As the adoption rate of electric vehicles continues to rise, the market demand for Advanced Driver-Assistance Systems (ADAS) has greatly increasedIn recent years, the number of vehicles equipped with LiDAR (Light Detection and Ranging) technology has skyrocketed, particularly in China, where electric vehicles are becoming ubiquitous across the countryMore and more cars are now being equipped with LiDAR, underscoring a significant shift in automotive technology.

In 2023, global passenger vehicle LiDAR shipments surpassed 500,000 units, with forecasts suggesting that shipments could reach 1 million by 2024. This figure represents a market penetration rate of approximately 1%, indicating substantial growth potential that lies ahead.

Since 2021, various automotive manufacturers such as XPeng, NIO, Li Auto, and others, have begun gradually to roll out models featuring LiDAR technology

The number of LiDAR units per vehicle ranges from one to four, and increasingly more manufacturers are planning to incorporate LiDAR into their offeringsAccording to Yole, by the end of the third quarter of 2023, 36 Chinese manufacturers had publicly announced the use of LiDAR technology, leading to around 106 models being launched, which accounted for approximately 90% of the global new car models equipped with LiDAR during this period.

Understanding LiDAR Technology in Automotive Applications

ADAS is categorized into five levels, ranging from L1 to L5, representing increasing levels of driving automationAs a result, the types and numbers of various sensors being used have also increased

• Common Psychological Pitfalls for Investors.
• Tongcheng Targets International Expansion After $8B Valuation
• Can A-shares Reach New Heights with Expected Rate Cuts?
• Global Chip Spending Jumps 31% in Q4
• How Billionaires Navigate the U.S. Tax System

In the earlier stages of the industry’s understanding of ADAS systems, it was widely believed that Level 4 (L4) required LiDAR, whereas many companies did not adopt LiDAR at Level 3 (L3) due to its high cost.

LiDAR can create a 3D view of the roadway by measuring the time it takes for light to travel to an object and reflect back, thus providing a 3D point cloud image of the surrounding environmentThis capability allows vehicles to recognize their surroundings effectivelyWhile there are differing opinions on the use of LiDAR within the ADAS industry, its ability to generate high-definition 3D point clouds of the environment generally positions it as one of the core sensing technologies for ADAS.

LiDAR primarily comes in three types: mechanical, semi-solid state, and solid state.

Mechanical LiDAR is notorious for being difficult to assemble and having a low scanning frequency

Semi-solid state solutions feature decoupled transmitting and receiving units that do not undergo mechanical motion, making them suitable for detecting certain fields of view, like forward detection, with a more compact design compared to mechanical LiDARSolid-state LiDAR includes scanning methods such as MEMS, Flash, and optical phased arrays (OPA).

The transmission module of LiDAR can utilize two main detection methods: the prevalent Time-of-Flight (ToF) and the emerging Frequency Modulated Continuous Wave (FMCW). The most common laser sources are Edge Emitting Lasers (EEL), while Vertical Cavity Surface Emitting Lasers (VCSEL) are gaining attention.

In the reception module, the leading detection technology is Avalanche Photodiode (APD), while emerging options include Single-Photon Avalanche Diode (SPAD) and Silicon Photomultiplier (SiPM). The disadvantages of APD include its large size, high power consumption, limited detection range, and inconsistency

SPAD solutions tend to significantly lower costs while also offering a more compact size and greater stability in performanceSiPM provides a significant advantage with a gain that is 10,000 times greater than that of APD and a sensitivity that is 2000 times better, all while operating at a notably lower voltage.

The Competitive Landscape of LiDAR Technology

Globally, prominent players in the LiDAR industry include U.Scompanies such as Velodyne, Luminar, Aeva, Ouster, Israel’s Innoviz, and Chinese firms like Hesai Technology, RoboSense, TuSimple, and HuaweiIn the early years, foreign companies dominated the technology and product markets in the development of LiDAR

However, with the rapid growth of the Chinese electric vehicle industry in recent years, local firms have poured substantial resources and funding into research and development, gradually catching up to—and in some areas surpassing—their foreign counterparts.

In 2022, Hesai held nearly 50% market share, maintaining its position at the top of the global LiDAR revenue ranking for two consecutive years, growing its share from 42% in 2021 to 47%. TuSimple claimed the second spot with a 15% market share, thanks to a steady supply to NIO vehicles, while Valeo and RoboSense took the third and fourth spots with 13% and 9%, respectively.

According to Yole, in 2023, the top five companies holding market share in vehicle LiDAR are all Chinese, with Hesai, RoboSense, and TuSimple collectively accounting for 80% of the market, while the remaining 20% is divided amongst Valeo, DJI, Huawei, Cepton, and Innoviz.

The Challenges Behind the Brilliance: High Costs and Losses

As it stands, automotive LiDAR remains in its early stages of development, with performance and cost being significant challenges the industry strives to overcome

Currently, the cost of a single LiDAR unit ranges from $1,000 to $3,000. When ancillary materials are included, the expenses for automakers can soar well above 10,000 RMB for LiDAR alone, contributing to the high price point of models equipped with LiDAR, generally around 300,000 RMB.

Despite the booming appearance of top brands in the market, the profitability of their products remains challenging, largely due to these high costs.

Taking Hesai and RoboSense as examples:

In 2022, Hesai reported revenues of 1.203 billion RMB, marking a year-on-year growth of 66.86%, and shipped over 80,400 LiDAR units—an astonishing increase of 467.5%. Their total revenue and shipping volumes dwarfed the combined figures from eight international competitors

However, their net losses also increased by 22.85% year-on-year to 301 million RMB, while their gross margin fell sharply from 70% in 2019 to 39%.

In 2023, Hesai reported a 108.5% revenue growth to 440 million RMB in Q2, with a staggering 946.5% increase in delivery, totaling 52,100 unitsHowever, their gross margin hit a historic low of 29.8%.

RoboSense’s IPO documents reveal that from 2020 to 2022, the company sold 300, 4,000, and 36,900 units of ADAS LiDAR, respectively, with revenue figures of 6.175 million, 40.089 million, and 160 million RMBDespite increasing shipment volumes, the profitability has remained poor, with the average selling price of each unit dropping from 20,500 RMB to 4,346 RMB and net losses growing from 59.93 million to 563 million RMB during the same period.

In other words, RoboSense is employing a strategy of sacrificing profit margins for volume sales to ensure sufficient shipment quantities

As of March 2023, RoboSense secured front-deck orders for 52 models across 21 vehicle manufacturers, placing them at the top globally.

These developments expose the awkward situation faced by LiDAR manufacturers, where despite expanding revenue, profitability has not kept pace.

For Chinese automakers, the fierce competition and overall economic downturn have made cost reduction an essential strategy, which is an unfavorable situation for LiDAR adoption.

In 2023, numerous manufacturers shifted their focus from high-precision mapping as a selling point for advanced intelligent driving towards a lighter mapping approach led by Battery Electric Vehicles (BEV) and Transformative technology.

From recent product releases, it is evident that aside from high-end models, the majority of new products are reducing LiDAR usage

Single LiDAR and dual LiDAR configurations are becoming the mainstream focus in the industry, with companies like Huawei, Li Auto, and NIO adopting single LiDAR, while XPeng and the Great Wall’s brands like Blue Mountain are opting for dual configurations, while configurations using three or more LiDAR units are becoming rare.

To tackle cost issues, companies, especially related chip manufacturers, are pursuing advancements across three technology pathways of solid-state LiDAR, striving to reduce prices down to a few hundred dollars, or even below $100.

Global Industry Pressures

With domestic LiDAR companies holding such high market share and experiencing dismal profitability, international players with smaller market shares face even tougher challenges.

Despite limited profit margins, the aforementioned local companies continue to ramp up shipments

As 2024 begins, several leading firms reported record high monthly and yearly delivery numbers.

Hesai announced that by December 2023, it surpassed 50,000 LiDAR units delivered, exceeding its previous guidance of 220,000 units for the yearRoboSense declared a milestone in December 2023, reporting monthly sales topped 70,000 units, with quarterly sales hitting 151,000, marking a year-on-year increase of over 545.30%, which was more than the total sales from the first three quarters combinedTuSimple stated that their vehicle-end LiDAR deliveries exceeded 150,000 units in 2023, exceeding a 100% growth rate.

Thus, according to disclosed sales data from these top players, total LiDAR sales in China for 2023 are projected to exceed 600,000 units.

As a result, foreign competitors are compelled to reconsider their presence in the vehicle LiDAR market.

In September 2023, Bosch, which had been investing in LiDAR development for nearly three years, announced it would cease its self-developed LiDAR endeavors.

Ouster reported its Q2 2023 revenue exceeding $19 million, marking a year-on-year growth of 88%. However, the gross margin fell to merely 1%, a significant drop from 27% in the same period of 2022, mainly due to the merger impacts with Velodyne that negatively affected company profits, leading to a net loss of $123 million.

Luminar also experienced substantial revenue growth in the first half of 2023, skyrocketing over $30 million and an increase of 83% year-on-year

Nevertheless, it remains in a loss-making position, reporting a loss of approximately $140 million in the second quarter of last year, a 49% increase year-on-yearTo address these challenges, Luminar is optimizing its automated factory in Mexico to meet the demands of downstream automotive manufacturers while announcing plans to establish a plant in Asia through a partnership with TPK.

In early May, Luminar announced plans to lay off 20% of its workforce.

The founder and CEO of Luminar, Austin Russell, stated in a public letter that although the company's core business areas related to technology, products, industrialization, and commercialization are stronger than ever, the company faces unprecedented challenges in capital markets

To date, the company has invested $1.8 billion to build breakthrough technological foundations starting from the semiconductor level and has successfully launched the Standard Operating Procedure (SOP). However, the current business model and cost structure no longer align with the company’s needs.

The 20% reduction in workforce is aimed at cost cutting, while restructuring will streamline the organization to improve financial flexibilityThe company is also seeking to sublease part or all of its facilities to reduce scale.

They anticipate this plan will be implemented immediately and, once completed by the end of 2024, will reduce annual operational costs by $50 million to $65 million

Over the next five years, they expect to achieve over $400 million in cost savings.

Challenges for Pure Vision Systems

In previous years, it was generally believed that for Level 3 and above ADAS systems, the incorporation of LiDAR was essential to guarantee system functionality and safetyHowever, over the last two years, this perspective has shifted, largely influenced by Tesla.

Tesla has remained skeptical about LiDAR for many years, insisting on a pure vision solutionTheir vast database and continuously improved AI functionality have suggested that achieving L3 and above levels of ADAS without LiDAR is feasible, presenting a clear challenge to the LiDAR market and its associated companies.

Today, the vehicle camera systems are primarily divided into visual and sensor-fusion categories.

The visual-based systems rely predominantly on cameras, requiring high levels of algorithm sophistication

Tesla, with its Autopilot 3.0 system, utilizes no LiDAR across its entire fleet, employing eight cameras, one millimeter-wave radar, and twelve ultrasonic sensorsThis arrangement includes three front-facing cameras, four side cameras, and one rear camera, providing the vehicle with a 360-degree view within a radius of 250 meters.

Conversely, multi-sensor fusion systems underscore the importance of hardware over software, with relatively lower algorithm requirementsThese systems are increasingly utilizing larger numbers of cameras; for instance, NIO’s ET7 uses 11 high-definition cameras with 8 million pixels each, while Geely's Zeekr 001 employs 14 cameras, including 7 at the same pixel count

Currently, the most used sensors for ADAS are CIS image sensors, known for their cost-effectiveness

In nighttime scenarios, infrared sensors are crucial, while ultrasonic radars are necessary for parking maneuversAs application demands evolve, millimeter-wave radars are increasingly appearing in vehicles, yet their high costs lead to controversy about their adoption, as seen with Tesla's reliance on CIS and superior algorithms for achieving ADAS capabilities without these advanced sensors.

Musk believes that the pure vision approach offers significant advantages, aligning with human driving instinctsIf a camera can see something, the system can accurately interpret that information to make sound judgments, which is often more reliable than data collected from various sensorsCoupled with developed high-precision maps, advanced ADAS functionality can be achieved.

Even high-end cameras only cost a few dozen dollars each

Thus, from a functional perspective for driver assistance, the pure vision approach presents a more cost-effective alternative compared to LiDAR.

Nonetheless, relying purely on visual information carries its drawbacks, such as 3D modeling limitations, which impede its ability to achieve higher levels of spatial perceptionTesla's counter strategy lies in leveraging AI and advanced software algorithms to resolve these challenges.

Conclusion

Faced with high costs and competition from pure vision solutions, LiDAR companies will need to optimize both cost and technology to ensure sustainable growth in the future.

To further reduce costs, strategies could involve optimizing designs, such as developing proprietary LiDAR chips and integrating System-on-Chip (SoC) designs to enhance system integration.

While each LiDAR company's development path may vary, most of the cost is concentrated in transmitters and sensor chips, representing around 50%-70% of total costs