Observers of the autonomous vehicle (AV) industry could be forgiven for thinking that the technology has hit something of a wall. Tesla disabled its Autopilot feature last month; Waymo’s AVs get into near-weekly collisions in its Phoenix testing area; Uber Freight halted its autonomous truck program. While important advances in fields like computer vision have been made by these technology companies, it’s still incredibly hard for AVs to achieve the situational awareness—recognizing hand signals, managing four way stops, making ‘eye contact’ with other drivers—that is instinctive to human drivers. There’s even a startup called Perceptive Automata devoted to helping AVs intuit human driver behavior by imagining how people think.
Moody’s (NYSE: MCO) released a white paper on Monday, “Autonomous driving efforts pick up pace; GM ahead with Honda investment in Cruise,” that suggests AV startups have inherent disadvantages compared to incumbent vehicle manufacturers. The white paper emphasizes that winning in the AV space will be predicated on an intricate orchestration of a whole suite of still-developing technologies including surrounding environment sensors, computing, external connectivity, and standard vehicle controls. Building a business that depends on bringing new technologies together is capital-intensive and risky.
The deal that Moody’s referred to in its title is a joint-venture between Honda (NYSE: HMC) and GM (NYSE: GM) on GM’s Cruise Automaton subsidiary. In May, SoftBank (OTC: SFTBY) invested $2.25B in Cruise; now Honda has said that it will immediately invest $750M in Cruise, and contribute another $2B over twelve years in fees and development costs.
“The deal between Honda and General Motors is consistent with how we expect risk- haring arrangements among automakers for developing technologies to take shape,” Moody’s wrote. “The industry's reliance on cooperative arrangements is credit positive because it will accelerate the introduction of effective AV technology and significantly reduce the business risk and capital required by each automaker.”
Moody’s expects “radical evolution of the autonomous driving landscape over the next five years”, citing changes in technology, new regulations at every government level, a wide array of consumer demands, and emerging competitors among automakers, parts suppliers, and tech companies. Managing those risks, in Moody’s view, requires large-scale cooperation and disciplined capital allocation to bring autonomous technologies to market safely and at scale.
“The capital requirements to support investments in each of these critical technologies would be prohibitively burdensome for many individual companies given the business and financial risks associated with core automotive operations,” Moody’s wrote.
It’s fascinating to consider the relative advantages and disadvantages of incumbents and upstart competitors in the terms laid out in Clayton Christensen’s The Innovator’s Dilemma. That book says that technology product development follows an S-curve, where improvements based on early iterations are minimal, then in the middle of the product cycle the improvements are steeper, but eventually improvements level off and further releases don’t substantially change the product. Where exactly on the ’S-curve’ is autonomous vehicle technology now? Is the technology rapidly improving or is it still in the early stage of marginal improvement, with no clear winner?
More importantly, there are crucial differences between, say, mobile phone technology and autonomous vehicles. Mobile phones can be introduced into the marketplace as a minimum viable product, and the steep section of the S-curve, where transformative improvements are introduced in successive releases, can take place in the public market in front of consumers.
We think that autonomous vehicles will have to be nearly perfected before widespread adoption takes place due to safety concerns. Even the trained contractors testing autonomous vehicles today have trouble maintaining the appropriate amount of attention (both visual and manual) while riding in Level 4 and Level 5 vehicles. It’s unrealistic to expect consumers to operate autonomous or semi-autonomous vehicles correctly because drivers today do not operate their cars ‘by the book’, whether the issue is regular maintenance or driving safely. Therefore, when truly autonomous vehicles are ready to be released, the product will be in the last flat stages of Christensen’s S-curve, which marks a mature product where further improvements will be marginal.
Our view meshes with Moody’s: because AVs will be released at an unusually late point in the product development cycle, risk-sharing joint ventures between incumbents make sense, and disruptive startups will burn through mountains of capital building products that may never make it to consumers.
Still, just because GM has taken a leading position in the space does not mean that it will be able to maintain it.
“A leading position at the start of this marathon does not assure leadership down the road. Nevertheless, GM has one of the most well-thought-out strategies, it is executing on this strategy, and is well positioned for the long haul,” Moody’s concluded.