Gravity Software Past, Present, and Future Interview w/ Emad Dlala

[…]
"we had to use hypervisors to allow for isolation of the systems, which is somewhat a new technology."
Fact checking statements is great when people try to deceive, mislead, or misstate material facts. I wish more people would fact-check. In this case, Diala was honest and forthright, so how about we give him a break on the exact wording he used during a live interview?

In any case, the substance of what Diala said is true. Hypervisor technology in cars is somewhat new, even though the first hypervisors used were in the 1960’s.

The Lucid Air prototype came out in 2016, production in 2021. The Land Rover Defender announced QNX hypervisor in 2020. I hope I have my facts straight!

That qualifies as “somewhat a new technology” in my book.
 
I took it to mean, it was “new” at the time they were designing Air. They did start about 15 years ago or so designing it. Perhaps it was new in use to autos back then? Just speculating…
But they didn't and don't use a hypervisor in Air - Air, as he said, has two separate compute units so there's no need. It's only in Gravity with the one Qualcomm chip that they're using a hypervisor to have both QNX for safety critical systems and Android for what I assume in infortainment/UI.

QNX introduced its hypervisor in 2015, so over a decade ago. In the high-tech world of software, that's not new, or even somewhat new.

how about we give him a break on the exact wording he used during a live interview?
OK, QNX did recently upgrade their hypervisor support with their version 8.0 - but that was QNX optimizing Type 2 hypervisors to have Type 1 efficiencies. This was in May 2025, so I doubt Gravity could be using it. My educated guess is that Gravity is using QNX hypervisor 2.2, which is part of QNX SDP 7.1. That's been available since 2022, so the timing seems about right. It sounded to me like he was making an excuse for maybe Lucid's developers not understanding QNX hypervisors.

But, hey, like I said, I'm probably just grouchy from still not yet getting the OTA for 3.4.1.
 
...

But, hey, like I said, I'm probably just grouchy from still not yet getting the OTA for 3.4.1.
Can't remember if you turned off wifi in your car. That seemed to precede the update for a number of us.
 
Can't remember if you turned off wifi in your car. That seemed to precede the update for a number of us.
I did earlier in the day before I got the update, although I'm not sure it mattered because a whole wave happened to come out later than night that I was part of.

I realized I needed to turn wifi back on because the car's cell signal now seems non-existent in my garage since the update. I couldn't do remote climate anymore, and it wouldn't get any audio services unless I left the garage and got into the clear sky. The update seems to have cut the car's cell signal in half or something (??).
 
Can't remember if you turned off wifi in your car. That seemed to precede the update for a number of us.
I have not turned off WiFi, but I did park the car for several hours away from any WiFi.

When I turn WiFi off at home, the app says “Vehicle Offline”
 
Last edited:
But they didn't and don't use a hypervisor in Air - Air, as he said, has two separate compute units so there's no need.
They did. It was a bad idea in Air for many reasons, so they don't anymore. v1 used hypervisors. v2 does not.
 
They did. It was a bad idea in Air for many reasons, so they don't anymore. v1 used hypervisors. v2 does not.
Emad Dlala: "We use for both Air and Gravity a QNX-based system for the safety for the critical, and Android. But the difference in Gravity is that right now they are implemented in one chip, where we have to use hypervisors to allow for the isolation of the systems, which is somewhat of a new technology. It comes with rewards and challenges."

So my bad, again, for believing Dlala, I guess.
 
Emad Dlala: "We use for both Air and Gravity a QNX-based system for the safety for the critical, and Android. But the difference in Gravity is that right now they are implemented in one chip, where we have to use hypervisors to allow for the isolation of the systems, which is somewhat of a new technology. It comes with rewards and challenges."

So my bad, again, for believing Dlala, I guess.
Guess you'd rather he didn't speak about the update at all then. Seriously, are you going to haul the guy over the coals for one thing he said in an unscripted interview. You're making a mountain out of a mole hill.
 
Guess you'd rather he didn't speak about the update at all then. Seriously, are you going to haul the guy over the coals for one thing he said in an unscripted interview. You're making a mountain out of a mole hill.
At this point, I'm trying to factor in @borski 's statement that Air used to use hypervisors, then changed because it was "a bad idea in Air for many reasons," and so removed it (which seems difficult), only to put in back into Gravity and have the VP claim it was part of the ongoing software problems because it was new. QNX typically provides excellent support to their OEMs and Tier1s, so if Lucid had a problem with QNX's hypervisor, I'd expect that Lucid would have QNX all over it.

Maybe 3.4.1 will be so great that I won't care, but right now, considering the state of Gravity software, I do care.
 
At this point, I'm trying to factor in @borski 's statement that Air used to use hypervisors, then changed because it was "a bad idea in Air for many reasons," and so removed it (which seems difficult), only to put in back into Gravity and have the VP claim it was part of the ongoing software problems because it was new. QNX typically provides excellent support to their OEMs and Tier1s, so if Lucid had a problem with QNX's hypervisor, I'd expect that Lucid would have QNX all over it.

Maybe 3.4.1 will be so great that I won't care, but right now, considering the state of Gravity software, I do care.
Okay. You don’t have to believe me. I’m perfectly fine with that.

He definitely couldn’t have simply misspoken. That would be crazy.

(There are plenty of very old threads about this)
 
Last edited:
Okay. You don’t have to believe me. I’m perfectly fine with that.

He definitely couldn’t have simply misspoken. That would be crazy.

(There are plenty of very old threads about this)
I did a search for hypervisor in the forums and didn't find anything relevant. Not that these forums are the indisputable truth anyway.

I've just got the 3.4.1 update, as well as the latest iOS update (2.13.0 (4028). I remain very disappointed in Lucid Software, just from the two things I've tried not actually driving the car, lol.
 
Emad Dlala: "We use for both Air and Gravity a QNX-based system for the safety for the critical, and Android. But the difference in Gravity is that right now they are implemented in one chip, where we have to use hypervisors to allow for the isolation of the systems, which is somewhat of a new technology. It comes with rewards and challenges."

So my bad, again, for believing Dlala, I guess.
I bet you love Elon Musk lies.....give him a break, maybe he mispoke, no big deal. Making a mountain out of a molehill. Now lets all get on with enjoying our Lucid vehicles!
 
I bet you love Elon Musk lies.....give him a break, maybe he mispoke, no big deal. Making a mountain out of a molehill. Now lets all get on with enjoying our Lucid vehicles!
No, I despise Musk. We terminated our Model X lease early due to unhappiness with the car (me) and unhappiness with supporting Musk (wife).

But,thanks for projecting my disappointments as some kind of Tesla-love thing.
 
I did a search for hypervisor in the forums and didn't find anything relevant. Not that these forums are the indisputable truth anyway.

I've just got the 3.4.1 update, as well as the latest iOS update (2.13.0 (4028). I remain very disappointed in Lucid Software, just from the two things I've tried not actually driving the car, lol.
try 'virtual machine' or 'vm'
 
try 'virtual machine' or 'vm'
OK, I find your previous post on the subject.

Now we have confirmation that Dlala mis-spoke:
• Hypervisors aren't new
• Hypervisors aren't new to Lucid
• Air did not "not need it" because it has two separate compute modules.
• If you're correct that Lucid rewrote stuff in Air to not use a virtual machine (which is what the hypervisor runs), then did they port a bunch of code from one Operating System to another OS, and if so, which OSes were involved?

I'm OK dropping this whole thing - to me it's just further indication of how much further Lucid needs to go, and I hope the company has/gets some real software expertise.

For instance, yesterday my wife was out and about and called me. She sounded great, and I told her so! Unfortunately, the car didn't reconnect her phone, so she wasn't using the mic/speakers in the car, but calling me directly from her phone.

Lucid's software engineers had better be burning the midnight oil.
 
then did they port a bunch of code from one Operating System to another OS, and if so, which OSes were involved?
I don't know the answer to this question, but I do know they essentially rewrote the stack. It took a long time, and there was some parallel work for a bit of time, but it was sorely necessary to improve the state of things.
 
I finally had time to watch the video. On the one hand, a lot of what Emad said was great, and it's also good to hear that there's another stability focused release coming 6-8 weeks later. But judging from the comments here showing that there are still a fair number of glitches after 3.4.1, I feel like Emad was too definitive about the software stability problem being resolved. Hopefully 6-8 weeks is enough time for another really substantial set of bug fixes in the next release. I decided in December that the level of instability was too high for my personal circumstances, and so canceled my order. While things seems greatly improved, I also still feel that they need at least one more big set of fixes before I'd be ready to buy, and 2 or 3 substantial OTA updates is more likely.

On the good news/bad news front: it's great they were able to accomplish all of 3.4.1's fixes with 2 months of engineering, but it's a bit shocking that it took them 5 months from the start of real customer deliveries for them to begin really digging into the bugs. Nevertheless, 3.4.1's level of improvement after two months of work is hopefully a positive sign of what the next release will improve after 6-8 weeks more effort.
 
Last edited:
I feel like Emad was too definitive about the software stability problem being resolved.
I thought the same. Things are way more stable now, but I still get the same errors (including cameras' responsiveness), just less frequently. He probably misspoke, but he shouldn't had been so definitive in his statement.
 
Back
Top