If the chips are sufficient, then why are they failing?
Because the package they're put in is failing.
It's much harder to build test equipment than a basic product. Incredibly more difficult and expensive.
Yes you can have dodgy parts, but ICs have pretty good QC as each part is approved and sometimes binned. They themselves are a product, and will rarely have limited batch issues because of good testing, flying probes and continuity etc. Performance measured at data sheet specced levels. Certificates of conformance attached with shipped goods.
What matters is that you can adequately test the full assembly to pick up on dodgy components or processes before shipping to end users. Chip companies can and must afford this. You can't stop everything but the goal is to try, and if more than 10% are defected then it is a serious issue and a failing of quality assurance. More control is needed. But at least you need to filter them.
Much easier when said than done.
S&B are receiving tested parts but they can't trust their entire built out device because of that.
If you ship a faulty product, it's not at the fault of or much concern to your supplier.
If it was just the chip it should be detected in product testing and nothing should pass.
If you have to do a recall on everything, because of functional issues/concern or because of a mandated recall from an audit or official direction, then something crucial was missed in R&D, then not flagged in testing and product testing needs to be fixed so that it will catch small issues, let alone the entire fleet being decided faulty after shipping internationally at scale.
The thought crossed my mind today, with the pace of resupplying, and the amount of errors, it seems like they either have no idea or are milking out a dodgy product due to economic pressure, and patch fixing their dodgy untested product in software that disables when certain related metrics are detected.
If you can do that, it's not because the chips are bad. It's because you've built a lemon and backed yourself into a corner with the complexity of the design and dynamic of business.
It's not a new phenomena and won't be the last time.
To see it from the biggest name in this industry is why I have the confidence to make my earlier post.
Once a product becomes this complicated, and is pushing power levels to the extreme and packing in hundreds of parts, the testing also has to match. If it did, then this basic product would work statistically better than a Grasshopper. Kind of similar limitations with these higher wattage compact heaters.
If you've ever done any overclocking, thermal read outs for chips are estimations compared to the actual silicon.
Mostly the hottest thing in a device is the chip, if you add a powerful heater in close proximity that is obviously going to get stressed in the field, you also need to test it in accordance.
Turning around after a product launch to recall everything and then not being transparent about the story does not lead me to think that it's anyone else's fault other than the company launching the product.
Whatever the case, hopefully they can sort it out for the end users sake. They owe them as much.
It works amazingly...the airflow alone is worth to endure all the back shit that happened with this unit. I love it and it is a perfect...'lets go out'....vape. 
is always available