Electric Meets Autonomous: Synergies for a Transportation-as-a-Service (TaaS) Future
[Disclaimer: The views expressed here are personal and meant for information/education purposes only and not representative of his employer directly or indirectly. Any mention of company names or use of images are for illustrative examples only. ]
This is part of a series of articles on e-mobility: Electric Vehicles & Ride Sharing Economics, Commercial Fleet EVs - Stealth Revolution, Towards Ubiquitous, Affordable, Ultrafast Charging, Electric & Autonomous: Synergies for a TaaS future, Towards Scalable EV Charging Infrastructure (Think Outside the Gas Station Box), Towards Scalable EV Charging: Hidden Costs of Level 2 Charging at Scale.
[This article was originally written before Tesla Autonomy Investor Day (slides, video). I have made some additional edits in the article below. ]
Transportation is going through a lot of transformations. First the drivetrain transformation from a less efficient internal combustion engine (ICE) to electric drivetrains (electric motors, on board power electronics, simpler transmission and a battery pack). This gives the vehicle immediate torque upon acceleration, battery packs for 200+ mile ranges, efficient energy conversion, lower center of gravity and rigidity for side impacts and larger crumple zones. Second, the vehicle platform can be updated via software remotely to add new features including autonomous features such as Navigate on Autopilot, Enhanced Summon etc. Observers and the company feel that with the rapid training of Nav on Autopilot with data from fleet of vehicles on the road, substantial improvements in safety and reduction in crash probabilities are happening, with a transition to feature complete autonomy just around the corner this year. Third, transportation is moving from a product model (you buy / own / drive a car) to a service model (you hail an ride share (Uber/Lyft/Ola/Grab) mobility service which subsequently becomes electric, and autonomous), enabled by connectivity & mobile apps Fourth: the quality of autonomous driving is improving rapidly with the availability of super-specialized hardware to run real-time deep neural networks, and multiple approaches to collect relevant training data to make the autonomous features better (but especially with the availability of more cars on the road collecting fine-grained driving behavior data).
This article unpacks a few synergies between the economics of electric vehicles, ride sharing mobility services and autonomous technology. This is not meant to be an exhaustive list.
Synergy #1: Low cost-per-mile in fuel costs & Higher Efficiency vs ICE
In a prior article I have computed fuel costs for ICE vs EVs for different sizes of vehicles, and it is attractive across all categories, though the impact is likely to be seen faster in smaller vehicles due to the availability of such vehicles on the road. For a car, assuming a gas price of $3/gallon vs electricity prices of 13 cents/kWh, and efficiency of 25 miles / gallon vs 2.5 - 3 mi/kWh, it yields an "energy" or "fuel" cost per mile of 4.3-5.2 cents/mile (EV) vs 12-15 cents/mile (ICE car). With a higher efficiency electric car such as Tesla Model 3 which has an efficiency of 4-4.5 mi/kWh (i.e. 222 - 250 Wh/mile), the cost drops further to ~3.2 cents/mile. This means that there is a differential of 9 -12 cents/mile vs a Tesla Model 3 and an ICE car doing 25 miles/gallon.
The low cost per mile will allow the transition to robo-taxis in two steps:
First, EV drivers in ride sharing can have a higher profitability per ride. Since the ride pricing will still be based upon gas pricing, and say it is $2.5 / mile and the owner-driver gets $1.5-1.75 / mile. They can get an additional savings of 10 cents / mile by riding in an electric car. In cases where their monetization is lower or if gas prices are higher, the 10 cents / mile is a larger part of the value. This is why in European countries where the gas prices are double (i.e. $6-8 / gallon), the additional contribution margin is 25-30 cents / mile which can be significant for a driver's unit economics. Note that the contribution of lower maintenance cost of EVs vs ICE is of a similar order of magnitude, i.e. 15-20 cents/mile in both US and Europe.The total savings combining fuel and maintenance is ~30-35 cents/mile. Since the driver share is $1.5 – 1.75 / mile, the 30-35 cents/mile represents an additional 20% impact to their bottom-line: not insignificant. The impact is higher in Europe (see map of gas prices below, and article: closer to 30-35% bottomline impact for drivers). Ride sharing companies will indirectly benefit as more drivers go electric, customers appreciate it (eg: via GreenMode publicity or otherwise), and as enough drivers adopt it, they may be able to pass on some of the savings to the customers.
Note: The relative contribution of maintenance cost savings is higher in US than in Europe due to lower gasoline/diesel costs. This drives a step change in unit economics. As mentioned in the prior article, for a 100K miles driven (~roughly 300 miles/day x 330 days/year), just this incremental unit economics will make a huge difference and lead to the choice of going electric. Eg: for 100K miles/year in ride sharing applications will net $17K-25K (fuel savings alone) A YEAR in Europe and $7-10K (fuel savings alone) in US. With the maintenance cost savings added, this savings figure will be of the order of $30-40K/year in Europe and $18-30K / year in USA for 100K miles/year driven (~roughly 300 miles/day x 330 days/year).
With superior unit economics, there will be a greater uptake of ride sharing with electric vehicles. Lyft is currently offering a "Green Mode" for electric and hybrid vehicles - and it is possible that customers may be willing to pay a slight premium for these vehicles as well. Anecdotal evidence from early use of Tesla Model 3 and model S is positive in terms of both revenue-per-mile realization and cost-per-mile impacts.
Once autonomy or limited autonomy is available, then the driver's load and stress can be reduced considerably allowing them to do more hours safely, and earn more out of the hours driven (due to the increased contribution margin). Tesloop has been pioneering inter-city services using EVs and this is an important factor of the overall experience as the inter-city services transition from ride share to point-to-point car-sharing. Of course, once the driver is not tied up in the vehicle, and the service is pure EV, then the pricing per mile can drop further, since the cost of the driver in the car is removed. Lower prices (eg: $1/mile) will allow users to substitute daily commutes (eg: to work, shopping etc) with autonomous taxis, and this has the potential to substitute a large fraction of vehicle miles-travelled (VMT) in urban areas. A single "robo-taxi" can have 5-10X the impact on e-VMT (electric VMT) and decarbonization compared to a single EV.
Synergy #2: Fewer Rugged Components: Longer Life of Asset (1 MM miles), Less Maintenance Opex
The electric car has fewer moving parts in its drive train. The moving parts such as the transmission is a simple single speed transmission, and the differential in gearing is also simplified (eg: see video).
With fewer parts and fewer & simpler moving parts, and the moving parts can be optimized for a longer life. Empirical studies on battery degradation in Tesla cars show the battery holding up well. For example, Maarten Steinbuch reports that Tesla Model S battery degradation data, averages: 1%/50,000km degradation rate, or 91% left at 270.000 km.
The drive unit and body being designed to be more robust for longer life and modular replacement of battery pack modules will also help.
As mentioned earlier, the maintenance savings could be of the order of 15-20 cents / mile for electric vehicles. The key components of these savings are from avoided oil and filter changes (and the opportunity cost of time spent getting them done), lower brake pad replacements (since you tend to use one-pedal driving with regenerative braking in EVs), and lower mechanical failure related replacements (eg: timing belt, transmission / gear box which happens every 100K miles in ICE vehicles etc). Rough estimates are $1500 in oil change for every 6000 miles which itself translates to 25 cents / mile! Given that electric cars have some occasional check up and tire rotations, I have calculated a more conservative 15-20 cents/ mile as the net maintenance savings. Source: AAA Study on True Cost of (ICE) Car Ownership. Ben Sullins has some more interesting data of a similar order of magnitude on a per-mile basis on his blog/ video. Recently Tesla has notified owners that service can be on demand and not scheduled (see video by James Cooke).
Now, for a 100K miles/year ride share driver this maintenance savings is worth $15-20K per YEAR contribution margin. It is interesting that the maintenance savings is also applicable for hybrid vehicles, even though the relative fuel savings may be lower. In other words, net net, there is a significant annual savings especially for applications like ride sharing / taxis which drive a lot of miles per year. Note with $40K Model 3 SR+ available in the market now, just the maintenance savings combined with fuel savings, and the existing government incentives ($3750 federal and some state incentives) can easily overcome any capex differential within a year for ride sharing applications. From then on, it is pure profit contribution. This is why it is a no brainer for a heavy-usage ride sharing or commercial drivers to choose their next vehicle to be an EV, provided they can get a good finance or lease scheme.
Synergy #3: Increased safety => lower insurance costs & decoupled from human in car
EVs with autopilot or similar autonomous features will have lower probability of accident; and lower probability of injury or death. For example, Tesla's Vehicle Safety Report indicates 2.87 million miles driven for every one accident with AutoPilot engaged vs NHTSA's data of 436000 miles between crashes in the general population. Currently this is not fully translating into reduced insurance, since the cost of repair etc need to be optimized. We can expect that as the population of electric cars with safety assist features on the road are increased and statistically significant actuarial data is available, the insurance pricing will start reflecting this in more markets.
Further, once there is full autonomy, and the driver is no longer needed, the insurance costs of the car per mile will be related only to the number of miles and where it is driven, and not associated with the driver. With a depreciated vehicle (eg: off lease vehicle see below), the insurance costs could also be lower since insurance costs tend to also depend upon the value of the vehicle and the value of the replacement. If more of the vehicle value is in software (easily replaceable with an over-the-air upgrade), the insurance costs will depend only upon the replacement costs of the hardware / metal / components etc.
The fact that the insurance is also decoupled from the occupant is crucial for transportation-as-a-service (TaaS), since today the insurance costs are significant for ride hailing companies. For instance Lyft in its IPO roadshows intends to reduce insurance costs by prioritizing safer routes over fastest routes for improved safety and lower insurance costs. Without the influence of a driver, the insurance costs will be a function of how good the autonomous driving is and its record on the road based upon aggregate fleet statistics. Significantly lower insurance costs will raise operational profitability of the service, and improve overall safety statistics.
In Tesla's Q1 2019 earnings call, they announced their intent to insure Tesla's cars (working with an insurance partner). Since Tesla collect granular data this will allow them to more accurately price risk; and subsequently bundle insurance cost with individual robo-taxi trips decoupled from the driver without being constrained by current auto insurance norms. [Update: Tesla Insurance was recently launched in California in August 2019. See here for an analysis.]
Synergy #4: Remote diagnostics, Software upgrades for maintenance.
Electric cars today have fewer components but all components are also fully observable and configurable by the firmware. This firmware can be updated by over the air (OTA) software/firmware upgrades. The ability to have granular controls, and update the controls via OTA upgrades allows the car to have newer safety or comfort features over time. The product becomes a living product.
From an operational point of view, the fine grained visibility by component and by time allows remote diagnostics and condition-based service of sub-components. Design for serviceability allows a small fleet of mobile service technicians to support a large fleet of vehicles (both individually owned and fleet in Tesla Network). Also it is possible that software upgrades can fix certain behavioral attributes that increase wear / tear and maximize service life of the equipment. This is a radical departure from the break-fix approach to maintenance, and high costs of scheduled maintenance for vehicles today; and also directly will translate to an autonomous ride share/taxi fleet.
Synergy #5: Finance Innovations & Leasing as a pathway to lower cost Autonomous EV fleet vehicles.
Financing matters for automotive sales and leasing. EV companies like Tesla offer a loan product with partner banks for 100% financing at 3.75% APR over a seven (7) year period. Nissan also offer an attractive lease option. This is an attractive option, especially for business users who can acquire the car for business purposes (see YouTube Video). Note that the depreciation claimed is higher based upon gasoline vehicles, but the reality is that the car depreciates lower, and has optionality for new cash flows with autonomy. The business user can arbitrage this, in addition to having a low cost per mile. In some states, there is also a per-mile writeoff of 50+ cents per mile, whereas the car actually has a lower cost per mile (fuel, maintenance costs) since it is electric! If in addition to primary business, one associates the vehicle with Turo or an autonomous taxi network, it will make the car far more affordable than the sticker price. This will drive up demand, and which in turn lower battery costs.
As an example of a lease, there is 3 year lease for their Model 3 vehicles where for $535/month ($6420/year or $19260 for three years) you can lease a $39500 vehicle for 15,000 miles a year (i.e. 42.8 cents/mile), plus a $3000 downpayment. [Note: the lease numbers are getting better with time as companies achieve greater scale]. When the user finishes the lease, they will not have an option to buy the vehicle, but instead the vehicle gets added to the autonomous fleet. In essence, the company will get back a 45K mile vehicle, which is depreciated down from $39500 to $17240 (after you subtract the $19260 + 3000 in lease payments).
If you compare this to the "real" market-based depreciation studies (eg: done by Autolist, see here or here, or pictured below) for Model S, X and projected for Model 3, these cars are expected to hold upto 71% of their value at 50K miles {compare that to over 50% paid off via lease payments above}. Note that this is only for the "hardware" of the vehicle; as the software value increases, and the ability to associate w/ the autonomous network for revenue streams, the depreciation curve will be lower sloping in general, and will be an appreciating asset for those who capture those revenue streams.
Since the Model 3 vehicle will be built in the future for a longer life (eg: claimed a million miles, and the battery pack has a life of 300k-500k miles), and battery replacement can be done at the level of individual battery modules alone at $5-7K (instead of changing the whole battery pack), a $17K post-lease depreciated asset in the company's hands can conservatively run for about 250-300K miles. At a revenue realization of $1 / mile (current ride share rates are $2.50-3.50/mile in US), this $17K asset will yield a revenue of >$250K across say 5 years, or $50K / year, implying an excellent asset turnover (or sales-to-capital ratio) and overall lifetime value (= 250k - 17K) of the asset.
I believe that if regulators do not allow full self driving with a human in the car quickly, they may allow the car to drive itself safely to be re-positioned WITHOUT a human in the car (as long as they are convinced of its safety to other vehicles and humans on the road. This means that even though you may need to "drive" the car, all you need to do is punch in the destination (or speak to the car) and have your hands on the wheel during the drive (and have a valid driver's license) to take over in an emergency, or can over-ride the auto-pilot for driving end to end or disengage it for a period of time. After you get off the car will drive itself to the next fare opportunity.
Think of this as similar to you getting a trip-specific rental car service (eg: from Hertz or Avis or Enterprise or Turo), but where the car comes to you. A similar model could also be followed for a 1-day rental similar to room sharing (eg: Airbnb, Oyo) for hospitality (eg: as offered on Turo today, see graphic) or very short term rentals (eg: an hour's test drive) or a subscription model (eg: a car for a week or a month across locations etc). Another demand lever is to also sign up users for a subscription program as part of its future autonomous fleet , which could be based upon a bundle of miles and time (eg: weekly or monthly subscriptions). While these initially do not need autonomy - the allure of the Navigate on Autopilot will attract a lot of users, and ideally improve the brand aura.
With improvement in battery costs and more cars deployed via leases or innovative financing across the world, the learning curves and economies of scale in battery manufacturing will lead to lower battery pack prices, which in turn drops the price of electric vehicles and the price of leases which will supercharge this cycle.
Similar trends are happening with other automakers. For example: companies like Waymo and Jaguar announced a partnership to partner and devote ~20,000 I-Pace vehicles towards the Waymo fleet from 2020 for autonomous ride sharing. GM's Maven ride sharing service launched in several cities is and is promoting use of electric vehicles as part of Maven Gig, including a dedicated fast charging infrastructure partnership with EVGo.
In Europe, Car2Go and DriveNow have joined forces to become ShareNow (JV of BMW and Daimler), and a key piece of the announcement is this: "One central component of the future strategy is electromobility. With four European cities operating with 100 percent fully electric fleets and 13 additional cities operating with partial electric fleets, SHARE NOW is already the largest provider of free-floating e-carshare. Over 120 million kilometers / 74.5 million miles have been driven worldwide to date in approximately 3,200 electric vehicles."
Summary:
I believe these are just a few of the synergies that will enable more options in a Transportation-as-a-Service (TaaS) future. Specifically, I believe we should not expect a single switch from human-driven cars / ride shares to fully autonomous ride shares. With different levels of autonomy, and efficiency / reliability, the operational efficiency and financing / insurance costs of an electric vehicle (EV) will drive more substitution of ICE to electric and more vehicle miles traveled by electric vehicles in ride sharing or rental contexts. The availability of convenient fast-charging and destination charging infrastructure will also be crucial to support his transition.
Regulatory approval could also come in stages. For instance if approval for a car without a human inside can be allowed to go on roads (or a limited number of roads), this will allow re-positioning for a short term rental service or pickup/drop off service even if the human has to drive it once the car arrives.
Note that we have not discussed Bidirectional grid services/V2G value streams or demand response (DR) possibilities of aggregating the private EV network fleets for demand response via smart chargers (eg: in the California ISO market) and sharing that value with the users.
The electric + autonomous vehicle can indeed become an "asset" in the true sense of the word, of yielding a stream of cash flows whose present value is larger than the cost of acquisition of the asset. For example, it has been suggested that a single heavily used robotaxi (capex of ~$40K) can yield a net present value of $200K (i.e. net of capex) over its lifetime, and $30K/year gross profit. In a secondary market, when sold, the asset value can appreciate since it will no longer be seen as a depreciating single vehicle, but as a fleet asset that yields residual cash flows. The appreciation is a direct function of the software capabilities that allow it to participate in multiple value networks.
With more electric vehicles on the road given the lower cost EVs and financing/leasing etc, there will be more data collected to train the deep neural networks. In AI, data is the lifeblood of improvement in accuracy and handling corner cases. The improvement in such algorithms' capability is super linear with the amount of data collected. For players who have fewer vehicles on the road, a combination of collecting data from ICE vehicles fine-grained driving patterns, and simulated data to expand the training data set may be necessary to reach adequate capability thresholds.
The future indeed is exciting, and may start happening faster than we think as the fleet of EVs with autonomy starts increasing on the road in major markets.
LinkedIn: Shivkumar Kalyanaraman
[Disclaimer: The views expressed here are personal and meant for information/education purposes only and not representative of his employer directly or indirectly. Any mention of company names or use of images are for illustrative examples only. ]
Twitter: @shivkuma_k
If you like this article, please check out these articles: Towards Scalable EV Charging: Hidden Costs of Level 2 Charging at Scale, Towards Scalable EV Charging Infrastructure (Think Outside the Gas Station Box), "Commercial Electric Vehicles (EV) Fleets: The Stealth Growth", "Towards Affordable, Ubiquitous, Ultra-Fast EV Charging: Part 1: Need & Battery Issues", "EV Taxi Fleets & Ride Sharing: Poised for Huge Growth", "Shared EV Transportation in India", "Understanding the Rs. 3/kWh bids in India in 2017", "Distributed / Rooftop Solar in India: A Gentle Introduction: Part 1","Rooftop Solar in India: Part 2 {Shadowing, Soiling, Diesel Offset}", "Rooftop Solar in India: Part 3: Policy Tools... Net Metering etc..." "Solar Economics 101: Introduction to LCOE and Grid Parity" , "Solar will get cheaper than coal power much faster than you think..", "Understanding Recent Solar Tariffs in India", "How Electric Scooters,... can spur adoption of Distributed Solar in India," "Solar + Ola! = Sola! ... The Coming Energy-Transportation Nexus in India", "UDAY: Quietly Disentangling India's Power Distribution Sector", "Understanding Solar Finance in India: Part 1", "Back to the Future: The Coming Internet of Energy Networks...", "Tesla Model 3: More than Yet-Another-Car: Ushering in the Energy-Transportation Nexus", "Understanding Solar Finance in India: Part 2 (Project Finance)", "Ola! e-Rickshaws: the dawn of electric mobility in India", "Understanding Solar Finance in India: Part 3 (Solar Business Models)" , "Meet Olli: Fusion of Autonomous Electric Transport, Watson IoT and 3D Printing".