The second quarter of 2020 has been pretty great for Phantom.ai. The Silicon Valley-based ADAS startup closed a $22 million Series A financing in April, led in part by Ford Motor Company. Today, they announced a partnership with Renesas, a Japanese Tier 1 automotive supplier, to develop “full stack Level 2 advanced driver assistance systems.”
This makes Phantom.ai one of the very few startups targeting what would seem to be a lucrative and promising market.
An oddity of the self-driving car revolution is that startups have so far had much more success tackling Level 4 full autonomy, compared to Level 2 advanced driver assistance.
Level 2 means that a driver still needs to be in control of the vehicle, which leaves startups with one of two difficult paths:
- Become a manufacturer and build vehicles for consumers.
- Sell ADAS packages into the existing automotive ecosystem, with lead times approaching a decade.
Faced with that challenge, and perhaps also for safety concerns, most startups have opted to instead work on Level 4 autonomy. This is a much harder challenge, but carries the potential of deploying robotaxis directly and probably (maybe?) avoiding the existing automotive supply chain.
The only company that has cracked this nut, Mobileye, cracked it in a huge way, exiting to Intel in 2017 for $15 billion. On the one hand, I would have thought more entrants would’ve been attracted to this space. On the other hand, it took Mobileye 18 years to achieve this success, highlighting how long the automotive supplier road can be.
That leaves Phantom.ai, which has survived on a mere $5 million seed funding round since 2016, and overcame a cringe-inducing 2018 rear-end collision with a press crew on-board. Kudos to them as the leading startup in the space.
Even Phantom’s own employees seem a little dumbstruck by this state of affairs. One anonymous employee wrote in a 2019 Glassdoor review:
“ If they exist, we don’t know who our competitors are, other than MobilEye. Is another company going to come in and steal our thunder? It’s my biggest worry. [Our competitors, by the way, are not Waymo, Aurora, Cruise, etc… their product is for a different market.]”
The main competitor I can think of is Comma.ai. I own their EON DevKit and have installed it in several different vehicles. The performance of the OpenPilot software it runs is impressive. I wish it would get to market in a bigger way than it has so far.
But cracking the automotive supply chain is tough.