Hesai Group (NASDAQ: HSAI): Lidar for Robots and Autonomous Driving
Hesai builds lidar — the spinning-and-solid-state laser sensors that let a machine measure the world in three dimensions instead of guessing at it from pixels. It has spent a decade in a brutal price war and came out the other side with something no competitor has: volume, and a profit.
Who they are
Hesai started in 2014 building laser gas detectors, pivoted into lidar, and became the default sensor supplier for the first generation of robotaxi programmes worldwide. That gave it something most Chinese hardware companies didn't have early on: a Western customer base of autonomous driving developers who needed sensors that worked, at any price.
Then the price collapsed. Automotive lidar went from tens of thousands of dollars a unit to a few hundred in under a decade, which killed several competitors and forced everyone else to choose between margin and volume. Hesai chose volume, built its own manufacturing lines, and pushed cost down its own product ladder — the AT, FT and JT series — until lidar became cheap enough to put on a robot lawnmower.
The company is dual-listed, on Nasdaq since 2023 and in Hong Kong under 2525, and manufactures in China plus a new Galileo facility in Thailand built partly to serve customers who want production outside China.
What they do
Two businesses share one factory. ADAS lidar goes into passenger cars — Hesai shipped 353,441 ADAS units in the first quarter of 2026 alone, up 141.9% year over year, and named customers span Audi, BYD, Chery, Great Wall, Leapmotor, Lotus, Pony.ai and WeRide. In 2026 it was confirmed as strategic lidar partner for Mercedes-Benz Level 3 programmes in Europe and China, and as the primary lidar partner for NVIDIA's DRIVE Hyperion 10 platform.
Robotics lidar is the part relevant to this chain, and it is the faster-growing half. The JT robotics series shipped nearly 240,000 units across 2025, and Q1 2026 robotics shipments were 118,282 units, up 137.8%. Management guided the robotics line to at least double again in 2026 and describes the robotics market as roughly ten times larger than ADAS in addressable terms, ranking first in humanoid, quadruped, robotaxi, robovan and robotic lawnmower segments per GGII, Yole Group and Frost & Sullivan.
The showcase humanoid reference is unusually literal: Hesai's JT128 lidar sat on Honor's humanoid robot "Lightning" when it broke the human world record at the world's first humanoid robot half marathon. A robot that runs a half marathon is a stunt, but the sensor that survives 21 kilometres of vibration is a product claim.
How they make money
Unit economics finally turned. Hesai reported the lidar industry's first full-year GAAP net income for 2025, and Q1 2026 marked its fourth consecutive quarter of GAAP profitability with net income of RMB 18.3 million on revenue of RMB 680.6 million. Gross margin was 39.1%, down from 41.7%, as the mix shifted toward cheaper products — the mechanical consequence of winning on volume.
For the first time, the company also split its reporting: the lidar business generated RMB 41.9 million of operating profit, while Strategic Growth Initiatives (SGI) produced no revenue and an operating loss of RMB 50.5 million. That is a useful disclosure, because it tells you the core business is profitable and the company is choosing to spend that profit on a new bet.
Full-year 2026 guidance calls for 3.0 to 3.5 million total lidar units. Management has been candid that blended average selling price will fall, from mix shift and standard annual automotive price reductions. Growing units while prices fall is the entire game here, and it only works if manufacturing cost falls faster.
Where it sits in the value chain
Hesai sits in Stage 4 — Sensing & perception, the stage where the industry has an open philosophical split. Tesla's Optimus is camera-only, consistent with house doctrine. Nearly everyone else pairs cameras with lidar, because unstructured environments punish vision-only stacks in ways a demo video never shows.
Hesai is the outlier in this stage in one important respect: while most sensing suppliers are in a design-win phase with tiny humanoid order lines — Vishay Precision Group booked its first $1.0 million of humanoid orders in Q1 2026 — Hesai is already shipping robotics sensors in six figures per quarter. The catch is that most of those units go into lawnmowers, delivery robots and robotaxis rather than humanoids. It is real robotics revenue; it is not yet humanoid revenue.
The bigger trend
Lidar is completing the journey from exotic instrument to commodity sensor, and that is exactly what makes it interesting for robots. At a few hundred dollars, a lidar becomes a default part rather than a budget debate — and once it is default, the market becomes a volume manufacturing business, which favours whoever already has the lines running.
Two risks sit on the other side of the ledger. The first is geopolitical: a Chinese sensor supplier serving Western robot and defence-adjacent programmes is exposed to procurement restrictions that have nothing to do with product quality, which is part of why Thailand manufacturing exists. The second is margin: if ASP declines outpace cost reductions, the volume win stops translating into profit. The Q1 gross margin step down from 41.7% to 39.1% is the number to keep watching, quarter by quarter.
What to watch
- Robotics lidar unit trajectory against the "at least double in 2026" guide, and the split between humanoid-class customers and lawnmower/delivery volume.
- Gross margin versus average selling price — whether cost reduction stays ahead of the price declines management has already flagged.
- Kosmo turning early orders into disclosed SGI revenue, and whether the SGI operating loss stays bounded.
- Ramp of the Thailand Galileo facility and any customer explicitly requiring non-China production.
- The ETX high-end model with the in-house 6D full-colour SPAD system-on-chip, with start of production guided for the second half of 2026.
Related companies
🇺🇸Integrating lidar and cameras into warehouse bots for lights-out operation — the customer profile for this stage.
🇰🇷The software layer that consumes what these sensors produce.
🇰🇷Robot builder whose platforms need perception hardware per unit.
Also in this stage but not yet profiled: Ouster (Nasdaq: OUST), Vishay Precision Group (NYSE: VPG), Novanta (Nasdaq: NOVT) and Aeva Technologies (Nasdaq: AEVA).
FAQ
Is Hesai profitable?
Yes, and it was first to get there in its industry. Hesai reported the lidar sector's first full-year GAAP net income for 2025, and posted RMB 18.3 million of net income in Q1 2026, its fourth consecutive profitable quarter. Its core lidar segment earned RMB 41.9 million of operating profit in that quarter, partly offset by a RMB 50.5 million loss in its new Strategic Growth Initiatives unit.
How much of Hesai's business is robotics rather than cars?
By units, robotics was 118,282 of 471,723 total lidar shipments in Q1 2026, or roughly a quarter, growing 137.8% year over year. Management describes robotics as an addressable market around ten times larger than ADAS over time, and the JT robotics series shipped close to 240,000 units in 2025.
Do humanoid robots actually need lidar if they have cameras?
It is a live disagreement. Tesla's Optimus uses cameras only. Most other developers pair cameras with lidar because lidar measures distance directly rather than inferring it, which is more robust in low light, on reflective floors and around glass — conditions that are common in warehouses and rare in demo videos. Cost has fallen far enough that the debate is now about architecture rather than budget.
Where is Hesai listed, and where does it manufacture?
It trades on Nasdaq under HSAI and in Hong Kong under 2525. Manufacturing is primarily in China, with a newer Galileo facility in Thailand that supports production for customers who prefer or require capacity outside China, including Mercedes-Benz Level 3 programmes.