We previously discussed the relationship between the pricing of a lidar sensor and its performance. In this part, we focus on the role of the market in constraining pricing — and the balance that has to be struck between customer expectations on performance and the attendant cost tradeoffs.
In addition to lidar performance (discussed in Part 1), another important factor to consider in lidar pricing is understanding the price that the market would be willing to bear at a given unit volume. To discuss this with some helpful context, let’s review one example from Sabbir Rangwala’s article at Forbes — “Money for Somethin?”. Rangwala stated that based on his analysis of “projected price”, he found that the price announced by one or more lidar suppliers was significantly lower than his projections. He discussed several reasons why this might be the case — such as, software vs. hardware mix, innovations in system design, scaling of bill of material (BOM) costs, and quality/manufacturability. Purely for the sake of analysis, let’s consider one of the lidar companies discussed by Rangwala — Luminar (a Cepton competitor in some markets).
Assessing Market Price Expectations
Rangwala noted that in their IPO-related filing, Luminar announced a unit price roughly ~4–5X lower than what he projected from his analysis. His chart reports Luminar’s announced price (in the IPO-related filing) at $1,300 and Rangwala’s projected price at $5,671. Luminar’s projected revenue is captured in slide 23 of their presentation and a screenshot is included below for quick reference.
Using the information in the slide, we’ve calculated Luminar’s projected average selling price (ASP) per unit for 2023–25. We also estimated the hardware (HW) versus software (SW) ASP per unit using the information in slide 23, assuming that each unit sale involves both HW and SW.
Luminar also stated in slide 23 that:
Revenue primarily driven by automotive series production programs, commencing in 2022E
With their end market clearly defined, it provides us a set of useful information to make some relevant observations tied to market pricing.
Let’s start with automotive OEM expectations. The highest end, partial automation features in the automotive market today — such as Tesla Autopilot and GM SuperCruise — are typically sold to consumers at a premium price adder in the $6,000 — $8,000 range. That pricing includes all sensors (lidar, camera, radar, etc.), computing and software required to support the full feature set. Hence, OEMs considering adding lidar to their sensor stack — to offer L2+/L3 ADAS/AD features — typically expect lidar pricing to be in the $500-$1000 range (with some exceptions where prices need to be lower than $500 and fewer exceptions where pricing slightly above $1000 would be considered). This price range typically applies at volumes approaching tens of thousands of units. This is an important consideration for OEMs because their ability to sell higher-end cars to consumers in volumes approaching tens of thousands of vehicles a year would generally require them to price their premium offering only a few thousand dollars higher than standard offerings.
With that backdrop, we can look at what Luminar is projecting for their business. In 2023, Luminar expects a total lidar ASP per unit of ~$3,960 (and a lidar HW ASP of ~ $2,700 per unit) at a volume of ~31 K units. If the lidar portion of the overall vehicle sensor stack costs that much at that volume, the OEM vehicle price premium for ADAS/AV features that use the lidar would likely be priced significantly higher than other similar offerings in the market. This is certainly a possibility, but it would be somewhat unique. (Even if the ASP is partly influenced by non-recurring engineering and development related revenue (NRE), excluding an NRE amount similar in magnitude to prior year estimates from Luminar still leaves the ASPs fairly high.)
Software vs. Hardware Pricing
Another interesting aspect about Luminar’s offering and pricing is their expectations regarding software revenue. Based on our estimates (see table above), their software revenue per unit is >45% of hardware revenue per unit in 2023. This fraction is expected by them to drop slightly in 2024 and then rise to ~75% of hardware revenue by 2025. The corresponding SW ASP in 2025 is projected to be ~$560 per unit at a volume of around 630 K units. To be clear, Luminar has a different business model from competitors and may have good justification to have such an expectation. Regardless, based on our experience in the electronics, systems and automotive industries, software ASPs that high — both in absolute terms and as a fraction of hardware revenues — are likely to be unique at such high unit volumes. Stated another way, a total ASP of ~$1300 per unit at volumes of 633K will be in the high end of the spectrum when it comes to OEM and market expectations.
The purpose of the example we used above is to illustrate the market pricing constraints that come into play in setting a competitive price for lidars. Even when OEM customers are in a position to accept higher prices, they have to contend with the reality of vehicle sales when their competitors offer similar advanced features at lower prices. As a result, we need to add this important consideration to Rangwala’s observations in his article — namely, the upper bound price constraints that are market driven, which therefore require lidar suppliers to be very efficient in their business models and operational execution at scale. This is one of several reasons why we believe collaboration with automotive Tier 1s becomes important, because many Tier 1s have figured out how to streamline operations to achieve acceptable margins at scale, and thereby achieve pricing levels (for non-lidar sensors) that would support high volume vehicle sales. For Tier 2 suppliers, there is an understandable concern with introducing a Tier 1 into the relationship with an OEM — namely, the price adder that comes into play. However, in our experience, as long as the collaboration is structured correctly, with the right intellectual property (IP) ownership and roles/responsibilities, the benefits substantially exceed the potential downsides.
Performance vs. Cost Tradeoffs
In the scenarios we considered so far, we considered two opposing factors that impact pricing — performance and market expectations. In reality, there is also a third factor, namely, the extent to which an OEM customer is willing to negotiate on performance in order to enable them and their suppliers to land within a mutually acceptable price band. Not surprisingly, our experience is that customers do consider performance — price tradeoffs seriously to avoid situations where a product is overdesigned to meet a specific requirement but is then priced in a manner that becomes a barrier to sell a high volume of vehicles. Let’s consider a few examples:
Range: Although it is common to see a 200 m (or higher) range requirement for 10% reflective targets when it comes to L2+ ADAS/AD applications, it is also common for customers to be willing to adjust this expectation downward for certain classes of applications. Higher range requirements are usually associated with higher vehicle speeds — in other words, the faster the vehicle is moving, the farther the visibility that is required since the braking distance to avoid an accident would be higher. So, if an OEM’s goal is to first introduce a Traffic Jam Pilot (TJP) type feature in their vehicles, before introducing a Highway Pilot (HWP) type feature, TJP could allow for a partial relaxation in range specs since this feature is used at lower speeds.
FOV: For a front-looking lidar (FLL) application, range and angular resolution near the edges of the lidar’s FOV tend to be less important than near the center of the lidar’s FOV. This could potentially alleviate constraints in lens and beam steering designs, which in turn have an impact on cost.
Size: An incredibly important property of the lidar that is often missed in general discussions of automotive lidar applications is size. OEM styling and design teams have to deal with the high complexity resulting from the growing sensor count in vehicles — hence, sensors that require more embedded real estate (e.g., surface area or volume) are harder to integrate. So, a sensor that is more compact and easier to integrate could have a disproportionately positive impact on other tradeoffs that an OEM might be willing to accept.
Power consumption: As consumer vehicles get increasingly packed with electronics and sensors, managing power consumption gets more challenging. Therefore, having sensors with lower operating power helps in two respects — it reduces overall power requirements and it reduces thermal heat generation, which might otherwise require more complex heat dissipation or some form of active cooling to keep the sensor operating properly, thereby leading to higher cost. This is another topic on which OEMs are likely to be more open when discussing performance tradeoffs.
These examples are just some of many aspects that come into play in detailed discussions with Tier 1s and OEMs to find the right balance between performance, embeddability, reliability and cost. Cepton’s ultra-compact, high performance lidar sensors are best-in-class in power consumption. They use our patented MMT® (Micro-Motion-Technology) to enable automotive grade reliability. By using a modular, licensable and highly manufacturable architecture that strikes an optimal balance between performance, size, reliability and cost, and through collaboration with our Tier 1 partner, we have been able to successfully enable smart tradeoffs for OEMs and significantly simplify vehicle integration. The Vista-X90 is the newest introduction to our autograde lidar family that embodies some of these design principles.
