BYD’s ADAS Liability Bet: The Full Breakdown

When I first read the headline, my honest reaction was: no. This cannot be true.

A carmaker promising to pay, in full, for the crashes its own driver assist system causes? There had to be something hidden behind it. A catch, fine print, a scenario so narrow that it almost never actually pays. So I went looking for the catch. And yes, there are a few, and I will show you exactly where they sit.

But even after I found them, I could not get my head around the bigger picture. This is not one prototype. Not a few hundred test cars. We are talking about scale: a mass-market maker with assisted driving cars already on public roads in the millions.

I have explained in detail, the BYD’s megawatt charger in the past. This time, I spent my time looking at this promise, and the small asterisks around it.

At BYD’s intelligent driving event in late May, chairman Wang Chuanfu announced a one-year liability pledge for City NOA accidents in China, on vehicles equipped with God’s Eye A and B. If the car is judged at fault and the accident is directly linked to the system, BYD says it covers the cost: own vehicle repair, third party property, and personal injury.

The slide behind Wang Chuanfu said the line that matters:

Sit with that for a second. A company producing millions of cars a year, is openly saying that they will take on L3 and L4-style liability while the system is legally still L2 in China. You do not say that out loud unless you are genuinely confident in the system. To make this bet, you have to be.

And here is what is easy to miss reading this from Europe. We picture orderly lanes and fairly predictable traffic. Chinese city driving is not that. I have just been through Beijing, Hefei, Chengdu, Shenzhen and Fuzhou, and the way traffic moves there is a sight to behold. You cannot blink. You are not only driving your own car, you are reading everyone else’s at the same time: who is about to cut in, which scooter is coming down the wrong side, who is stepping out between moving cars. E-bikes appear from nowhere. Streets narrow to nothing. That is the environment BYD is promising to stand behind, and it is what makes this bet extraordinary.

Before going further, the simplest version, and this is my own working understanding. The whole thing sits on one question: who is responsible when the car is driving?

Almost every ADAS system you can buy or subscribe to today is L2, or some form of it. The driver watches the road, and if anything goes wrong, the burden stays with the driver.

Everyone is chasing self-driving. Almost no one will stand behind it. The rule, for now, is mostly the same everywhere: if the system gets it wrong, it is still on you.

That is the backdrop that makes BYD’s move land.

When I said there had to be a catch, here it is. And if your first instinct is that this is negative, I would not necessarily paint it that way. I actually think it is a very good way to go about this precisely because BYD knows the limits. That is why it is not throwing this open to the whole range. It is putting it only on the top tiers, where it trusts the hardware most.

C is the volume tier. Having looked at BYD’s sales mix, my rough estimate is that most BYD cars sit here. So the slice this pledge actually covers, A and B, is a smaller piece of the pie today.

But hold two things in mind. First, BYD already says it has more than 3 million cars with assisted-driving systems on the road across A, B and C. Second, a C-tier buyer is not necessarily locked out when buying a new car. They can step up to B by adding the LiDAR hardware, a 12,000 CNY upgrade, roughly $1,770.

I do not think existing C-tier cars already in the field can simply be retrofitted into B-tier cars. And I also do not yet know how many C-tier buyers will actually pay for the B-tier upgrade on new cars. That matters because the door is open, but we should not assume that it is wide open.

For existing A and B cars, BYD says a software update to version 5.0 will bring eligibility for the pledge. New eligible cars also receive one year of coverage from delivery. This is a controlled way in: start at the top corner, where you trust the hardware most, and widen from there if the numbers work.

And before you think I am selling this as flawless, I am not. This is where the catch I promised actually lives, and it is bigger than one phrase.

The conditions are specific, and they lean BYD’s way. The driver has to use the system properly and keep monitoring it. This is still legally an L2 system, so the driver is still expected to supervise. The real grey zone is takeover. The terms require the driver to retake control in time once the system prompts a handover, but the public Q&A does not clearly define how much time “in time” actually means. A few seconds? Longer? Situation-dependent? That is exactly where disputes will sit.

There is also the question of who judges the technical cause. From what I have seen, BYD’s authorized service center makes that technical assessment. So the company writing the cheque is also heavily involved in judging whether the cheque should be written.

I will be honest about my own read, because it changed as I dug. My first instinct was that this was a masterstroke, and at the level of scale and ambition, I still think it is. But the deeper I went into the conditions, the more careful I became.

If you crash, you are still the one who calls the police, gathers the evidence, and reports the case. BYD also requires the owner to contact the company within 24 hours. BYD advises owners to report to both BYD and their own insurer at first, in case the accident is later ruled outside the guarantee. Only once BYD accepts the case does the owner withdraw the insurance claim. So the “no premium hike” benefit only fully lands if BYD says yes.

And this was exactly my own misunderstanding at first. I initially thought there would be no need to involve the insurance company at all, because BYD was supposed to take over all crashes related to its ADAS. That is not quite how it works.

So yes, this is bold. But it is not open ended.

Is BYD a company that avoids risk? Not really. And honestly, even from where I sit, this one leaves me a little beyond words.

Because this is not the first time. Last year, BYD did something similar for parking. I did not know it until I started researching this topic. Its smart parking feature lets the car park itself, and BYD pledged to cover losses if that system caused an accident.

What I found really interesting is that the parking pledge was not limited to A and B. It went across the entire lineup, including C. Again, no cap on liability for damages caused by the system.

Then came BYD’s own number. BYD says smart parking usage jumped from 21% to 93% of owners once the pledge was in place. I cannot independently prove that figure. It is BYD’s own data. And to be transparent, I do not know how many cases were reported or compensated by BYD. But if that number is even close to true, it is incredible. Same hardware, same feature, more than four times the usage.

Think about what that tells you. Those owners were not avoiding the feature because it did not work. They were avoiding it because they did not want to be the one paying if the car clipped a wall or another car in a tight space. The moment BYD said it would carry that risk, the fear went away.

The barrier was never only the technology. It was trust. And liability did not just remove the barrier, it changed how people behaved.

Go back to God’s Eye A and B. Yes, it is a smaller, more premium slice. But every one of those cars is out on the road collecting exactly the data BYD needs, in exactly the conditions that matter.

BYD says it already has more than 3 million cars with assisted-driving systems across A, B and C. I do not know exactly how many are A and B, and honestly, that is the number I would watch.

So let me use a deliberately modest example. Since BYD’s relevant volume is in China, let’s say it sells around 3 million cars a year there. If just 10% are in the relevant A and B tiers, that is still around 300,000 new cars a year. If each of those cars drives roughly 3,000 km a year using City NOA, that alone is around 0.9 billion km of real world city driving exposure.

And this is not clean test track data. This is city traffic. Pedestrians. E-bikes. Narrow roads. Bad cut-ins. People doing things you cannot predict from a dashboard.

That is a learning curve that compounds. At that volume, a system caused crash slowly stops looking like a blind gamble. It becomes something you can estimate, model and price.

That is why I think this is more than marketing. The pledge builds trust with the customer, yes. More people are likely to use it because of the pledge or because of building trust in this technology, and that usage will compound. But behind it, BYD is also building the learning loop that could make the pledge easier to sustain over time and give it an incredible edge over competitors.

The cars collect the data. The data improves the system. The improved system reduces the risk. The lower risk makes the liability promise more believable. And the promise gives customers more confidence to use the system.

That is the flywheel.

I am not going to narrow this to Europe. The whole world needs to see what is happening here.

Tesla is doing a lot in this domain, and maybe in an year or two, we get genuinely reliable full self driving across Europe and the US. But on the other side, you have a huge Chinese competitor effectively telling its customers: drive, I am behind you. Of course, with asterisks.

That is the China way. That is China speed. And in a few years, BYD will surely be among the companies to watch closely in ADAS. This will also push other Chinese OEMs such as Nio, Xiaomi, Li Auto and XPeng to raise their game, which means a net win for consumers and extreme competition for German, US, Japanese, Korean and French automotive OEMs trying to keep up in China, and eventually on their own turf as well.

Because the competition for ADAS may no longer be only about who has the best features but something much harder: who is willing to stand behind the system when it makes a mistake?

BYD’s pledge is not full self driving. It is not even a clean L3 liability model. But it is one of the clearest signals I have seen that the ADAS race is moving into a new phase.

Not just capability. Responsibility.

And that may be the part that changes customer behavior first.

1 2 3 4 5 6 7 8 9

Also kindly share your feedback, if you liked this Newsletter. This will help me to improve.

Thanks and till next week,
Haseeb