I really appreciate your constant LIDAR analysis in these posts. I always expect to see a comment from you when posts are about self driving or lidar advancements.
I'm all in favor of LIDAR, but I don't see what makes this company worth $3.4 billion as a public company. The investors are putting in $570 million.
Here's a list of the top players in automotive LIDAR: Delphi Automotive, Continental, ZF Friedrichshafen, Infineon Technologies, Velodyne Lidar, Texas Instruments Incorporated, First Sensor.[1] Not seeing Luminar on that list.
Their 2022 product seems to be yet another rotating scanner. 120 degree horizontal field of view, 30 degree vertical. Possibly with some steering on the vertical axis to deal with hills. Nice renders of the packaging. Why is this worth $3.4 billion?
There are lots of other cheap LIDAR real soon now companies. Quanergy. LeddarTech. InnoViz. Continental bought Advanced Scientific Concepts, which had expensive solid-state LIDAR working well. Continental put that product on hold because nobody wanted to order a few hundred thousand of them. Continental is an auto parts maker, and they do volume products only. They will sell you an expensive prototype if you want one. Delphi seems to be taking a similar position.
There's another problem with this. In the end, they're an auto parts company. Auto parts companies get hammered on price by automakers. Hard. They do not get to have high margins.
Silicon Valley companies are used to being able to sell things with essentially no warranty. Here's what a major auto company demands of its suppliers:
"Seller warrants and guarantees that the goods covered by this Contract will conform to all specifications, drawings, samples, descriptions and quality standards (i) furnished by Buyer, or (ii) furnished by Seller and approved by Buyer in writing, and will be merchantable, of good material and workmanship and free from defect. In addition, Seller acknowledges that Seller knows of Buyer’s intended use of the goods covered by this Contract and warrants and guarantees that such goods have been selected, designed, manufactured or assembled by Seller based upon Buyer's stated use and will be fit and sufficient for the particular purposes intended by Buyer. Unless otherwise set forth in this Contract, the duration of the warranty provided by Seller to Buyer for the goods will begin on the date of receipt of the goods by Buyer and end on the later of ... (b) expiration of any warranty applicable to the goods provided by Buyer to Buyer’s end customer for the vehicle into which the goods are incorporated.[1]
Yes, the parts supplier is on the hook for warranty costs during the warrantied life of the vehicle.
Oh, and as to liability:
In the event that any goods fail to conform to the warranties set forth in this Contractor the product specifications incorporated by reference in this Contract, or if Seller otherwise breaches any of its obligations under this Contract, Buyer will be entitled to recover from Seller any and all damages, including, without limitation, any direct, indirect, incidental and consequential damages and all legal and other professional fees and costs incurred by Buyer as a result of such breach or failure, including, without limitation, costs, expenses and losses incurred by Buyer ...
Right. If a LIDAR failure causes a car crash, the supplier pays. Everything. Complete reverse of typical Silicon Valley terms. There's a lot more like that, and it's all in favor of the buyer, the auto company.
Agree with you that it looks odd. Strikes me like a scramble to compete with the Velodyne-GRAF deal. From the Luminar SEC filing[1]:
* Volvo and one other OEM are main customers
* $15m in revenue this year. Claimed $837m revenue in 2025. Sounds like when Uber ordered 100,000 S-Classes for ATG. [2] In contrast, Velodyne was at $100m last year, yet has only projected to $600m for 2024 [3].
* Will not make 1,000 sensors per year until 2023.
* Today, Luminar claims the TAM for their market is $4B. So the $3.4B valuation suggests they somehow have already captured a majority of their market. Well, they certainly captured a majority of their own deal with Volvo.
The sensors have some great performance, but the economics look like pure information arbitrage.
I liked the Advanced Scientific Concepts technology when I first saw it in 2003 or 2004 in Santa Barbara, when it was on an optical bench. No rotating machinery. A later version is used by Space-X to dock the Dragon spacecraft with the ISS.
Their device was very expensive because it required custom ICs made using an unusual InGaAs process. But that's the kind of thing that becomes cheap when production gets into millions. Continental is an auto parts manufacturer, where every penny per part is carefully counted. If they thought they could get the cost down in volume, they probably could.
But no market appeared. Continental appears to be able to make the things; they have demoed and offer a prototype. Setting up an assembly line to bang them out has to wait for a big order. Automatic driving isn't far enough along to generate such orders.
Which is the big problem in all this. Nobody needs a million LIDARs yet.
I understand Reverse-Mergers generally, but I don't really understand how the money works.
?People? buy stock in an existing 'shell' company.
The shell company buys the target company and immediately re-brands itself as the target company.
I guess I don't understand valuation and dilution and who has ownership after the acquisition. It would seem that the founder would want to have a lot of the stock (more than 50%), but the owners of the shell company would also want to retain the ownership of their stock... Where does the stock come from for equity for the founders and employees?
Lidar is almost always discussed in relation to self-driving vehicles.
What are some other applications that stand to benefit significantly from a price reduction? Submersibles come to mind as a cost-prohibitive tech where this could help somewhat.
LIDAR has interesting space applications for Guidance, Navigation and Control (GNC).
For instance, I worked on the design of a mission concept for multi-target Active Debris Removal [1], for which flash LIDAR was an important part of the GNC package, in addition to stereo visual cameras, and various other sensors. For non-cooperative targets, LIDAR is a pretty important technology, due to possible varying lighting conditions.
LIDAR has been used for close proximity GNC for the International Space Station too [2].
Compact, low-power flash LIDAR will open up a slew of applications like on-orbit satellite servicing and refuelling.
Solid setup's for accurate BIM model building, the current scanners need a bunch of time to use because it is cost prohibitive to have more than one, this makes the scanning somewhat costly as you're paying for a techs time to setup/wait for scanning.
Build a mesh network of a dozen sensors and join up the results to scan entire buildings in a few hours and you'll have some very happy engineers.
It's hard to see submersibles ever being more than a tiny niche market. There just aren't very many. And commercial and military submersibles often have to operate in poor visibility conditions where lidar range is near zero.
They may sell later, but not at the time of IPO itself. Often, there's a lock-up period, where current stockholders are not allowed to sell to the public until a certain amount of time has passed.
Not surprising imo. There are a lot of lidar companies that have come out and just like how self driving car teams/startups are almost required to be strapped with an OEM or big co (torc with mercedes, zoox with amazong for example), specialized lidar companies like Luminar and Ouster are pretty much on a straight path to a merger or acquisition.
Awesome stuff. I'm actually more excited to see what all the autonomous vehicle compute and sensor stuff will be used for in different domains. Just like cheap SoCs and MEMS from smartphones ended up being integral to the drone industry taking off.
I see the applications/overlap with driving down the costs/size of multi-mode precision-guided munitions: guided missiles and drones that use a mix of laser, radar, and infrared sensing for all-weather accuracy. I'm not really a hardware guy but I'd love to prototype something Spike-sized with COTS sensors, then try to setup a factory in a developing nation.
An SPAC lets the company negotiate a firm, fixed price, whereas an IPO is (semi) exposed to the market. Of course that comes at a cost - the company accepts a lower average price in return for reducing its risk. You can understand why that has become a bit more appealing in the last few months.
Ive read on HN SPAC mergers are good for “clearing out saints” ie firing people otherwise harder to terminate. It should in theory minimize volatility of the price after the merger as opposed to an IPO.
How do we tell whether autonomous cars are actually making progress? Is this a big bang final reveal, or is it possible to tell that every quarter, something on the way to full autonomy is developing?
It seems that these spinoffs and acquisitions of Lidar companies, etc. all say they solve a piece of the puzzle, but it still seems no closer to a real system.
> The CEO at Luminar, Russell, is 25. At age 17 he received a Thiel fellowship to build out Luminar.
Does the Thiel fellowship mean he dropped out of College? I am not familiar with the founder but this feels odd and wrong to me given it's describing a (technical?)founder for a autonomous driving startup? Fully autonomous driving is a really hard technical problem and I believe you need a deep understanding of many advanced topics understand the boundaries of this space, yet alone innovate.
Yes, that was the point of the Thiel fellowship. It gives out grants if you apply for people who decide to drop out / not go to college and pursue something ambitious.
Taking a risk implies that you are going to be personally impacted if everything falls through.
Peter Thiel is worth $2.1b and almost all VCs are investing institutional money.
So I think it's ridiculous to even mention them in the same breath as founders who are typically working 70+ hour weeks, dealing with immense stress and are poorly paid for an idea that is statistically likely to fail.
> Founders deserve the credit for their startup's success not the VCs.
That varies. Often founders are truly independent. Occasionally a founder is a glorified employee of the main investor. In these cases the investor may also bring customers, technology, and/or employees, but the products tend to be less interesting.
I only bring it up because it’s an interesting thing I didn’t expect until I participated in SV.
While it would be nice to be addressed as a Nobel laureate, I'd actually prefer the title of MacArthur Genius Grant winner. Both named for dead guys whose fortune paid for the award. Go figure.
