‘s forecast of $70 kWh in 2030 is the price of the raw materials that go into a battery. Very roughly 0.8kg of lithium goes into a 1 kWh battery, and that about of lithium currently costs about $10. The most expensive metal used in current generation lithium-ion batteries is cobalt, which at one stage got as high as $90 per kg, but is now around $62. From the chart below you can see that different battery technologies use different amounts of cobalt. Tesla’s Model 3 used NMC 811 batteries (eight parts nickel to one part cobalt to one part manganese). A 50 kWh battery uses 4.5kg of cobalt or about 0.1kg cobalt per kWh. That is worth about $6. Nickel current trades at about $13 per kg. So for an NMC 811 battery we are looking at about $10 per kWh for the nickel. Thankfully, manganese is relatively cheap, so isn’t a big swing factor in battery pricing.
Nevertheless, if you add the three most costly metal components together, we have a floor cost of $25. And these are the raw materials for just one component in the battery: the cathode! You then need to complement the cathode with a graphite anode, an electrolyte and a separator. Then these components have to be combined into cells, which are linked into battery panels, and finally fabricated into battery packs. And the whole system requires a very sophisticated charge management and heat control system. In sum, getting battery prices below $100 per kWh with current lithium-ion technology will be tough.
With that caveat in mind, I will ask the following question to finish this series of 20 blog posts on the future of EVs:
“Is Tony Seba delusional in seeing EVs totally displace ICE vehicle sales by 2030?”.
My answer to that would have to be “no’. Ninety-five percent plus penetration is certainly a stretch goal since it would require current lithium-ion battery technology to be pushed right up against the boundary of what is possible in terms of price and performance. But I feel it is just about doable.
Does that mean that I think it will likely happen? That is a different question. I’m a probability guy and prefer thinking about the range of possible outcomes as opposed to giving one point estimate. For me, saying EV penetration will definitely be X percent in 2030 when looking at such a dynamic industry is a meaningless thing to do. My own guesstimate is that the EV penetration rate will likely be somewhere between 60-80%, which still puts me completely outside of the mainstream.
Nonetheless, in the spirit of Tony Seba hyperbole, I am happy to go out on a different type of limb. So in my words (rather than Tony Seba’s):
“All those who think EV sales penetration will remain down at around 30% in 2030 (the consensus view) are completely, utterly and certifiably crazy!”