The two markets for mobility startups: today’s carmakers OR tomorrow’s robotic ride-sharing companies

Why mobility startups must to choose early on between 1) selling features to carmakers or 2) technology to robotic ride-sharing companies.

5 min readMay 31, 2017


Mobility and autonomous startups are exploding on the frontier-tech startup scene. Aspiring engineers are scrambling for the requisite training to join the ranks of Zoox, Waymo (recently assumed to be valued at $70B), Argo, Cruise, and the plethora of startups that are solving problems related to sensing, perception, planning, vehicle-to-vehicle communications, security, and in-car services. Unfortunately many startups don’t appreciate the distinction between selling into the automotive supply chain, vs. selling into the nascent robo-taxi service business. I’ve blogged about the harsh reality of selling into the automotive supply chain in the past.

To make it simple for founders, I’ve split the market opportunity for startups into two categories: autonomous as a feature, and autonomous as a service, the former being the traditional auto supply chain, and the latter being the Wild West. I encourage founders to gain an intimate understanding of the dynamics of these markets before choosing one, knowing that the business they build to pursue one may not be appropriate for the other.

Many talented founders are building magical technology and services to accelerate driverless transport to the masses. Unforutnately, some founders are attempting to sell products into BOTH 1) the conventional automotive supply chain i.e., GM and Toyota, as well as 2) nascent robo-ride-sharing companies, such as Uber, Zoox, and Lyft/Waymo. Selling into the automotive supply has been traditionally challenging for startups; it took Mobileye many years! Meanwhile, robo-taxis represent a nascent market, and are very vertically-integrated, similar to car manufacturers in their early days (and what Tesla is aiming to become). My advice to founders: Either focus on the car companies, OR the robo-taxi companies, but not both.


Autonomy features are just an extension of the existing automobile technology. It started with cruise control in the middle of the twentieth century. In the seventies, anti-lock braking and traction control made it safer for humans to pilot their cars in poor weather and emergency conditions. In the past couple decades, we’ve started to see adaptive cruise control and adaptive parking take more work out of the hands of the driver. Automatic parallel parking, lane keeping, and Tesla’s Autopilot put driving into the hands of the computer, under drivers’ close supervision.

Car companies have made it clear that they want their cars to do more of the driving, evidenced by Ford promoting it’s driverless tech chief to run the entire company. The big brands: Mercedes, BMW, GM, Toyota, and Volkswagen are on the prowl for the hardware and software that will bring more automation to their cars. Intel hurried to purchase MobilEye to position itself as a vendor into this supply chain. Unlike startups, Intel has a massive balance sheet to support the arduous qualification and validation process imposed upon technology vendors by the traditional automotive supply chain. After all, cars are among the most complex, and most heavily regulated products in one of the most cut-throat markets in existence.

Startups are building lidar and radar sensors, perception algorithms, maps, simulation tools, security, and vehicle-to-vehicle communications software to accelerate autonomy. However, most companies I speak to feel as though the automotive supply chain will give them special treatment relative to other vendors, whom are subject to the arduous validation process I alluded to above. Perhaps some innovative startups will be able to command high prices and short sales cycles; however, it’s no secret that selling into the automotive supply chain is a long, arduous, low-margin business. Startups will lose margin to integrators higher up in the automotive supply chain with the wherewithal to validate and integrate the novel technologies startups are inventing into cars.

For example, many perception, visualization, and planning companies expect to license their algorithms for hundreds of dollars per month to car companies and Tier 1s (e.g., Conti, Bosch, Denso, and Delphi). If the software is expected to go on a Toyota Camry, that itself commands a $300/month payment, how can they expect manufacturers to pay hundreds of dollars per month for autonomy software, and pass that on to the consumer at a markup? Tech startups need to be mindful that if they are selling into the auto supply chain, they won’t be regarded any differently from other software/hardware vendors, and should expect the same, low margins.

At Lux, we invested in a stealth startup building special hardware AND software as a complete driverless solution for OEMs. Its team and culture are being built from the get-go to manage the difficult qualification and validation it needs to undergo before being able to sell a single product. However, the vertical nature of the product lends itself to being higher-up in the supply chain, which enable it to capture the attractive margins that we expect to underpin an iconic company.

Expect Autopilot-style driver assistance, and perhaps even automated driving, to become an option on window stickers in the next decade. The technology vendors for those features are likely to be the usual suspects in today’s Tier’ed automotive supply chain. Startups offering truly unique technology, that have the time, money, and expertise to undergo rigorous validation and qualification, have a shot at selling into the conventional automotive supply chain.


This is a nascent business, and should be regarded as a market that probably won’t be significant for at least another 4 years. This market is also very concentrated: Uber and Lyft, the latter probably with the help of Waymo, are the only U.S. companies with the distribution to offer such a service to consumers. Meanwhile, startups like Zoox also aspire to offer robo-taxi services in the future. Like Henry Ford at the dawn of the Ford Motor Company, these few players are building most of what they need in-house: algorithms for perception and planning, simulation tools to build and test those algorithms, vast labeled datasets, maps, and vehicle-to-vehicle communications. Most are also building their own sensing hardware, as evidenced by the recent lawsuit filed by Waymo against Uber for allegedly stealing their lidar technology.

This tight integration is also reminiscent of telecommminications until 30 years ago: when AT&T invented and owned everything from the copper in the ground to the telephones in our living rooms. Though I do not expect autonomous taxi companies to become regulated monopolies like the telecoms of yore; I do expect the level of complexity, and nascency of the technology to follow a path of more vertical integration for the first few decades of robo-taxi existence. As a counterpoint, there could be a possible WinTel (alluding to the Microsoft Windows and Intel Microprocessor of the PC era) combination that creates a vibrant ecosystem of vendors selling hardware and software to robo-taxi operators. I guess the next question is: what happens when you have a blue screen?

The robo-taxi business is nascent and fertile ground for new business models. Unfortunately the number of companies with the $$ and distribution to offer robo-rides are few: Zoox, Uber/Otto, Lyft/Waymo, and Nutonomy, are a few that come to mind. Furthermore, for the time being, most of these robo-ride companies are building most of what they need to operate in-house: perception, maps, simulation tools, etc. Is there a future where what’s viewed as “core” is outsourced from vendors, and startups with cutting-edge technology? Possibly, but it will take some time.

I expect many iconic companies to emerge as me move towards our autonomous future. I expect many more startups to be acquired for their top talent. At Lux, we seek founders aiming to build lasting, iconic institutions. I expect many iconic institutions to emerge in each category. The stealth Lux investment I alluded to above has invented unique sensors and software that would make adding autonomous as an “option” as easy as adding an upgraded stereo or cruise control for automakers: thereby falling under “autonomy as a feature.” In addition, Lux is invested in Zoox, falling under the “autonomous as a service” category, which plans to launch a fleet of robots that will create unforgettable driverless transport experiences for consumers.

Shahin is a gearhead and Partner at Lux Capital aiming to empower founders seeking to build a fantastic future driven by technology. You can reach him at for follow him on Twitter @farshchi