The Year 2040 – Somewhere in Silicon Valley #
It’s 8:07 am and my next-door neighbor, cheapskate Charlie, has been waiting outside his door for a few minutes for his ride, which is guaranteed to be at his house within a 10 minute window. He looks at his garage and is reminded that he will soon be renting it as storage space to his neighbor, Rich.
As the electric Gee-Auto arrives, Charlie notes that another neighbor, tightwad Tom, is joining him today and on their journey, they will pick up parsimonious Paula. Despite sharing a vehicle with two to three people each day, the efficiency of a packet network of autonomous vehicles has reduced his average commute time from 30 minutes to 23 minutes, eliminated the need for auto insurance, and given Charlie the opportunity to play his virtual piano on his morning commute, instead of focusing on the car in front of him.
Parsimonious Paula likes the Gee-Mobile service as she no longer has to rely on the discontinued and obsolete county transit. Her monthly subscription to the Gee-Mobile service is comparable to what she used to pay for a monthly transit pass and she doesn’t have to walk half-a-mile in the rain to catch a bus. It would make bringing groceries home easier, but Gee-Autos have been delivering goods directly to homes for decades.
It’s 8:15 am and across the street, just like every workday, a Gee-Auto meets my spendthrift neighbor, Rich, at his doorstep exactly as he opens his front door. He hops in the Gee-Auto and waiting for him is a morning latte, a freshly toasted bagel, along with morning news, entertainment and education tuned especially for his viewing, listening and olfactory pleasure.
Rich has a tinge of disappointment that his 15 minute commute (which used to be 30 minutes before the arrival of self-driving, always-connected vehicles) couldn’t be just a little longer, as he really enjoys this daily ritual of breakfast and relaxation in a moving pod. That disappointment is soon forgotten, as he realizes today is the day when a contractor and his team of droids will begin the conversion of his garage into a tricked-out, man-cave.
Along the way, the Gee-Auto’s speed is constantly and automatically adjusted to traffic conditions. The queuing algorithms are working especially well these days and intersections that were formerly regulated by stoplights are now sophisticated roundabouts and it will be a non-stop trip for Rich. There is one stop for the Gee-Auto transporting Charlie and that is to drop off Paula at her banana stand.
Like most days, Rich and Charlie arrive within a few minutes of each other at the Acme Anvil Company (Charlie is the CFO and Rich is in marketing). They wave adieu to Tom, who works about a half-mile away, and go about their day. In the meantime, the Gee-Auto that had transported Rich to work slips into the median, where an embedded wireless charging pod rapidly recharges the hybrid super capacitor-graphene battery system, before receiving its next assignment to pick up groceries for delivery to another Gee-Mobile subscriber.
[Note: The above scenario of an automated people mover seems ridiculous, but it wasn’t too long ago that the idea of talking to one’s phone to get directions would be absolute lunacy. The idea of an on-demand transit system providing door-to-door transport goes back to at least the mid-1970s, as the first major expansion for Silicon Valley’s public transit system was such a service, Dial-a-ride (dial-a-ride used the old school telephone to beckon a mini-bus directly to one’s residence). Dial-a-ride didn’t scale, however, as the staffing and equipment costs were greater than the traditional public transit approach of aggregating people at transit stops.]
Although fictional, the above story isn’t science fiction, as the technology now exists to make the above scenario real. Many companies could potentially implement such a people transport system, including car manufacturers, auto-rental and logistic companies, but it is likely to be outsiders (Amazon, Walmart, Google, etc.) that disrupt this multi-trillion dollar industry.
The focus of this article is Google and how the elements it already has in place could be stitched together to create an end-to-end, subscription (as well as Pay Per Ride) people transport service that generates tens of billions of new revenue, while building upon its existing businesses.
One of the oft-cited barriers to the autonomous car is the question of who is liable in the case of an accident (e.g., the manufacturer, the driver, etc.)? A subscription model doesn’t remove the liability factor, but by taking a holistic view of the driving experience and owning the “last mile” transport method, Google could greatly reduce its exposure.
Like its cloud services, Google would have complete control over the design (ensure no single points of failure), the maintenance (no mechanical error by ensuring equipment is always up-to-date) and the software (e.g. secure it from hacking).
Further, removing the constraint of having to accommodate a driver would allow for a rethinking of a vehicle’s design (see the above video). There is no need for a steering wheel, which could change the form factor, while improving the safety of the passenger who occupies the driver seat.
The need for windows goes away and could be replaced with electronic screens, such that one could choose the environment that he wants to see (think advertising space for Google). Without windows, presumably the vehicle’s body could be made stronger (e.g. more cross-members where the windows would have been). Additionally, the seats could be placed backward as there is no longer a need to face forward.
Last week’s announcement that they have designed their own prototype car is consistent with other initiatives, like Google Fiber, where they want to control the entire experience. A custom design also reduces vehicle cost by eliminating overhead that an individual consumer normally pays when she buys a car from a dealer (which passes on the sales, marketing, engineering and other overhead costs of the manufacturer, along with the dealer costs, etc.).
Google, along with other entities, have a number of initiatives that set the stage for a subscription-based, autonomous transport system, including:
- Google has proven it can create an autonomous vehicle that can drive hundreds of thousands of miles without an accident.
- Google’s Waze application, coupled with their Google Maps already provides a real-time view of traffic allowing drivers to select the best route. Having a vehicle automatically make the decisions as to the best route is the next step (and safer). The more vehicles that are directed in this manner, the better, in terms of route optimization (i.e. traffic reduction); the Gee-Auto and their control become more and more like the Internet, as the underlying signaling improves the throughput of the overall transportation network.
- Google, as well as Amazon and others, are investing heavily in on-demand delivery of goods. This effort is a great testing ground to understand the best routing of vehicles. As Google is wont to do, they are also building the associated apps and signaling technology via the broadband network to ensure orders are relayed through the delivery chain. If Google can prove this model with a driver, then eliminating the driver via an autonomous car makes the model work that much better.
- The idea of a subscription service for a car rental isn’t new, as evidenced by the rise of ZipCar in urban areas. The autonomous car would allow this concept to spread into suburban areas, as the cars would automatically appear at the subscriber’s house [Added 12/23/14 – the idea of an on-demand, shared, last-mile car service may become reality in 2015, as Singapore is looking to open up one of its neighborhoods to such an experiment].
- Relay Rides, Uber and Lyft provide models for the electronic dispatch of vehicles – albeit with human drivers – using a Smartphone or tablet. It isn’t a stretch to envision the elimination of the driver. It is important to note that Google Ventures is already an investor in Relay Rides and Uber [Note, since this article was published, Uber has suggested that the elimination of the driver could be part of their long-term plans – one industry executive even predicted that Uber might purchase a auto manufacturer, so that they could control the experience and have cars that last a million miles].
- The concept of a relatively low-cost ($24k), electronically controlled electric pod car is close to reality with the soon to be released vehicle from LIT Motors; a small San Francisco start-up that promises to disrupt the auto industry with its Silicon Valley business model.
- Building a car with screens, instead of windows, provides Google with an opportunity for more “ad-space”. This is ad-space that is not only location-aware, but location-directed (e.g. sensing the rider might be hungry for a certain food item, it would be easy to automatically reroute to one’s favorite restaurant and provide incentives for stopping at said restaurant).
- A Google Fiber/Wireless backbone, although not necessary, could be tuned to off-load signaling information emanating from the vehicles peer-to-peer communications systems. These two networks (P2P vehicle and the Fiber backbone) could become an integrated central nervous system for the network of vehicles. [Added 6/6/2014] Google’s request for a Statutory Temporary Authority from the FCC for the nationwide testing of millimeter frequencies (77 GHz) looks to be part of an effort to detect objects around a car. [Added 8/25/2014] Further, the conversation on Vehicle to Vehicle communications continues with the NHSTA’s release of it’s Advanced Notice of Proposed Rulemaking.
Why would Google ever want to jump into such a seemingly tangential business model of being a Subscription Vehicle on Demand service provider? Simply, a project of this sort could move their revenue needle, produce great margins and augment their advertising business. As importantly, the notion of organizing the world’s atoms is akin to its initial mission of organizing the world’s information.
For simple modeling purposes, let’s assume the IRS reimbursement rate of 55.5 cents per mile (gas, maintenance, amortized car payments, etc.) and that the average person drives 10k miles per year (AAA estimates 59.5 to 97.5 cents per mile for 10k miles/year for a small to large sedan, respectively). That would mean $5,550 a year in transport costs per car or approximately $460 per month.
It isn’t too difficult to imagine a 3 tier subscription offering, similar to what Google is doing with their broadband offering to meet the needs of the various customer profiles:
- The Parsimonious Paula Offer – $125/month – Gee-Auto guaranteed within 10 minutes – have to share with others, advertisements, plus goods delivery within 8 hours – 500 miles per month (overages apply).
- The Mainstream Mary Offer – $300/month – Gee-Auto guaranteed within 5 minutes, sometimes have to share depending upon demand, limited advertisements, plus goods delivery within 4 hours – 1,000 miles/month limit (overages apply).
- The Regal Rich Offer – $1,000/month – Gee-Auto is ready when the person opens their door, no sharing with others and no advertisements, plus goods delivery within 1 hour – Unlimited distance per month
Further, assume take rates of 10% for the Regal Rich offer, 40% for the Parsimonious Paula offer and 50% for the Mainstream Mary offer, the weighted average would be $300 per month per subscriber (33% less than the assumed conservative average of $460/month in transportation costs).
Assuming a 5% market share of today’s 18+ population, this would mean approximately 9 million subscribers or about $2.7B monthly or > $32B annual business, not counting any uplift to existing businesses (e.g. advertising, broadband, etc.), on-demand business (taxi-replacement business) or fleet/logistic replacement.
Because of the sharing nature of the business, Google’s costs would be lower than the IRS reimbursement rate of 55 cents/mile. Even the most expensive option in the above scenario would be shared (e.g. once a Gee-Auto pod drops off one person, it could pick up another nearby person). Assuming a sharing ratio of 1/3 (one Gee-Auto for every 3 people [8/19/2014 update – When I wrote this, the 1/3 ratio was a gut-feel guess. As it turns out, some MIT scientists using mathematical algorithms and real-data from Singapore determined that a 1/3 ratio is about right as summarized here about their white paper]), the costs, based on the IRS figures would be $153/subscriber/month (1/3 of the single driver’s cost of $460) or almost 50% gross margin ($153 costs versus $300 in revenue/subscriber); not a bad business and with $30B+ in revenue, a business that is approximately 50% of Google’s current business.
Granted, there would be significant capital costs to such an endeavor, but, because electronics and software are the significant cost components associated with the above scenario, cost reductions would more closely follow Moore’s Law than the traditional cost for building automobiles. There are also costs associated with upgrading roads, etc. that would need to be factored in as part of a capital build. Still, by building this on a city-by-city basis over time, much like Google Fiber, the capital costs would drop with each deployment. Even at $20k per vehicle, the capital costs to create 3 million vehicles would be $60B; not insignificant, but within the realm of possibility given current costs for low-end electric cars.
There are several upsides, both the aforementioned uplift to their existing businesses, as well as opportunities to reduce expenses relative to traditional transportation systems, as well as find new revenues:
- Lower Insurance Costs: Google would probably self-insure, given the sheer volume of business, as well as the confidence they would have in their technology and the indemnification clauses their attorneys would include in their subscription agreements. Self insuring would remove the costs of the insurance company middleman. Additionally, given the potential improvements in safety from autonomous vehicles (Google suggests that human error causes 90% of the 1.2 million vehicle deaths each year), the effective cost of insurance would be lower than the costs for insuring human-driven autos.
- Lower Operational Costs: Being all-electric, the operational costs from maintenance and fuel would be less than traditional hydrocarbon vehicles. Additionally, it wouldn’t be a stretch for Google to create a network of its own power stations (which, as alluded to in the above story, could be in medians and other non-usable areas).
- Local Subsidies: At $125 per month, the Parsimonious Paula tier is more than 10% cheaper than the existing Silicon Valley public transportation option (a monthly pass on Silicon Valley’s VTA is $140). Given that public transit authorities operate bus systems at a loss, it might be cheaper for a transit authority to pay Google on a variable cost basis and retire the bus systems (particularly in suburban areas). Google probably would run the transit system without subsidies, as the political benefits of saving the local taxpayers money would outweigh the marginal revenue.
One of the brilliant insights from the Google Fiber management team is its understanding of the importance of speed; not just speed in broadband access, but speed to market. The longer it takes to deploy Google Fiber, the higher the costs of make-ready and the more opportunity competitors have to thwart its efforts. As such, one of the most important factors in determining where they deploy Google Fiber is the willingness of local cities and agencies to work with them to smooth out the barriers to deployment (e.g. obtaining permits, rights-of-way, etc.).
The Google Fiber project has forged the sort of local relations that would be necessary to implement such a revolutionary approach to transportation. A project of this scale would require working with local government to support infrastructure improvements, such as distributed power charging stations (or some equivalent, such as solar roadways), improvements in traffic light signaling (making it more dynamic, based on real-time traffic demands or [link added 8/17/14] eliminating it as seen in this video) and other road improvements (e.g. roundabouts).
It’s not too difficult to imagine Google pursuing a nationwide competition like it did when it introduced the Google Fiber concept. If Google were to target a community with a population of 100,000 adults and assuming a 5% subscription rate, with a 1/3 ratio of vehicle per subscriber, they would be looking at 1,700 vehicles and assuming a near-term cost of $100k per vehicle, this would be a $170M investment; an amount that is pricey, but would provide a good field test and refine the commercial project, just like Kansas City did for Google Fiber (this is in the realm of possibility, as Google recently announced that they will be building 100 prototype vehicles for testing purposes).
Like the Google Fiber project, which received over 1,000 applications from communities of all sizes, a Gee-Auto contest would grab the attention of forward-looking cities and Google would probably have its choice of cities to pilot such a project. By staying on a city or regional basis, Google might be able to avoid the regulatory reach of some federal and state agencies. As they cross beyond county or state lines, however, the regulatory environment would become more complicated.
Assuming the above tack where Google starts local, policy makers would have many things to ponder over such an application including:
- How to create an open network, such that vehicles from multiple operators can traverse the same roadway and still communicate in such a way that all operate in a seamless fashion, regardless of the underlying transport technology?[added 08/03/18 – RideOS seems to have the type of transportation orchestration platform envisioned in this bullet and is starting to get traction.]
- Who controls the signaling system and should that entity be a private operator (e.g. Google), a quasi-private entity or a municipal entity?
- Should the entity that controls the signaling system be able to prioritize traffic, based on subscription tier, (e.g. public safety vehicles would still get first priority), etc.?
- How to charge operators for the use of the roadways (e.g. pay per mile) and pay for ongoing infrastructure upgrades as well as upgrades that such a transportation system would entail?
- What are the privacy implications of knowing a person’s movement at such a micro-level?
A shared vehicle society has long-term implications for local planning officials and could change how they plan for parking, design roads and the economic implications of the hollowing out of the traditional automobile trade.
- The Gee-Mobile service could potentially reduce the number of parking spots at a given building. Garages in single family residences might no longer be necessary in the autonomous auto world. At night or other times of slack demand, the Gee-Auto would park itself in unoccupied locations, which wouldn’t have to be near a residence. Further, because a Gee-Auto is dynamically assigned, parking lots could be structured to eliminate the space between cars (Last-In, First-Out). Additionally, parking lots and charging stations could be located in what are currently unusable spaces (e.g. in a median).
- The roads could be optimized for the autonomous vehicle. For instance, because it would be possible to create a narrower vehicle (LIT Motors, as an example), as well as pack the vehicles closer together, it might be possible to effectively create, say 3 lanes where there are two. These high density lanes could dispense with painted lines, as electronics would keep the autonomous vehicles in place. These virtual lanes would only be for the higher speed, autonomous traffic and not traditional motorists [Added 8/25/2014 – similarly the number of lanes for a given direction could be dynamically assigned, depending upon time of day – e.g. a 4 lane road might use 3 lanes for one direction in the morning and change the direction of those lanes when the traffic pattern changes in the afternoon).
- The local economic impact of the reduction of traditional automobiles will be huge. Of course, gasoline taxes to pay for infrastructure go away (an issue with electric cars that need to be addressed, regardless). The bigger impact might be on the restructuring of local economies. From the local auto shop to the gas station to the car dealer to the insurance agent, the traditional automobile has a huge economic impact on a community and the lost revenue would have to be made up with new opportunities from existing and new employers.
One aspect that a local economic agency could tout when trying to get those new jobs is the superior quality of life (e.g. not having to fight traffic, lower cost of transport, freedom for senior citizens and those with physical disabilities to leave their domicile without depending upon others, etc.) Like with gigabit broadband there will be a first-mover advantage for those communities that successfully implement an autonomous vehicle network. Eventually, however, being a “Smart Transport Community” will become a must-have.
Although all the technological elements of the so-called Gee-Mobile service exist today and the pricing is even within striking range, the bigger barriers will probably be business model and regulatory. It looks like there is a path to a business model (particularly as autonomous vehicle costs fall). Google has proven that it can work with local governments with its Google Fiber initiative, which would be helpful in getting past regulatory concerns. The biggest question in the above story is what jobs will cheapskate Charlie, parsimonious Paula, tightwad Tom and regal Rich will be driven to in 2040?
It’s 8:30 a.m. and, at the same time Rich and Charlie arrive at their office, I am sitting down to work from home in a virtual environment via my 10 terabit connection. Just as I am about to start, I am pleasantly surprised by the appearance of a Gee-Air, the flying drone that whisked my 113 year old mother from her engagement residence (the term retirement home was retired from the vernacular decades before), located some 60 miles away. She had decided to surprise me with freshly made cinnamon rolls for breakfast. But that’s a story for another time.