< Part 1: The trip economy
Consultants have created a framework for mobility disruption: Connected-Autonomous-Shared-Electric (CASE). It's nice: it spells a word. It can be made into a different word if you rearrange the letters around: ACES. Some people prefer this second word and others prefer the first. But unfortunately neither captures what makes these particular forces disruptive or how they interact.
The terms we use to describe the technologies that make up the CASE acronym - electric vehicles, connected cars, autonomous vehicles - assume that the natural way to deploy them is in the form of an automobile. But the fourth term - "shared" - is the most important and it hides a forest by disguising it as another tree. Business models are what make markets and determine the flow of technology. The trip economy gives a different way of thinking about new technology rollout. From the perspective of a component supplier rather than a carmaker, there are many kinds of vehicles that might make use of a certain technology: sidewalk robots, bicycles, mopeds... it's simply a question of standardization and distribution channels and these are determined by business model. Just as the building blocks for cars have become more modular, so too the building blocks for all vehicle types.
So let's look at these technologies - which are all pretty neat in and of themselves - and how they tie into the trip economy.
For a brief moment in 1900, electric taxis made up about a third of vehicles in New York City (another third were powered by steam and the remainder ran on gasoline). But it was gas-powered, internal combustion powertrains that won out. Internal combustion worked because fuel is dense in energy and easy to move and as the technology scaled, costs came down. But we're now entering the final act of the internal combustion engine.
The "peace dividends of the smartphone wars" is a neat way to describe how smartphone innovation has disrupted other sectors through software innovation as well as cheap, low power silicon and sensors (which we'll get to later). But consumer electronics and then smartphone innovation have also served as a pathway along which battery innovation has been catalyzed to create a massive industry of low cost, reliable lithium battery cells with a supporting network of power outlets and an energy grid to support them.
This evolution of batteries has helped birth a new energy innovation battleground centered around electric vehicles. Technology is still helping to improve fuel efficiency and reduce emissions from internal combustion vehicles, but the tailwinds behind electric vehicles - including regulation, infrastructure build-out, consumer mindshare and Elon Musk cat memes - have become very powerful. An electric future is now inevitable.
As we reach diminishing returns on conventional vehicles and the opportunity for electric vehicles comes to the fore, where does the arbitrage lie in shifting to a new powertrain? The answer to this question is related to utilization.
Internal combustion engines have low upfront costs, but fuel is relatively expensive so costs go up for each mile that is driven. Electric vehicles are the opposite: batteries are expensive, but each additional mile is cheap. This means that the more miles an electric vehicle drives, the more cost-effective it becomes to operate it.
This tradeoff has significantly improved over time as batteries have become cheaper: according to Bloomberg New Energy Finance battery prices have more than halved since 2013 and are on a steep downward trajectory driven by both investment and policy. Based on current and projected costs of electric and internal combustion vehicles (including upfront costs, maintenance and fuel) the current utilization required to make the average electric vehicle more cost effective is around 25,000 miles per year. That number is on track to decrease dramatically to about 12,000 miles by 2030 (see chart above).
Part of the reason electric vehicles are cheaper at high utilization rates is that they have a third as many components as gasoline powered vehicles. This relative simplicity reduces servicing costs and improves reliability since there are fewer points of failure which further strengthens their value in fleets that are constantly in use.
Relative simplicity also makes EVs easier to design and manufacture. Complex internal combustion powertrains have served as defensive moats for incumbent carmakers and suppliers.
As these barriers have come down, a slew of new electric carmakers have followed in Tesla's wake with alluring branding and catchy names like Rivian, Lucid, Arrival and Rimac. China sees an opportunity to ride the next wave of technological innovation and has given significant financial and regulatory support to a new wave of domestic startups with similarly impressive vehicles but somewhat bewildering names like NIO, Li Auto, XPeng, Weltmeister and Byton. These vehicles all have a more fundamentally digital architecture which adds to their appeal and complements the sensor-pervaded, silicon-powered direction in which vehicles are moving. Billions of dollars are being poured into all of these new carmakers and several others besides.
As scale increases and battery costs continue to decline, the value proposition of electric vehicles will only improve, accelerating the flywheel of new investment and adoption. Already the majority of municipal bus sales are electric (given that short, predictable routes make their total cost of ownership significantly lower when electrified) as are passenger vehicles in Norway (helped by generous subsidies and strong incentives). The low energy density of batteries which limits range combined with delays when charging make electric vehicles less fitting for heavy duty commercial use cases, but passenger vehicle and light commercial adoption is expected to explode over the coming years according to BNEF.
Given that trips are provided by fleets of vehicles, it is within the trip economy that electric vehicles find a natural home. This is especially true for fleets that own their vehicles and can directly calculate and control the full costs of ownership.
Technology is transferable. Just as battery technology developed largely for consumer electronics has accelerated the electrification of cars, so too the electrification of cars can help reinvent the form of other kinds of vehicles. Cars are big because they are decathletes, but vehicles right-sized for particular purposes are in many ways a better fit for electrification.
This right-sizing extends most clearly into micromobility. Electrification works particularly well with small, lightweight vehicles since they require less energy and can be recharged quickly. Since scooters are in some sense large toys, it makes sense that they are powered by the same thing that powers gameboys. Replaceable batteries also make fleet recharging easier.
This rollout is also easier to scale. Electric vehicles can more than double the power draw from a household, creating challenges for the grid especially at the local level where infrastructure has not been designed to support this scale of demand. Fleets of larger vehicles can create even greater strain. The significantly lower charging requirements of micromobility form factors bypass this challenge.
The potential to port core EV technology into new form factors is already starting to play out. 48V electrical systems were developed to improve fuel efficiency and performance of internal combustion vehicles through stop-start motors and turbochargers. Now the same technology is being used by suppliers such as Valeo to power electric rickshaws and motorbikes in India and to create an integrated electric bicycle motor targeted at the European market where electric bicycles are exploding in popularity. Similar trends are democratizing battery cells, packs and battery management software developed for the passenger vehicle market. In this way, electrification is a force that accelerates the trend of unbundling and optimization of vehicles according to the specialized needs of the trip economy.
Many of the benefits of electric vehicles also apply to hydrogen fuel cell vehicles: a simplified, silent electric drivetrain powered by an energy source that doesn't create pollution in the process of moving the vehicle.
While it is clear that internal combustion engines are losing (and diesel has already lost), it's not yet clear whether fuel cell vehicles can share in the winning. They are still far too expensive. The new Toyota Mirai at $68,100 (or about $58,000 in Europe after subsidies) costs significantly more than comparable non-hydrogen vehicles. Fuel cell vehicles at scale are still a long way off.
In contrast, electric vehicles are already price competitive with internal combustion vehicles today (especially when factoring in subsidies) and are rapidly growing their market share, creating a challenge for fuel cell vehicles which are competing with EVs for attention, regulatory support, investment and consumer mindshare.
Besides the higher costs of manufacturing fuel cell vehicles (which lack the kinds of economies of scale of EVs have started to achieve), the current upstream costs of producing hydrogen through electrolysis are still too high to justify most applications. However, there are many viable uses including heating, agriculture and industrial processes such as steel production that create a pathway to scaling demand and justifying investment and regulatory support to bring down energy costs.
Supported by sufficient demand, hydrogen has the potential for long term costs reductions. Hydrogen can be produced in a diversity of ways, helping to support scaling its production into the broader energy value chain. However, the primary focus is on lowering the cost of so-called "green hydrogen" which is produced through electrolysis in a process that does not emit carbon.
According to McKinsey, the cost of green hydrogen in Europe (currently around $6/kg) can be reduced by 60% primarily through scaling and improving electrolysis while decreasing the cost of offshore wind energy production. This would bring the cost to about $2.6/kg making it cost competitive for a wide variety of use cases.
Hydrogen has broad applicability because it is essentially a much denser battery. In transportation, this ties directly to refueling and range. EV batteries are very efficient but they have relatively low energy density and take quite a lot of time to recharge (even with fast charging). This means they work best when vehicles are lighter or cover shorter distances or have significant downtime.
In contrast, hydrogen's massive energy density, about triple that of diesel or gasoline and ten times that of lithium-ion batteries, means shorter refueling times and more uptime. It also gives hydrogen vehicles greater range while adding less weight to the vehicle, meaning they can go further without needing to refuel. This makes fuel cell vehicles most effective in use cases where electric vehicles are weakest such as trucks with heavy payloads or that cover large distances.
In addition, hydrogen's discrete energy economy can help offset the grid-dependent nature of EVs while also bypassing the long tailpipe problem. Electric vehicles require that the grid be able to pick up the energy demand that has thus far been satisfied through fossil fuels, creating strain on particular points in the grid network and a significant increase in overall peak demand.
In contrast, hydrogen is produced and moved independent of the grid (much like fossil fuels). Though hydrogen fuel requires specialized refueling infrastructure, an elaborate system for moving and storing flammable liquid fuels already exists which can be adapted to hydrogen. This combined with higher energy density and much shorter refueling times means that for fleet use cases where range is important, the cost of hydrogen infrastructure is lower than EVs on a marginal basis and is easier to scale since it can be built around the fleet.
So while in the short term, fuel cell vehicles suffer significant challenges achieving scale, in the longer term they have potentially more appealing economics when considering the total cost of ownership since there is more room to create cost efficiencies.
Given strong pressure to achieve climate targets and hydrogen's applicability to complementary use cases to electric vehicles, there is reason to believe that government support, which is already significant in Japan and Europe, will continue to grow giving fuel cell vehicles a viable track to scale alongside EVs, even if right now they still seem stuck in the starting blocks.
Connected is the C in CASE and also the C in ACES. That's nice, but "connected" doesn't really tell the whole story. Digitization is a better term for the set of things going on here, and in combination with T for Trips (as per the preceding section), you can spell DATE, which is an acronym worth swiping right on.
But back to digitization: what does it mean? Three things.
The foundation for digitization is silicon. Computational power and therefore functionality is defined by the processors available. Silicon with low power requirements and high performance has become widely available along with modems to enable connectivity.
The challenge is integrating powerful processors into vehicles. Because the existing network architecture of vehicles is convoluted, with weak processors distributed across the vehicle, it is hard to move to a more rational, centralized, software-defined architecture. Furthermore, this dependence on numerous kinds of low margin microcontrollers has made the automotive industry particularly vulnerable to a recent supply squeeze in the semiconductor industry.
The wheels are starting to turn, albeit slowly. Most automotive suppliers are positioning for a shift away from discrete systems bundled with one or multiple ECUs towards a more centralized architecture. Aptiv has been the most eloquent in outlining a vision for this shift to what the company calls Smart Vehicle Architecture (SVA), which enables centralized compute, software-defined functionality and rationalized power supply.
Major acquisitions (and attempted acquisitions) over recent years in the silicon sector —Nvidia/ARM ($40B), Intel/Mobileye ($15B), Qualcomm/NXP ($47B, blocked) and Broadcom/Qualcomm ($117B, also blocked on political grounds)— have all had an eye towards the coming opportunity in automotive silicon. The blocked acquisitions and US/China geopolitical tensions (see: Huawei, SMIC) help highlight how critical the semiconductor industry is and its role in mobility will be going forward.
All of these changes on the silicon level form a foundation for software that can increasingly virtualize functions currently done by hardware. Much like the iPhone became the platform for a shifting ecosystem of apps, the software stack of vehicles is likely to serve as a platform for new software-enabled features and services, especially for next generation EV manufacturers such as Tesla, Rivian and NIO and for micromobility vehicles such as scooters and e-bikes.
Software is flexible and dynamic and is a fundamental enabler of virtually every other trend transforming mobility, from the mobility marketplaces that power the trip economy to electrification through battery management software and the perception, path planning, mapping and teleoperation layers that form the autonomous technology stack. Software also allows information to be generated that powers business insights or vehicle data monetization. To the extent that the platform is open and extensible, third party developers can create new experiences and expand the array of offerings built on top of this digital layer. This is especially true when software is organized into an operating system with clearly defined functions that can be used create new experiences safely while allowing distribution to a large audience of customers.
Software becomes a lot more useful when combined with connectivity. There are a few ways in which this is creating value.
The first is broadly called telematics, which leverages data going out of the vehicle such as driver performance and GPS location to generate business insights (When will a truck arrive? Is the vehicle being driven safely? Does this scooter need to be repaired or repositioned?) and is especially useful for fleet operations.
In addition to data going out of the vehicle, data coming into the vehicle can enable new functionality, especially in the form of over the air (OTA) software updates.
Amongst carmakers, Tesla is quite far ahead in leveraging this functionality and customers love it. Meanwhile, mobility fleet operators such as scooter sharing companies use OTA to manage their fleet (e.g. dynamic geofencing) or address issues (like braking problems).
Teleoperation is another process which relies on data transfer through connectivity and ties directly into autonomy, which we will cover soon.
Finally, connectivity also allows onboard silicon and software to offload certain functions to remote data centers running powerful servers. This is particularly useful for processes like machine learning that require significant volumes of data and computational resources. The cost of moving data between the vehicle and the cloud along with latency limits this process, but faster networking (like 5G) and volume discounts for moving data improve this tradeoff and will enable new functions.
There's another sense in which the smartphone wars matter: smartphones are at war with every other product category. They define consumer expectations, setting them high, while a rapid rate of evolution allows them to constantly absorb new functions.
When it comes to digitization in any vehicle type, the litmus test is whether a given functionality is better or more useful than what you can already do with your smartphone. Remember that your smartphone includes the latest silicon, 5G connectivity (whatever that means exactly) and actively leverages the cloud. When it comes to software, smartphones have become so mainstream and the supporting software ecosystems so broad that any developer needs a very good reason to bother building for a different platform.
How much built-in digitization makes sense depends on a variety of factors including vehicle type, value of the asset, available power and use case. Sometimes having a well-positioned smartphone holder is the right choice. Often the question is how to make the vehicle interact most effectively with smartphones and their ecosystems of apps. For instance, carmakers are increasingly relying on Google and Apple to display smartphone functions such as music and navigation on a screen within the vehicle in a way designed to support driving (or at least minimize distractions). Similar considerations apply to fleets.
Accounts of the Messiah vary widely. So too with autonomous vehicles. On the one hand, their full promise is a distant vision. Full, hard core, "Level 5" autonomy - in which a safari vehicle can successfully veer off the road to track a lion through the underbrush - is more a theoretical concept than a foreseeable reality. Meanwhile other versions of autonomous vehicles are here amongst us, blessing our lives with their benign presence. Take for instance Rio Tinto's fleet of autonomous haul trucks, larger than two story houses, carrying iron ore around mines in Pilbara, Australia. Or Kiva robots moving goods around Amazon distribution centers. Already 42 cities operate a total of 64 fully automated metro lines (half of which are in China). And cars driving on the road today can manage long stretches of highways without any driver intervention (though Elon Musk's prophecies about the near future happen to make reality harder to discern).
There's been a lot of hype (and projections) around autonomous technology and the glorious future it will usher in. This vision involves broad scale optimization, beating the swords of inefficiency into productive autonomousploughshares. The Messiah is whatever you want it to be and yet also broadly misunderstood.
A way to understand autonomy that steers clear of magical thinking is through the lens of labor arbitrage: a robot that does the work of a driver is replacing the work that would otherwise have to be done by a person. Framed differently, full autonomy is on the far end of a spectrum of labor efficiency, with several preceding steps along the way in which automation improves driving efficiency.
Labor efficiency in transportation can be measured in time needed from a driver in order to complete a specific task. Road throughput for instance matters because it increases driving efficiency, reducing the time people have to spend driving in order to complete a trip. It can also mean that the task is done more effectively with fewer issues.
Labor efficiency can also be viewed as the ratio of operators to vehicles. Normally there's a ratio of one operator per vehicle. But autonomy can reduce this ratio. For instance, ADAS in passenger vehicles takes over a portion of the driving task such that the driver is required to dedicate less mental effort to driving, a benefit consumers are willing to pay significant amounts for.
Now consider teleoperation, the ability to remotely operate a vehicle. When a driver doesn't need to be in the same location as a vehicle, the role of labor changes. Teleoperation improves driving labor efficiency by:
But teleoperation makes most sense in combination with autonomous functionality. Initially, augmented information can improve the performance and increase the safety of remote drivers much like ADAS can. As autonomy improves, the ratio of operators to vehicles drops as some tasks are done automatically and the operator need only intervene at moments when the autonomous system runs into issues. At low levels of autonomy, the operator to vehicle ratio is close to one but tends towards zero as the autonomous system reaches maturity.
These steps of autonomy combining with teleoperation to improve the efficiency of labor in mobility are a bit like a caterpillar taking bites of a juicy green leaf. Each bite is an incremental building block of data and functionality that increases the amount of work that can be done without human involvement. At the end of this process, autonomy, like a butterfly, emerges, creating efficiencies that would not have been possible without the incremental support of teleoperation.
Once autonomy is framed as a spectrum with teleoperation supporting its growth like a cocoon, it's possible to be more granular about where it will be deployed and the value it will create. Fundamentally, complexity is at tension with scalability.
The broadest problem in autonomy is replacing the work people do driving passenger and commercial vehicles. This problem is hard because road infrastructure is diverse and a very broad array of unexpected scenarios can arise. The prospect of solving this problem is exciting since in the US alone there are about 220 billion daily trips meaning that a huge amount of labor could be freed up by autonomy. This is the goal companies like Waymo, Cruise and Aurora are pursuing. But for now, the where of these vehicles is quite limited (Chandler AZ, San Francisco CA) and the when keeps being pushed further into the future.
In the meantime, as we await this massively generalizable form of autonomy, the same approaches can be used to solve discrete, constrained problems. One way to reduce complexity is to reduce speed. Another strategy for constraining the problem is to operate in closed environments like mines and warehouses or trucks driving only on highways. Through this lens of solving smaller problems, autonomy is effectively bubbling up from the bottom and is stitched together as part of broader systems rather than applied top down to solve all problems at once.
As autonomy reaches maturity, it has the potential to accelerate the trip economy's tendency towards right-sizing and optimization, especially in mobility services managed by fleet operators. In general, vehicles are designed around drivers; autonomy introduces a very different engineering paradigm. Furthermore, in cases like ridehailing, where drivers currently own most of the vehicles in the network, autonomy will force fleet operators to directly manage their fleets rather than outsourcing this work to drivers. This will give them significantly greater control over the fleet and accelerate the optimization feedback loop in the process (Part 4 has more on this). It will also create some thorny questions around the role of labor in the transportation economy, which we will cover in Part 5. But first, let's consider bundles, both furry and fluffy.
Share this chapter with others: