Lucid Investor Day - March 12, 2026

[Need4Sleep]

Agree, there is no way we can go from reading a book to driving within milliseconds (or even seconds.) Anything requiring that quick a reaction time would be eyes-on L2, IMHO.

L3 is indeed complicated, littered with questions about how quickly we are expected to regain situational awareness after the system tells us to take over. I believe L3 on highways is within striking distance, though, with “hey human, start paying attention and be prepared to take control because I’m about to leave this well-controlled highway zone.”


Look at the size of the Uber/Lift/Taxi industry. Waymo has shown L4 is a reality today and is slowly gaining market share. Someone has to supply those vehicles as the entire industry shifts to robo-taxis. From an investor viewpoint, Lucid being one of the key suppliers of L4 vehicles represents a huge growth opportunity. Very attractive for investors if Lucid can pull it off.

For consumers, it’s an indirect benefit. If Lucid can capture a big piece of the L4 market, we’ll stop wondering if Lucid will go out of business. Also, I may not want L4, but if Lucid can deliver L4, some of that technology will apply to consumer-level L3 (which I want for highway driving!)

Same here, I have no interest in L4 autonomy, but I can easily see the learnings applied downstream. I truly hope that Lucid can deliver and learn from Uber and Nuro well

 

[Need4Sleep]

Looks like Car and Driver saw a mockup of Cosmos, though all reporters had their phones locked away.

Earth is expected to be followed a year later.

Kyle Conner also has a pretty detailed description of what he saw.

I noticed some conflicting information between C&D and Kyle Conner's information. Specifically, C&D said the front and rear drive units are identical, while Kyle said front is induction and rear is permanent magnet. Hopefully there will be some clarification on that.

 

[borski]

Looks like Car and Driver saw a mockup of Cosmos, though all reporters had their phones locked away.

Earth is expected to be followed a year later.

Kyle Conner also has a pretty detailed description of what he saw.

I noticed some conflicting information between C&D and Kyle Conner's information. Specifically, C&D said the front and rear drive units are identical, while Kyle said front is induction and rear is permanent magnet. Hopefully there will be some clarification on that.

C&D also said it was on Michelin P Zero tires, which does not exist (they mean Pirelli), so perhaps not the greatest source of legitimate information. Dno.

A few more sentences in and it claims the Air lowers its frameless windows, which… it does not.

 

[Need4Sleep]

C&D also said it was on Michelin P Zero tires, which does not exist (they mean Pirelli), so perhaps not the greatest source of legitimate information. Dno.

A few more sentences in and it claims the Air lowers its frameless windows, which… it does not.

yeah, all these inconsistencies feel like everyone was in a mad scramble to type out whatever they could remember

 

[TahoeMichael]

Thanks for this. It’s definitely worth putting the proper definitions back into the conversation.

To me, Level 3 is the part that never made any sense. And I’m guessing that’s why so many are stuck calling things Level 2++++ and such.

If I don’t need to pay attention at all, there’s no chance I’m going to be able to react in time in an emergency. You’re basically telling people to play video games or read a book, but still be ready to react to an oncoming semi in milliseconds?

Meanwhile, Level 2 is this massive variance of different skills that should have been broken into at least three separate levels.

Level 4 makes perfect sense for Robotaxis. But why Lucid is setting it as a goal for consumers buying a Lucid is strange to me. So I can sit in the back of my car, but only in certain places? What happens when I want to go somewhere just outside the geofence?

Anyway, thanks for the clarification.

Saksak's summary of what the terminology means is a useful primer for those unfamiliar with the way autonomous driving systems are classified. But the levels aren't intended to be precise definitions, and they leave a huge amount of room for different companies to approach autonomy in different ways. Generally, L3 is meant to be about autonomous systems that can give a human reasonable warning that they're going to have to take over. But there's no specific definition for how much warning is required. Most assume that the warning needs to be enough time to allow for a human who isn't paying attention at all to look up and gain awareness of what's happening (i.e. not requiring FSD style attentive supervision), but not so much warning that a human could go to sleep and potentially need a much more extended time to wake, become alert, and take over. So it would be okay to read a book, or play a game on your phone, but not okay to sleep, or drink alcohol, or move to the back seat.

 

[Smorgasbord]

Your are right. We were starting to go from healthy debate to non-constructive territory. Sorry about that.

As you pointed out, there is a lot of confusion on the levels of autonomy. It’s worth getting rid of some of that confusion in this EV forum.

The levels of autonomy are well defined by the Society of Automotive Engineers and is fairly straightforward.

• L1: Partial control. Human partially in control. Car does some of acceleration, brake, lane centering, but not all. Example: Automatic cruise control.
• L2: Eyes-on. Car has full control. Human must monitor constantly and instantly override the car.
• L3: Eyes-off driver. Human does not need to pay attention. However, human must be in the driver’s seat and be able to take control when prompted by the system.
• L4: No driver, constrained. Car is fully autonomous in geofenced areas and under specific conditions. Human is a passenger.
• L5: No driver, everywhere. Car self-drives everywhere and in any environment.

There is huge variation within a category, especially in L2. Lucid Air DDPro, Lucid Gravity DDPro2, Ford BlueCruise,GM SuperCruise, Tesla FSD, etc. vary wildly in capability, yet are all eyes-on L2. This is why we are seeing vaguely defined and informal L2+ L2++ L2+++, and where most of the confusion lies.

Also, cars may have different levels under different conditions. For example, most systems are L2 on highways in good weather, but not in cities or poor weather. Tesla FSD one of the few that is L2 (however many + symbols you want to add to it) in cities. Waymo is L4 but only in select cities and conditions. The San Francisco blackout is where Waymo spectacularly failed. If Waymo was L3 with a remote human driver ready to take control, the human would have taken over when requested by the system and it wouldn’t have made the news.

Many companies are trying for eyes-off L3 on highway, where the driver is effectively a passenger and can watch videos. Those system will probably be L2 in the city. We’ll then be looking at cars with L2/L3 designation.

I hope this helps in getting to a common understanding and “language” as we talk about self-driving for Lucid.

The SAE has published a summary chart: https://brx-content.fullsight.org/s...t/news/blog/sae-j3016-visual-chart_5.3.21.pdf

Note that one thing missing from the chart and many summaries is that the "drive everywhere in all conditions" phrase is actually caveated to add "in which a skilled driver can safely drive." Which leaves that up to interpretation and therefore varying future claims of adherence.

FWIW, the SAE doesn't have an "eyes-off"/"eyes-on" differentiation, just when supervision is/isn't required. SAE doesn't specify driver monitoring, either. Oh, and Waymos in the SF Bay Area and parts of LA are L4 on freeways now.

As @Saksak points out, in practical terms, Level 2 is very broad - too broad IMO. And misses consumer desirable features like "point to point navigation." Now, maybe I read/heard it wrong, but it seemed to me that Lucid is trying to have a "point to point within a city" feature, which confuses things even more, IMO. The whole original idea behind "point to point," and what everyone else means by it, is that it's not restricted to cities or to highways.

Also, Level 3 with the narrow ODDs (Operational Design Domains) that Mercedes and BMW had turns out to not be very useful, as Mercedes essentially admits by dropping it from future vehicles. An L3 that disengages when traffic speeds up past 40mph or there aren't enough cars around you or that it detects the sound of rain (puddles?) is almost more annoying.

With the focus on robotaxis, I suspect we're not going to see any new L3 systems for a while. The need to notify a human in the driver's seat to take over within several seconds (not defined in the SAE spec, but given as an example, and no-where is "milliseconds of notice" indicated as OK) is difficult to properly implement, as knowing when conditions are going to change several seconds in advance is difficult. The transition is ripe for failure on both machine and human, and assigning blame via the courts is, well, not optimal.

 

[The_Don]

Not sure if anybody else caught this, but seems like even the initial midsize vehicles will not have the hardware necessary for L3/L4 capabilities. They'll be updating that in 2028. See the question starting at 3:29:45 in their investor day video below.

 

[The_Don]

Also they clearly stated Gravity hardware will be coming to the Air, and the way they said it made it seem like that would happen soon. They were a bit more vague on midsize hardware coming to Gravity/Air, likely because that won't even be finalized until 2028, as noted in my post above.

 

[TahoeMichael]

Not sure if anybody else caught this, but seems like even the initial midsize vehicles will not have the hardware necessary for L3/L4 capabilities. They'll be updating that in 2028. See the question starting at 3:29:45 in their investor day video below.

I'd take even that with a grain of salt. Until Lucid has a functioning L3/L4 system, they won't be entirely sure what hardware is necessary. Up until that time, it's a science project. We should learn the lesson of Tesla getting that wrong year after year, and Lucid getting it wrong with the Air, and not get our expectations up.

 

[The_Don]

I'd take even that with a grain of salt. Until Lucid has a functioning L3/L4 system, they won't be entirely sure what hardware is necessary. Up until that time, it's a science project. We should learn the lesson of Tesla getting that wrong year after year, and Lucid getting it wrong with the Air, and not get our expectations up.

Totally fair, though now that Waymo has had L4 for a few years and continues to expand into different markets with increasing pace, I think we do have a good idea of what does work. What remains to be seen is whether L4 can be achieved with a much smaller sensor suite, and if so, just how small that sensor suite can be. I don't think it's unreasonable to think the industry will have a good idea of that by 2028 given the pace of advancement now.

 

[Need4Sleep]

Yeah, at this point of the maturity everyone is still trying to figure out the autonomy tech

 

[TahoeMichael]

Totally fair, though now that Waymo has had L4 for a few years and continues to expand into different markets with increasing pace, I think we do have a good idea of what does work. What remains to be seen is whether L4 can be achieved with a much smaller sensor suite, and if so, just how small that sensor suite can be. I don't think it's unreasonable to think the industry will have a good idea of that by 2028 given the pace of advancement now.

Yes, trimming down the sensor suite is an open question. But so is exactly how much computing power is needed. I think we're at a point where engineers can make a reasonable estimate of what's necessary. But Waymo is also a cautionary tale, in that their vehicles do still sometimes need to "phone home" and ask for pointers, and a consumer autonomy system probably will be intended to be more independent. Though maybe I'm wrong, and the talked-about subscription fee will be intended to cover the cost of a central operations team.

 

[ThinkTalk]

Though the presentation slides and recording are both published, I will still share my summaries and notes to deliver on my promise. Feel free to inform me of corrections that need to be made.

Erwin Raphael (SVP of Revenue)
-> With Model S and X discontinuing, Lucid intends to introduce this segment of customers to the Lucid line up.
-> Lucid Air was the top-selling luxury sedan in the US, outselling the Tesla Model S, BMW i5, and Porsche Taycan.
-> Even across traditional gasoline vehicles, the Lucid Air was the third best-selling luxury sedan in the US behind the BMW 5-Series and Mercedes E-Class. It even out-sold the BMW 7-series and Mercedes S-Class.
-> As midsize scales up, we can expect Gravity costs to go down as well.
-> November and December ramp-up demonstrate that the 25k-27k production target is possible.
-> Lucid is bullish on Europe because Europe sees a 20% increase in adoption, with Germany being of particular interest to Lucid.

Derek Jenkins (SVP of Design)
-> Acknowledges that midsize is a price-sensitive segment.
-> Midsize customers separated into 3 distinct customer segments: on-road performance/dynamics (Lucid Cosmos), any-road adventure-focus (yet to be announced), and a middle ground between the two (Lucid Earth).
-> The three vehicles from the midsize platform intend to tackle each segment.
-> Identified customer segments plus Air and Gravity market account for 60% of the total EV market.
-> Cosmos drag coefficient to be below 0.22, will focus on sportiness and movement, yet will still target being a leader in terms of space in its category
-> Interior layout will have just one large screen (but with a few tactile buttons) with mechanical analogue door handles, both inside and outside (one personaudibly cheered in the crowd at this lol).
-> AI assistant to be integrated to perform both edge and cloud computing, plus will be LLM agnostic.

Emad Dlala (SVP Engineering and Software)
-> Focus on efficiency, as it will dictate battery size and mileage driven.
-> Midsize will have 800V battery and zonal architecture.
-> Atlas drive unit officially revealed, with goal of lower cost, higher volume, lower weight, and fewer parts.

Zach Walker (Chief Engineerg)
-> Even relative to Gravity, midsize has multiple components merged into one, then assembled directly onto the battery as a single part along the vehicle line.
-> Note that the 69kwh on the slide is not the official battery size, instead it is a reference point to show what is needed for midsize to reach 300 miles compared to the competition.
-> Factory footprint has also improved on the midsize (Saudi Arabia plant) when compared to factory footprint of Air and Gravity (Arizona plant), leading to 40% better footprint efficiency and 50% better labor and overhead.
-> Midsize ECU count reduced to 3 ECUs, with 40% fewer wiring harnesses compared to competition, even claiming that Lucid's team took only 4 minutes to install the harness.
-> Though midsize will not be using gigacastings, it will still use large megacastings with spot-focus use of steel and aluminum (such as extrusions)/
-> Use of steel and aluminum extrusions will allow for cheaper insurance costs and customer repairability/
-> Midsize still features reduced part count for manufacturing cost.

Marc Winterhoff (Interim CEO)
-> Lucid Earth and Cosmos will share 95% of parts
-> Third model will have as much commonality as possible.

Kai Stepper (VP of Autonomy and ADAS)
-> Personal autonomy for consumers, robotaxis to be focused for business-to-business
-> L4 readiness for midsize will be available directly from the factory. (A prior conference discussion with Marc and Taoufiq mentioned that Air and Gravity will require further investment to reach L4 capability)
-> Uber partnership still on track for deployment later in the year.
-> L2++ to L4 autonomy for Dream Drive Pro will have a subscription model.

Marc Winterhoff + Andrew McDonald (Uber President and COO)
-> Uber partnership extends beyond midsize.
-> "Lunar" is a robotaxi concept an design study based on the midsize platform.

Taoufiq Boussaid (Chief Financial Officer)
-> Lucid will be turning from the capital-intensive development stage to focus on the scale stage.
-> Expect 2028 onwards to be when midsize will take the significant portion of production.
-> CapEx was front-loaded for manufacturing infrastructure, expected to shrink in 2027 and drastically reduce by 2028.


My own personal notes
-> Maybe I'm reading too deep into this, but Derek, Emad, and Zach all were wearing Lucid Sapphire pins on their jackets. Given the goal of parts sharing across the midsize line up and the reduction in part count and decreased costs of the Atlas motors, I think a Sapphire trim for the Lucid Cosmos is very much still possible. Previously, Lucid's marketing material mentioned how the Air Sapphire's learnings helped inform the Gravity's development. I think we can expect the same process with midsize, where a Sapphire Cosmos could be released, and the learnings from that can help guide the development direction of Earth and the unannounced adventure vehicle.
-> Happy to see Lucid emphasize repairability with midsize. This is especially important because this will affect insurance rates, and this segment of the market is more sensitive to changes in insurance premiums.
-> Excited to see Lucid expand more into Europe, where it can further prove its credentials in driving dynamics. I think the Cosmos in particular will sell well in Europe, given it's on-road focus.
-> I'm not surprised that 2026 and 2027 will still see a sizeable amount of CapEx, as there is still the need for investment to scale up production.
-> I think the three models across midsize's lineup give Lucid the flexibility and breathing room it needs to take on the adventure-focused Rivian buyer and also the on-road focused performance buyer (such as the BMW Neue Klasse ix3). The reduced costs and focus on parts sharing/commonality could help with creating specialized trims for each of the midsize vehicles (such as a Sapphire trim mentioned previously or even an off-road focus trim like the Lucid Gravity X concept).

Great. Thank you so much. Are situated in Germany? I am about to get more into Lucid. In the next days. Some private informations about me and my "Car" history? Lets talk more, but privately first.

 

[Need4Sleep]

Great. Thank you so much. Are situated in Germany? I am about to get more into Lucid. In the next days. Some private informations about me and my "Car" history? Lets talk more, but privately first.

Unfortunately I live in Canada, but if you want to discuss further, we can continue through direct messages

 

[jedwin]

Waymo is also a cautionary tale, in that their vehicles do still sometimes need to "phone home" and ask for pointers, and a consumer autonomy system probably will be intended to be more independent.

This seems backwards to me. Consumer autonomy should be easier than driverless autonomy. A consumer autonomy system can simply ask the driver to figure it out when things get confusing. I suppose some people want full L4 so they can sleep or drink or make whoopie, but won’t most of us be willing to become the “outside expert“ when needed? My impression is that Waymo‘s “phone home“ fallback is not about split second decisions, but edge case uncertainty about some situation (with a pause in proceeding if needed) so an onboard human, even when not paying attention, should be able to respond in a similar way as the remote Waymo operators.

 

[dbsb3233]

This seems backwards to me. Consumer autonomy should be easier than driverless autonomy. A consumer autonomy system can simply ask the driver to figure it out when things get confusing. I suppose some people want full L4 so they can sleep or drink or make whoopie, but won’t most of us be willing to become the “outside expert“ when needed? My impression is that Waymo‘s “phone home“ fallback is not about split second decisions, but edge case uncertainty about some situation (with a pause in proceeding if needed) so an onboard human, even when not paying attention, should be able to respond in a similar way as the remote Waymo operators.

This is why I don't like using any self-steering until it's so perfected that it's 100% driverless (no steering wheel at all). It's all or nothing for me, because "supervised" means one of two things: (a) I either have to pay just as much attention and be ready to jump in at any split second (which is actually more stressful than just manually steering IMO), or (b) I won't be paying nearly as much attention as I'm supposed to, which means it's really often unsupervised and the risk level goes way up.

Neither of those is preferable to me over just steering myself. And we all know that everyone that uses it is really doing (b), even if they don't admit it.

 

[sri]

Seems investors day is huge success for short sellers, stock just hit another 52 week low. Dont know when will Lucid stop surprising us .
Its literally 91 cents pre split and seems Lucid wants the stock to be low. Lucid how low you want to go

 

[socktucker]

Lucid how low you want to go

Private maybe?

 

[sri]

Private maybe?

Seems thats the plan all along.

 

[socktucker]

Seems thats the plan all along.

That might be the case now but I doubt that was the plan, considering they lost like 5 billion dollars already.