This week, the chip shortage crisis escalates to take an US president executive order to help auto industry. Easy to say a crisis can come with opportunities, difficult to really challenge the status quote to make a real difference for a better future. For example as Nikkei’s Tesla Model 3 teardown suggests: it’s years ahead of other cars. e.g. Model 3 uses a few Electronic Control Units/chips, where as other cars have dozens. At the time like this, I hope more car makers to switch to new architectures earlier rather later. Bonus: it’ll help to catch up the competition & also to be better future-poof too. So the crisis is not wasted 🤞. Just in time as an Chinese poetry:
不經一番寒徹骨，怎得梅花撲鼻香。
No bone-freezing pain, no plum-blossom gain.
For 2 most exciting areas making cars smarter: In-Vehicle-Infotainment, & Self-Driving under the hook, what if they somehow converge. Which further simplifies the HW/reduce chips & powers more innovation by SW? Let’s consider it from two ends: 1) what are the easy paths forward & 2) what it could be at a better end.
Evolution Paths Forward
In the short term, it’s easier to assume tomorrow is the same as today. Not just it’s easier for human brains to compute linearity, but also likely right most of time for the imminent future. Even so, the industry can still push the boundary to be 10% better each year. If you still do 3+ years product development cycle, go back to the drawing board to speed it up first.
Infotainment systems
Better embrace & leverage smartphone tech & ecosystems as much as possible. Because automotive is a niche vertical in term of sales, 5% (80M vs 1.5B) of smartphone each year. For example as Bloomberg pointed out: By itself, only making chips for cars is not a sustainable business. So if you can not complete with smartphone for priority on chips: capital-intensive & software: talent-intensive, you might just join them 😉.
Still, anyone may reinvent the wheel for 10x better. I do wish them luck & hope lessons from the history help, e.g. Intel’s smartphone & smartwatch “experiments” 🧪. There’ll be a day when “the time is right & planets are alined”. But if you don’t have such money to burn, nor enough time to build, you may tap into established platforms & ecosystems instead. Besides, it’s more valuable to solve more unique problems for more users than building another “me-too” platforms/technologies. Even you build, they might not come especially when the current alternatives & substitutes are good enough. Also in a digital economy, only top 1 or 2 matter because the marginal cost of scaling is almost zero for them. Hopefully together, the industry may have 3 open ecosystems to bring more new smiles to more users sooner 😀.
Self-Driving
I know even less about Self-Driving. So it’s just my regular-Joy observations. Self-Driving reached the Peak of Inflated Expectations in 2015 & dropped out from the Hype Cycle for Emerging Technologies in 2020. While it’s wise to take them with a grain of salt, it does signal a long road ahead. As wise VC friends suggested: behind the hypes, there are continuous drum beats to production. Which will move the world forward in some niche markets earlier than the consumer market.
But for consumer, I bet Tesla’ll continue to lead the way because of its “invincible speed”, powered by its vertical integration strategy. Whereas the platform game is a wild west. Many promising players, e.g. WAYMO, CRUISE & etc., but still everyone’s game. Two thing for sure: car makers seldom play the platform ace card right. And it takes a village to make a platform & ecosystems prosper.
A Revolution Way
In a longer run, everything goes with cycles. And there will be a few moments in cycles, it turns instead of stay on the same linear trend. That’s the time to think disruptively if to be a giant slayer or a giant to be slayed 😉.
So what if both infotainment and Self-Driving systems run on the same computer unit? So one less board & chips to worry about. Also with 2 worlds bridged by SW, new opportunities for SW-defined innovations may come without any HW cost & lead-time. It’s not even a new idea & many have tried. A few challenges are how to achieve the performance & safely without overwhelmed by the complexity.
OS Virtualization was ahead of its time for smartphones since 2011 & may still overkill for most mobile users for a while. But it could be a good solution for auto, especially with better chip performance & matured hardware-assisted virtualization technology. So, Self-Driving & Infotainment systems can run on their own virtual machines & share the same hardware. Which brings at least 3 advantages:
- Lower the total HW cost for multiple systems. It’s no brainer as Apple M1 chip saves billions for 15+M MacBooks. The math may not work as the top 2: Volkswagen & Toyota ship about 10M cars each year. Not surprise some may call Tesla craze to design their own chip for HW3.0. But you gotta to admire the vision & courage 👍.
- Better isolation for Self-Driving & Infotainment systems & less painful to restart each virtual machine/system. Especially when 100% bug free is not realistic, better architectures alway pay, only sooner or later.
- Easier to update each system independently & reuse your Information & Self-Driving SW investment across car models. So, you may serve/reach more cars/users. These are new goldmines for application & service innovations. Car makers still have sometime to lead before too late.
I bet this may be too early for Tesla HW4.0. But I’m happy to be surprised, in Q3 2021? Or someone to “retrofit” an M1 machine into a car for a prototype, maybe 😉?
Full Disclosure
The opinions stated here are my own, not those of my company. They are mostly extrapolations from the news. I don’t have insider knowledge of those companies, nor an EV expert.
