Actually that's not true either. MobilEye terminated its relationship with Tesla after the deadly Autopilot accident. MobilEye CEO's statement said that Tesla was using its technology in an unapproved, therefore unsafe, manner and MobilEye had no choice but to stop supplying Tesla with hardware that was being purposely used beyond its intended purpose. This strongly implies that Tesla was taking unsafe risks with technology that was never designed to do the things Tesla was attempting to deliver. Out of concern for its own reputation, MobilEye severed its relationship with Tesla. The articles are all over the web. MobilEye was recently bought by Intel for $15B. I'd say they are doing quite well.
This is probably going to derail real quickly but I'll bite:
Honestly I think the truth lies somewhere in the middle. Of course MobileEye would rather upsell the EyeQ4 and beyond systems which are more robust and reduce the risk to MobileEye at the expense of Tesla accepting a slower roadmap than they'd like, and Tesla buying more (expensive) hardware from a vendor.
And as soon as MobileEye saw that Tesla was investigating their own vision system, I think it's only wise of MobileEye to forbid Tesla from co-training their vision system using the EyeQ3 as backup. AI / neural-net cloning is quite easy to do, and a huge legal gray area in terms of how IP laws play into it.
Bottom line, each had their own reasons for doing what they did. And each side is telling a story that makes them sound in the right and the other sound in the wrong. With the way that MobileEye is courting other vendors and pointing to AP1 (and arguably
a lot of Tesla work fine-tuning steering control loops and TACC brake ramping) as their success story, it seemed lose-lose for Tesla and MobileEye to work together any longer.
But hey, we just saw a Nissan Leaf prototype truck-lust, and the 2017 7 Series with MobileEye based lane holding seems to chicken out as soon as the lane lines start curving. I think the rational conclusion is that a lot of credit goes to
both MobileEye and Tesla for how well AP1 performs. And I think both parties will take a short-term setback from breaking off the relationship, but I also believe both parties will end up fine in the long run.