"4x" Hesai: Why has the lidar, once abandoned by Tesla, now 'seen the light' again?

$Hesai(HSAI.US), a leading company in the lidar industry that has attracted significant attention since its inception, achieved a global lidar revenue market share of 33% in 2024, and a market share of 61% in the L4 Robotaxi market!

However, this lidar star stock has also faced challenges. In early 2024, when the market was embroiled in the "pure vision solution vs. lidar" technology debate, Hesai Technology's market value hovered at a low point—due to high lidar costs, low willingness of car manufacturers to adopt, and unclear market space amid technical route divergences in the robot and passenger car lidar tracks, With the shadow of the U.S. Department of Defense's 'blacklist' hanging over it, the prospects for overseas expansion are clouded.

But just half a year later, Hesai achieved a transformative comeback in its stock price.

The starting point for the previous round of valuation surge was the arrival of the industry's breakeven point: In the third quarter of 2024, due to better-than-expected profitability, the stock price began to rise significantly, and the fourth quarter report again exceeded expectations with a single-quarter Non-GAAP net profit of 170 million (exceeding guidance by $20 million), becoming the first profitable leader in the global lidar track.

As of the close on July 14, 2025, Hesai's stock price had increased fourfold compared to the starting point of the previous round of valuation surge before the third quarter report last year!

Heading into 2025, competition among car manufacturers for intelligent driving is becoming increasingly fierce. It began with BYD's grand "intelligent driving equality" initiative at the beginning of the year, but Xiaomi's SU7 intelligent driving accident poured cold water on the field, essentially due to the immaturity of intelligent driving technology. However, this incident further heightened consumer attention to lidar, with the YU7, launched in July, featuring lidar as standard across all models.

At this turning point where vehicle-end intelligent driving is about to explode, and facing the vast space of future robotics, Dolphin Research once again focuses on this leading lidar stock, with core questions about Hesai:

1. What kind of company is Hesai?

2. Why did Hesai achieve a market value explosion after the third quarter of 2024, and what did Hesai do right?

3. How to view the market space for lidar?

4. Is the lidar track a case of the strong getting stronger? Can Hesai maintain its leading position?

5. Is there still room for investment in Hesai?

I. What kind of company is Hesai?

1. Actively positioning in rapidly growing demand tracks

From the founding background of Hesai:

1) Started as a "niche track" gas monitoring equipment company: The company was founded in 2014 based on the severe smog problem in China, focusing on laser spectroscopy technology to develop PM2.5 pollution map cloud monitoring nodes, thus positioning itself as a "gas monitoring equipment company." However, the laser gas analysis industry remains a niche track with a small market size (according to QYR, the global laser gas analyzer market size is about $545 million, with an expected CAGR of only 4.7% from 2025 to 2031).

2) Officially switched to the "lidar" track in 2016: Subsequently, Hesai saw the broader tracks of intelligent driving and humanoid robots as early as 2016, both of which require lidar as "eyes" for external perception (although there is still a divergence between pure vision solutions and lidar solutions in these two tracks).

In 2016, Hesai launched its first domestically produced high-performance radar, Pandar 32, officially entering the lidar track. From 2016 to 2020, several Pandar and QT/XT series products were launched, serving two scenarios: Robotaxi and robots (such as factory forklifts/unmanned delivery fields).

a. Lidar first shipped in the L4 Robotaxi scenario: However, in 2020-2021, the Pandar series became the company's main source of income, marking the first time the company seized the hot opportunity of L4 Robotaxi, achieving a 70% growth in hardware revenue in 2021.

But from the supply side, the Pandar series, due to its mechanical lidar structure, still has certain issues, summarized as expensive, difficult to mass-produce, and prone to failure:

① High cost and difficulty in mass production: The average ASP of the Pandar series is around 100,000 yuan, mainly due to the complex structure and low integration, as well as cumbersome assembly processes, leading to high production costs.

The complex structure results in long production processes, making large-scale rapid mass production difficult.

② Short lifespan of mechanical components, prone to failure: Due to its structure, mechanical radars generally have a short lifespan and are prone to failure;

From the downstream demand side, the main application scenario of the Pandar series is still Robotaxi, with major customers including Zoox, Aurora, Auto X in the U.S., and China's Apollo Go, Pony.ai, WeRide, etc.

In 2020-2021, these players all started directly with L4 high-level autonomous driving, but due to the inherent defects of traditional rule-based algorithms, even using mechanical lidar to supplement the shortcomings of 3D modeling in algorithms, the difficulty of running through the business model is still very high (rule-based algorithms have weak scalability, coupled with the high cost of traditional mechanical lidar), making it difficult for the Robotaxi downstream demand side to achieve explosive growth under the L4 leapfrog algorithm route.

b. Launched the AT/FT/QT series to enter the passenger car ADAS intelligent driving track:

When L4 Robotaxi demand was limited, Hesai launched the AT series in 2021, positioning itself in the passenger car intelligent driving track with greater demand, a broader market, and gradually increasing user demand for intelligent driving. The first AT series product, AT128, began mass production and shipment in 2022.

1) From the downstream demand side:

Due to the increasing competition among car manufacturers for intelligent driving, and the current passenger car manufacturers adopting a gradual route (led by Tesla, Huawei, and emerging automakers), the technical route through a data-driven end-to-end route has a greater possibility of truly running through high-level intelligent driving (L4-L5) compared to the leapfrog route players (Waymo, Apollo Go, etc.) (PS: For a detailed analysis of the two technical routes, see Dolphin Research's analysis in "Ultimate Questioning, Can FSD Really Support a $1.5 Trillion Tesla?").

With the end of electrification innovation approaching, competition among car manufacturers around intelligence is also emerging. Following Tesla's clear mention in its earnings call that smart cars are the future, and cars without high-level intelligence are like flip phone variants that will eventually be eliminated like smartphones (compared to iPhones), competition around high-level intelligence is the trend, and this turning point is not far off.

Although car manufacturers have not yet shown convergence in intelligent driving technology routes (most have switched to end-to-end routes, but are still divided into pure vision (represented by XPeng, Tesla) and lidar factions), as lidar costs gradually decrease, some car manufacturers, even if their intelligent driving technology is in a leading position, still equip lidar as a safety redundancy (such as Huawei/Li Auto).

2) From Hesai's supply side:

Hesai provides semi-solid-state lidar AT series to car manufacturers (detailed product information summarized later), achieving a significant cost reduction compared to solid-state lidar Pandar, and costs are still decreasing annually, better meeting the dual demands of downstream car manufacturers for performance and cost, thereby achieving a significant increase in passenger car ADAS lidar shipments.

After mass-producing lidar for passenger car ADAS in 2022, ADAS became the main force of Hesai's lidar shipments. By 2024, ADAS radar accounted for 91% of total lidar shipments, and ADAS lidar accounted for 68% of total radar revenue, mainly driven by AT128 shipments.

In 2025, Hesai guided total lidar shipments of 1.2-1.5 million units, with 1-1.3 million being ADAS lidar shipments.

This guidance also implies that in the ADAS lidar field this year, Hesai continues to achieve a year-on-year growth rate of 120%-186%, and the main player in ADAS shipments has shifted from AT128 to the lower-priced ATX, which began mass production in the first quarter of 2025.

c. Continued launch of the JT series, targeting the robotics field

In January 2025, Hesai continued to launch the JT series JT16 and JT128, targeting the robotics field. Currently, the JT series is mainly applied to the Dreame intelligent lawn mowing robot and Agtonomy unmanned agricultural vehicle (mainly JT16 shipments), and has not yet been applied to industrial robots/humanoid robots and other broader tracks.

However, Hesai's guidance for the general robotics field in 2025 is 200,000 units, with a year-on-year growth rate of 340%, potentially becoming Hesai's next growth point after ADAS passenger car radar, bringing broader application scenarios for the robotics track.

Summary:

Therefore, it can be seen that Hesai's lidar layout is relatively complete, and the company is good at seizing hot tracks to achieve product switching. Initially, it used the Pandar series to layout the L4 Robotaxi track, and after the track cooled, it catered to the high growth trend of lidar in the passenger car track, using the AT series to layout the passenger car ADAS track (with ATX as the main shipment product in 2025), achieving rapid volume growth in 2024-2025.

In the general robotics field, the company launched the JT series, with current orders mainly for lawn mowers, but still bringing high growth in revenue/shipments for 2025 (Dolphin Research expects robotics revenue to double in 2025), while broader application scenarios in the robotics track are still to be unlocked.

2. The company's product layout in three major tracks is relatively complete, and the ADAS track successfully binds major customers, bringing high growth

① In the ADAS passenger car lidar segment, which currently contributes the main revenue for the company: The company's ADAS radar products include the AT/ET/FT series, with the AT series being the main shipment product, the ET series mainly used inside the car (installed inside the windshield, mainly for aesthetics), and the FT series, due to the immaturity of the Flash pure solid-state solution, has lower precision and detection distance, mainly used as a blind spot radar.

From the company's main shipment AT series, shipments in 2024 were mainly AT128, and after 2025, the new generation ATX achieved another performance improvement, reduced size, and continued price reduction, becoming the main shipment of the AT series (Dolphin Research expects the ATX/AT128 ratio to be 3:1 in 2025, with AT128 no longer shipping after 2026).

From the company's own release of the passenger car ADAS solution, the company hopes to achieve shipments with two versions of low price + high performance (ATX targeting the L2+ market, AT1440 targeting the L4 market), while hoping for a significant increase in the number of lidars equipped in high-level intelligent driving (still to be verified, currently L2+ intelligent driving solutions are mainly equipped with one lidar).

From the downstream customers of Hesai's ADAS radar in 2024, the top three shipment customers are Li Auto, Xiaomi, and Leapmotor, accounting for 93% of ADAS shipments, achieving rapid shipment growth in the passenger car track by binding major customers.

② In the robotics field:

Hesai's main products are the XJ and JT series, with the XT series being the product launched by the company in October 2020. However, due to lower line density, mechanical lidar detection precision and distance are relatively limited, so it is mainly positioned as a flagship product for the delivery robot and mapping market, suitable for automated warehousing, logistics, robots, mapping, and low to medium-speed autonomous driving scenarios.

In January 2025, the company released the JT series, designed specifically for robots and industrial applications with a miniature 3D lidar. This series is compact and lightweight, making it easy to install and hide in various robot applications, currently leading in the lawn mowing robot field, with plans to expand to industrial robots/service robots/AMR logistics/delivery robots and other scenarios.

The JT series is also the main shipment product for Hesai Robotics, with over 100,000 orders (mainly for Dreame lawn mowers), and has also received orders from Agtonomy, mainly applied to unmanned agricultural vehicles, the JT series is the main source of incremental shipments for Robotics in 2025.

③ In the Robotaxi field:

Hesai's main products for the Robotaxi field include the Pandar series and QT series, with main products being Pandar128 (mainly for L4 Robotaxi) and QT128 (unmanned minibuses), and in September 2024, the OT128 was further expanded. However, the application of downstream Robotaxi is still constrained by technical maturity issues, with weak scalability.

Domestic L4 Robotaxi manufacturers have also switched from the previously expensive and short-lived Pandar series to the cheaper and better-performing AT series (such as Apollo Go/Pony.ai, both equipped with four AT128 series radars), so the revenue/shipment growth brought by this income segment is relatively limited.

II. Why did Hesai achieve a market value explosion after the third quarter of 2024, and what did Hesai do right?

From the above analysis, it can be seen that the market value explosion of Hesai after the third quarter of 2024 was directly driven by the rapid growth of passenger car ADAS lidar shipments, as well as the continued high growth trend of ADAS radar shipments implied by the guidance of 1.2-1.5 million units in 2025, and the incremental shipments brought by the JT series on the robotics side.

For the main source of Hesai's current revenue, the passenger car ADAS series, due to the extremely competitive nature of downstream car manufacturers, there is great pressure on costs, so the demand for lidar can basically be summarized as: requiring both high performance and low cost.

The core reason for the significant growth in ADAS lidar shipments of Hesai's AT series is precisely because it meets these two demands:

① Correctly positioned in the technical route, AT series products have stronger performance compared to competitors:

Before discussing the technical route, let's take a look at the main players in the passenger car ADAS lidar track and their market share situation:

There are actually only four main players in the domestic lidar market: Hesai, RoboSense, Huawei, and Seyond. The market concentration of lidar is currently very high.

Since Huawei's current lidar customers are mainly in the smart selection car mode and HI mode (integrated smart package shipments), there is basically no external supply, and Seyond's lidar is mainly supplied to Nio (the 1550nm wavelength solution is relatively expensive), so in the domestic ADAS lidar third-party market track, the actual competition is focused on the competition between the leader Hesai and the second player, Robosense.

From January 2024, Robosense's lidar shipment market share was still in a state of surpassing Hesai, but subsequently, Hesai's market share in domestic ADAS lidar continued to increase, achieving a surpassing of Robosense in July 2024, and after the ATX shipment in the first quarter of 2025, Hesai successfully topped the domestic ADAS lidar leader position in terms of shipment volume.

From the downstream customers, the main reason for the continuous decline in Robosense's market share is due to two major customers: Huawei and XPeng. Huawei has gradually switched to its own lidar route, and XPeng has also started to switch to a pure vision solution, leading to the loss of orders from these two major customers.

From the performance comparison of the main shipment products of the two companies' ADAS radars in 2025:

The prices of Robosense MX and Hesai ATX are basically the same, both around $200, and in terms of actual product performance (the main parameters of lidar measurement are ranging capability and point cloud quality—the fineness of 3D modeling), Hesai slightly surpasses Robosense.

Currently, the maximum ranging of ATX is farther, and the point cloud rate and regularity may be relatively better than MX (due to Hesai's all-in initial cost being higher, but the actual performance is stronger with the scanning mirror route), so in 2025, the number of fixed points obtained by car manufacturers will be more, and it seems that the actual performance of Hesai's ATX with the scanning mirror technology route will be better.

From the comparison of the technical routes of the two companies, the current lidar technology end has not yet shown a convergence state:

The classification of lidar technology routes is relatively complex. In terms of ranging principles, the mainstream lidar is currently based on TOF, and the principle of TOF ranging is as follows:

Lidar actively emits laser pulses, which are reflected back to the receiving end after encountering the target object. The system measures the time difference between laser emission and reception (time-of-flight method, TOF), combined with the speed of light, to accurately calculate the distance to the target object. At the same time, lidar continuously scans the environment to obtain a three-dimensional point cloud, achieving high-precision perception of the surrounding environment.

The current market mainly classifies lidar based on scanning methods: Currently, Hesai mainly uses a one-dimensional scanning mirror solution in the scanning module, while Robosense still uses the MEMS solution.

Hesai all-in on VCSEL (emission module) + semi-solid-state scanning mirror solution (scanning module) in ADAS lidar, currently seems to have relatively stronger performance at the same price:

Hesai has always all-in on the VCSEL + semi-solid-state scanning mirror solution in the technical route of ADAS lidar. Although initially, it did not have the cost advantage of the EEL + semi-solid-state MEMS solution used by Robosense (a more mature solution at the time, with a clearer cost reduction path, mainly requiring the simplification of the number of EEL lasers), so Robosense has always adopted a low-price strategy to seize the ADAS market share (Robosense's gross margin and ADAS lidar prices are lower than Hesai).

However, the VCSEL + one-dimensional scanning mirror solution adopted by Hesai, because each laser line needs to correspond to a real laser, initially had higher costs compared to the more mature EEL + MEMS solution due to the separation of components, but later achieved significant cost reduction through chip architecture iteration + increased integration. Therefore, in the end, the main ATX/MX products of the two companies achieved almost the same pricing, but ATX still slightly outperformed in overall performance (farthest ranging distance, point cloud regularity and density—corresponding to higher 3D modeling accuracy), so currently, the VCSEL + one-dimensional scanning mirror solution adopted by Hesai has both performance and cost advantages.

As a result, Robosense also switched from the MEMS solution to the scanning mirror solution in the iteration from M1 to M3 products, but the technical consistency currently seems weaker than Hesai. Hesai's current technical route may be stronger in point cloud regularity and algorithm adaptability (such as stronger night and extreme scene processing capabilities).

However, it should be noted that since the technical solutions of the lidar track have not yet shown a convergence trend, there is still the possibility of switching technical routes at any time, so it is still too early to judge who the ultimate winner is from the current perspective.

② Achieving significant cost reduction through chip integration and increased integration:

From the current average price of Hesai's lidar, the ATX price is only about $200 (1,400 yuan), which is a 74% decrease compared to the price of the AT128 series when it was first launched in 2022 (around 5,000 yuan), and is also significantly lower than the AT128's $350 (about 2,500 yuan) price, but in overall performance (ranging capability and point cloud rate), it is almost on par with the AT128 series (ranging capability is farther, but the point cloud rate of ATX is slightly inferior to the AT128 series).

In terms of size and weight, compared to AT128, the ATX series has reduced its size by 60% and its weight by about 50%, making it a more cost-effective, aesthetically pleasing, and high-performance choice. As of February 2025, it has received design orders from 11 car manufacturers.

The core of the lidar price reduction lies in the company's significant cost reduction through chip integration and high integration:

In the cost structure of lidar, according to data from the China Academy of Information and Communications Technology, the optoelectronic system accounts for about 70% of the total cost of lidar, consisting of the laser emission module, laser reception module, timing module (TDC/ADC), and control module, with the laser transceiver module accounting for a higher cost, about 60% of the total cost of lidar, being the most core component of lidar.

Hesai adopts a chip design, gradually integrating the originally numerous and large laser control circuits, signal acquisition and conversion circuits, waveform processing circuits, and hundreds of electronic components into a few small chips, achieving high-quality control and calculation of hundreds of laser emission/reception channels through chips. Hesai has already achieved self-research and production of laser emission chips VCSEL, laser reception chips SPAD, and control chips ASIC.

Due to simplified structure, fewer assembly steps, more holistic optical calibration, and higher integration, it has achieved the evolution from discrete lidar to integrated lidar, resulting in a multiple increase in production efficiency and a significant reduction in production costs.

At the same time, the chip design also enables lidar to achieve "Moore's Law": chip performance doubles every 18-24 months, and prices halve, allowing Hesai's lidar to achieve smaller size, lighter weight, longer detection distance, and higher precision advantages.

In the next article, Dolphin Research will answer the remaining three questions not covered in this article: 1) How to view the market space for lidar?; 2) Is the lidar track a case of the strong getting stronger? Can Hesai maintain its leading position?; 3) Is there still room for investment in Hesai? Stay tuned!

<End here>