Published on June 28th, 2016 | by Julian Cox150
Tesla Model 3 & The Economics Of Autonomy — Why An Autonomous ICE Vehicle Is Relatively Pointless
June 28th, 2016 by Julian Cox
This is a long-form reply to a group of questions in comments surrounding the sub-topic “autonomy.” For context, this relates to the Keynote Presentation I gave in Berlin for the 1st Cleantechnica Revolution Tour conference — the video and description are available at this link.
In my opinion, it is enough of an initial shock to conventional wisdom to get to grips with relative total cost and value of ownership of the Model 3 vs a BMW 328i and figure out that the Model 3 is just a cheaper and better car (see link above) without straining credulity and mind blowing people about the implications of practical networked ROBOTICS.
Robotics was capitalised deliberately on my whiteboard presentation (forgiveness for the whiteboard is asked, by the way).
The Model 3 is the first truly major “Thing” in the “Internet of Things.” 12 Cents per mile will cover the entire job — including operations centers, automated valeting and multiple interior swaps before end-of-life recycling (as well as, of course, the car, energy, finance, insurance and maintenance). This 12 cents is less than the cost of putting gasoline in a $100 beater, taxiing, insuring, and maintaining it. BUT … the market value of a mile in the Model 3 will be vastly higher than 12 cents so long as competition with conventional car ownership, leasing, ridesharing, and taxi services remain.
The market value of mileage in a car with a built-in driver will allow for autonomous ICE vehicles for a while, but the staggering additional profit margins between, let’s say, 12 cents of total cost per mile and $1–2 of total customer value per mile in an autonomous EV — hundreds of percent of net profits — while freely undercutting Uber (at $2.00 per mile) by a factor of two (at $1.00 per mile) — will predictably cascade capital into autonomous EVs like no gold rush that the world has ever seen.
More than this, if Tesla shares a percentage of this profit gold rush with its customers (while retaining sufficient to rapidly capitalise new build capacity) it will become profitable for customers to buy a car from Tesla and to let that car earn money full time or part time, and it will be profitable to buy a car from Tesla for exclusive use and to sell it second hand into fleet service. This profit potential takes the Model 3 total cost of ownership (TCO) from soundly beating the BMW 328i and a close call with a Toyota Corolla (with a standard ownership cost and depreciation model — per Edmunds.com data for the BMW and Toyota) to a complete contrast – not a cost of ownership but a profit of ownership. The Tesla, post full autonomy, is a moneymaker — a productive asset like a condo that you can rent out while you are away — and nothing at all like the economics of an ordinary car purchase. Literally the more cars you can buy, the richer you get. With car purchase as a customer asset rather than a customer cost item, Tesla will be able to grow exponentially on cash raised by customers to buy cars and on profit sharing with the customers via Tesla’s OTA-enabled fleet services. Not corporate bonds and bank debt, not government grants, not the sale of equities. Although, it can be said that lenders, governments, and shareholders will no doubt throw themselves at Tesla for a chance at a piece of the pie. Tesla autonomous vehicle factories will undoubtedly be granted land cash and labor allocation in a national bidding war far more heated than state bidding wars for the first Gigafactory, and those national factories can conceivably produce cars that simply drive off the line to their purchasers or cars that just route themselves cross country to wherever customers are hailing rides.
The basic premise is this: If Tesla (and others) comes through with a million-mile drivetrain and rolling chassis, as planned, then every $10,000 of lifetime cost represents 1¢ (1 US cent) per mile — by simple division. It is easy to see, therefore, that the cost of the car to build (let’s say $30,000) will cost 3¢ per mile. If it’s $40,000, then it’s 4¢ per mile. Wholesale, in-house-generated electricity is a trivial cost: 5¢ per kWh tops — probably Tesla with SolarCity on board can produce an excess at a net profit to the entire transport system. At 5¢ and 29 kWh per 100 miles, that’s about 1.5¢ per mile (one and a half cents, and is probably closer in reality to zero net cost after profits of selling electricity as well as mileage). In-house maintenance: let’s say we allow $20,000 per car for tires and washer fluid, etc. That’s 2¢ per mile. So, this is a total of 5.5¢ per mile and a balance of $5,000 for finance and insurance (0.5¢ per mile) for an all-in cost of 6¢ per mile.
The BMW 328i pulls 27 MPG (EPA) average gas mileage (whether it is autonomous or not). $1.80 gasoline is 6.66¢ per mile. Hence, if the car was a 27 MPG $100 beater and not a brand-new BMW 328i, it would still be cheaper to operate the $30,000 brand-new autonomous Tesla — and much nicer, most likely, for the passengers, who would unarguably be willing to pay more money to ride in the Tesla than in the $100 beater. The other significant implication is that there’s not even $100 (LESS THAN NOTHING per mile) after $1.80 gasoline to pay for producing a car of any description to compete with the autonomous EV. The only opportunity for an autonomous ICE vehicle is competition with legacy non-autonomous vehicles.
The all-in per-mile cost of an autonomous Corolla is not 6 cents per mile — more like 30–60 cents with rapid mileage depreciation, third-party for-profit fuel, and legally mandated third-party for-profit dealership maintenance and valeting. A 30–60 cent cost basis can compete temporarily with Uber and taxi services, but only until the autonomous EV gold rush wipes it out. Nobody in their right mind would throw capital at such an obviously diminishing return given the obviously more financially attractive alternative of the EV — most importantly, that includes consumers. It also probably includes car dealers, who would be far better off becoming Tesla fleet owners than sinking their capital into more dealership inventory on a sales lot. With a cost basis of 60¢ per mile, there is absolutely no problem running Tesla Model X vs an autonomous Corolla — with plenty of money to spare and a far more attractive ride to offer to passengers.
The other counterintuitive thing to consider with autonomous vehicles is this: 30 MPH (50 km/h) average (three times higher than many city centers), 70% duty ratio (plenty of time for autonomous charging), 24/7/365 = 5 years and 5 months to reach 1 million miles — and only 1 year and 7.5 months to reach 300,000 miles, after which the typical ICE drivetrain has reached the end of its service life. If that car needs servicing every 10,000 miles, it will be in the shop 29 times prior to recycling, or every three weeks.
At 27 MPG for the BMW 328i, it will have consumed 19.2 times its own weight in gasoline and produced multiples of that in non-recyclable emissions. 100% of the EV — not including tires and washer fluid (and possibly replacement seats and interiors) — consumed is available for recycling at 1 million miles. The recycling value of the EV net of lifetime consumables is not inconsiderable. Let’s say it is $10,000 worth of recycling. That deducts 1¢ per lifetime mile. There is nothing to be recouped from 6.66¢ per mile spent on gasoline — and at that price ($1.80 per gallon, coinciding in the oil market to about $40–50 per barrel) most of the world’s oil producers cannot make a profit on oil, requiring closer to $85 per barrel to break even and $100 to be profitable.
When contemplating Apple or other newcomers to the autonomous EV space, it is also worth contemplating that a car limited to city commuting is absolutely fine — that is, after the Model 3 has proven the transition to regulators because it is capable of everything. Successors to the Model 3 do not necessarily have to demonstrate an improved safety record for autonomy over a manually controlled version of the same car, and hence they could conceivably introduce vehicles with no manual version or even a manual override. An autonomous EV that is not a driver’s car does not have to be capable of all things that every driver could want. A journey can be electronically connected from a low-range, low-performance, but richly connected and entertaining city “bubble” car to a Tesla Model S or X for, say, an Interstate run — or maybe to a Hyperloop run instead. This will allow EV newcomers like Apple and probably LG, Samsung, etc., to forgo pushing the limits of vehicle dynamics (and even battery range) and still have a product that ICE city car ownership and leasing cannot possibly compete with. The only real argument against a proliferation of low-performance autonomous bubble EVs is that phenomenally great autonomous EVs will be under no margin pressure from legacy vehicles and won’t be before at least half the world’s road fleet is replaced. So, given the choice — money no object — why not get in an S-Class EV for an extra few cents per mile? Still cheaper on a cost basis than getting into a very second-hand ICE car.
For those contemplating city planning, a vital consideration is the quantity of traffic on the roads that is seeking a place to park or taking up road lanes by occupying onstreet parking. An autonomous transport system only needs to park in a charging garage, ultimately alleviating all of this from city streets. For those concerned about a need for home charging in cities — even those that keep an autonomous EV for exclusive use will have a car that can go away and come back charged, alleviating the need for a personal place to park and charge. Conceivably, an autonomous EV can drive to an out-of-town wind farm, or cheap land connected to one, and return in the morning or on demand, charged. This is one of the reasons I think Tesla Motors could arguably contemplate turning its drivetrain development business over to wind power generation — a simple adaptation of the already firmware and OTA-software-integrated regenerative braking systems developed for the vehicles. A Tesla regen-based wind turbine would be free of rare earth magnets due to AC induction, could use a (Nikola) Tesla turbine as a kinetic capture device — for the sake of poetry, that is. Or just ordinary composite blades and stir-welded towers would be fine if anyone knows a manufacturing company with those kinds of competencies along with some appreciation of aerodynamics. Not particularly useful for Mars is the only objection I could come up with.
The implications for the SpaceX 4000 satellite Global LEO OTA Internet network are another thing that is breathtaking to behold when contemplating ROBOTICS of transport and energy. That would probably be to take mind blowing to an unacceptable level. Just to hint at it: An OTA connected robotic system is not limited to routing an autonomous Model 3 from A to B without a driver. There is nothing to prevent the concept extending to performing useful functions at A and at B. It does not have to be limited to parking or moving on to the next journey. There is nothing written in stone that the shape of every autonomous vehicle on the OTA network should be for carrying passengers, or even car-shaped, to perform those tasks at A and B. The most basic of which would be an autonomous subunit capable of loading and unloading the autonomous EV. It would stand to reason to produce an autonomous subunit that fits in a seat designed for a human passenger. I think maybe you can see where this is going, which is why I thought it wise to limit my presentation in Berlin to the more palatable step one of the transition from the oil and gasoline era to the robotics or “Internet of Things” era, which is: Why EVs that you can order now in the form of the Tesla Model 3 are already better and cheaper than gasoline vehicles — no longer more expensive with compensating benefits felt only by environmentalists.
I am confident that all Model 3 owners will have a car that is a software update away from embarking on all of the above — from the get go — meaning a vehicle with better value for money in any case and highly likely to feature strongly positive appreciation over 5 years as a result of a full autonomy update. Not depreciation. I would also predict that this, along with a fully retracting steering column, is the content of Tesla Model 3 Unveiling Part 2.