Hesai Group (HSAI) Zen Horizon Research Report

Hunting ten-year five-baggers among great growth stocks — pressing the upside question: "Can it get much bigger?"

• QHow large is its market ceiling? Is it expanding an existing pie, or creating an entirely new market?6/10

  A

  Conclusion: Hesai is mainly “expanding an existing pie,” not creating a market from scratch. Automotive LiDAR already exists as a category, and the demand inflection has already happened. Hesai’s job is to lift penetration and market share. The part that truly qualifies as “creating a new market” (robotics / embodied-intelligence LiDAR) has large optionality, but for now it is still a single-digit revenue option and cannot support the ceiling narrative. On Baillie Gifford’s most prized “new market” label, Hesai earns only half a star.

  Start with the size of the pie. The most credible yardstick is Yole’s Automotive LiDAR 2025: the global automotive LiDAR market was about 859 million dollars in 2024 and is expected to reach 3.56 billion dollars in 2030, a CAGR of about 24%. MarketsandMarkets uses a broader definition and reaches about 9.59 billion dollars by 2030 (CAGR 41.6%), nearly 3 times Yole’s figure. That gap itself shows that the main uncertainty around the ceiling is “how low ASP will be driven.” The report explicitly recommends anchoring on Yole, treating M&M as the optimistic upper bound, and rejecting the secondhand summary that “the 2025 China market reaches 31.3 billion dollars” (likely a RMB figure or a definition that includes the full ADAS stack). Strictly speaking, even by 2030, Hesai’s “realized hardware market” is only a several-billion-dollar market, not a hundred-billion-dollar track.

  Why call it “expanding an existing pie” rather than “creating a market”? Because the LiDAR category, automotive-grade certification system, and OEM procurement channel already existed before Hesai entered. Hesai is doing three things that “make the pie larger”: ① pushing price from about $4,100 in the report down to <$200 for ATX, bringing LiDAR into RMB150,000-class vehicles and driving L2 penetration (the report cites about 69% L2 driver-assistance penetration in China’s new passenger vehicles, with about 2.5 units per vehicle on average); ② taking the high-ASP pie in Robotaxi (the report cites Yole’s 2024 estimate that Hesai had about 61% Robotaxi share); ③ thickening the market through shipment volume itself, with 1.62 million total units shipped in 2025, up +222.9% YoY. This is a classic “penetration + share” play, not the definition of a new need that customers previously did not know they had.

  The only part that qualifies as “creating a new market” is robotics / embodied-intelligence LiDAR: exclusive supply to Dreame/MOVA for lawn-mowing / floor-cleaning robots, JT128 installed in Unitree humanoid robots, and, as cited by the report, 239,000 robotics LiDAR units delivered in 2025, up +425.8% YoY. The TAM for humanoid-robot sensor components could reach about 10 billion dollars by 2035 (IDTechEx). But the report is honest in classifying this as “an option rather than a current profit pillar”: estimated current contribution is only about $25-50M, a single-digit percentage of total revenue.

  So the Baillie Gifford-style ruling is: Hesai’s ceiling is “directionally upward with high visibility, but the risk that the scale is overstated is large.” The pie is genuinely getting bigger, and Hesai is indeed enlarging it, but the hard constraint on this pie is “volume-price divergence”: the report notes that shipments rose +222.9% in 2025 while revenue rose only +45.8%, so revenue growth will structurally lag shipment growth. Whether Hesai can upgrade from “expanding an existing pie” to “creating a new market” depends entirely on whether the robotics second curve can turn from an option into a pillar over the next few years.

  Jun 10, 2026
• QCan its revenue at least double over the next five years? Will growth be driven mainly by volume, price, or new businesses?6/10

  A

  Conclusion: Revenue is very likely to double within five years, but it may well do more than double, and it will not be easy. The driver mix is “volume as the main engine, price as a drag, new businesses as optional upside.” Rapid shipment ramp is enough to carry a doubling, but the marginal revenue contribution of each additional unit sold is heavily eaten by ongoing price cuts. That is the most awkward part of Hesai’s growth story: it is one of the few growth stocks where “the more it sells, the lower the unit price gets.”

  Start with the math of doubling. Hesai’s TTM revenue is about 461 million dollars (+41.9%), and FY2025 full-year revenue was RMB3,027.6M / about 433 million dollars, up +45.8% YoY. The near-term growth trajectory recorded in the report is downshifting: Q1'26 +29.6%, and Q2'26 guidance of RMB850-900M (+20-27%). Even using a conservative 25% compound growth rate, revenue would be about 3 times higher in 5 years; the report’s bull case gives 2027E revenue of about $1.0B, already more than double FY2025. So “doubling in five years” is not a high bar. The real questions are quality and upside. That is where Hesai falls short of Baillie Gifford’s ideal of a company that doubles easily while earning more per unit.

  Driver breakdown (the core of this question):

  Volume (the main engine, but depreciating). The report’s FY2026 shipment guidance is 3.00-3.50 million units (+85-116%), and Q1'26 already delivered 471,723 units, +140.9%. Volume visibility is high (160+ vehicle-model design wins, with automotive-grade cycles locking in volume for several years), but the report exposes “volume-price divergence”: 2025 shipments rose +222.9% while revenue rose only +45.8%, implying blended ASP fell from about RMB4,139/unit in FY2024 to about RMB1,868/unit in FY2025, or about -54.9%. So volume is the engine, while price is the reverse thrust.

  Price (a continuing drag, not a driver). This is the opposite of most growth stocks: for Hesai, “price” is not a positive contributor to revenue, but is continuously suppressed by automakers’ multi-supplier strategy. The report directly cites Hesai’s SEC filing warning that it faces “continuous and substantial price reduction pressure from OEMs.” So hoping for revenue growth through price increases is largely unrealistic. The key is whether price can stabilize and stop falling at an accelerating rate.

  New businesses (the core upside, but currently options). Two lines: ① Robotaxi. The report cites Pony.ai’s fleet target raised from >1,700 to >3,500 by year-end, and Baidu Apollo Go operating driverless service in 10+ cities. L4 vehicles can use up to 8 high-ASP LiDAR units per vehicle, giving much greater revenue elasticity than ADAS. ② Robotics, with 239,000 units delivered in 2025, +425.8%. If these two areas scale, they can support ASP and gross margin amid “price declines,” turning “doubling” into “tripling.” But the report honestly marks their current revenue contribution as only single-digit %.

  Ruling: revenue at least doubling over five years is a reasonable base case and may even be conservative. The driver structure is “volume up more than threefold > price down more than half + new-business optionality.” But Baillie Gifford would ask whether this is high-quality growth. The answer is somewhat weak: it relies on volume, on price cuts to win the market, and on new engines that have not yet paid off. Compared with textbook growth stories such as PDD / Tesla, where volume and price move in the same direction and unit economics improve, Hesai’s doubling deserves a discount.

  Jun 10, 2026
• QAfter five years, what will take over as the next growth engine? Does this “second curve” exist today?5/10

  A

  Conclusion: The second curve does exist today, and it is a real, growing seedling (robotics / embodied intelligence + high-ASP Robotaxi LiDAR), not a concept in a slide deck. But it is still too small to take over in scale today. It is a “real but immature option.” Whether it can truly take over five years from now depends on the scaling pace of humanoid robots / lawn-mowing robots and the Robotaxi inflection. Both directions are right, but neither timing nor slope has been proven.

  Why say it “already exists” rather than calling it storytelling: Hesai’s robotics LiDAR is not a demo business; it has real customers and real shipments. The report records JT16 as exclusive supplier for 10 million Dreame/MOVA units (lawn-mowing / floor-cleaning robots), JT128 installed across Unitree’s humanoid-robot lineup, cumulative deliveries exceeding 200,000 units, and 239,000 robotics LiDAR units delivered in 2025, up +425.8% YoY. Q1'26 alone shipped 118,282 robotics LiDAR units. This is a real second curve already generating revenue and growing far faster than the main business, not something sitting on a roadmap.

  Why say it “cannot take over today”: the report honestly frames it as an option at present. It estimates current robotics revenue contribution at only about $25-50M, a single-digit percentage of total revenue, and about 15% of 2025 shipments (including robotics and some non-ADAS units). A segment that contributes a single-digit share of revenue may multiply several times and still may or may not become a “successor engine” alongside the ADAS main business within five years. The question is essentially slope. The report’s long-range anchor is IDTechEx’s “humanoid-robot sensor-component TAM may reach about 10 billion dollars by 2035.” Note that this is 2035, not 2030. The time window is long, making it a “next decade” story rather than a “next three years” story.

  The second potential successor line is Robotaxi / L4 high-ASP LiDAR. The report notes that L4 uses high-channel-count, high-ASP LiDAR, up to 8 units per vehicle; Hesai’s Robotaxi share is about 61% (Yole 2024), and it supplies 8-9 of the global top ten operators. The report identifies 2026 as a “Robotaxi inflection point”: Pony.ai’s Q1'26 Robotaxi revenue grew +395%, with a year-end fleet target of >3,500. L4 LiDAR is “low volume but high ASP and high margin,” exactly the counterweight to the ADAS price war, and it is the nearer-term and more realistic second curve.

  The hard constraints shared by both curves (the risks Baillie Gifford would watch): ① Tesla’s pure-vision route. Cybercab uses pure vision, creating structural LiDAR leakage within Robotaxi additions and threatening the long-term TAM. ② The robotics TAM is large, but whether LiDAR will be partly replaced by cheaper vision / other sensing solutions remains unresolved.

  Ruling: the second curve exists and is a genuine source of imagination that differentiates Hesai from pure ADAS hardware vendors. Robotics is the fastest-growing segment, and Robotaxi is the most elastic segment. This is stronger than many companies whose “second curve” still lives in a slide deck. But today it is “option value, not a profit pillar,” and taking over within five years is “possible, not proven.” An honest Baillie Gifford assessment: the second-curve direction is strong, but its current share is weak; it is positive, but not enough to raise the rating on its own.

  Jun 10, 2026
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next three to five years?5/10

  A

  Conclusion: Hesai’s core advantage is the combination of “cost / scale leadership + self-developed chip-based technology + massive design wins.” Its relative moat is real and is still widening over the next three to five years (competitors dying, share concentrating, L4 / robotics extensions). But its absolute moat is capped, because hardware economics, automaker buyer power, and geopolitical walls prevent its lead from translating into “sustained excess profitability immune to competition and price pressure.” The report’s composite 3/5 (“the best house in a tough neighborhood”) is honest and apt.

  Start with the hardest moat: cost / scale barrier, which is a real barrier. Hesai is the only player in the industry that has converted scale into positive profit: the only company in the industry with full-year GAAP profitability in FY2025 and gross margin of 41.8%, while in the peer set RoboSense’s full-year gross margin was only about 26.5%, Ouster/Aeva are still loss-making, and Luminar filed for bankruptcy in 2025-12. Hesai’s gross margin is about 13-15pp higher than RoboSense’s, a structural cost lead (self-developed chips + scale cost reduction at the Maxwell factory, with the report noting a cycle time of about 10 seconds/unit). In an ASP-halving price war, this is the foundation for Hesai surviving and making money.

  Technology barrier (strong, but not insurmountable). The report cites KnowMade’s 2025 Automotive LiDAR Patent Landscape, which ranks Hesai first globally by patent holdings and patent-family count (about 2,071 granted + pending, third globally in technology influence, behind only Bosch and Waymo). Four generations of self-developed chipization (ASIC) bring high consistency and fast iteration. But the report is honest: LiDAR is not software. Technological lead must be maintained through continuous R&D, and RoboSense / Huawei can follow with same-generation chipization. This is “running faster,” not “unbreachable.”

  Customer lock-in (medium, and weakened). The report records 40 brands / 160 vehicle-model design wins + automotive-grade certification cycles lasting several years, forming an entry barrier and giving volume visibility. But Hesai’s SEC filing warns that it “faces continuous and substantial price reduction pressure from OEMs.” Design wins give volume, not pricing power, and automakers’ multi-supplier strategy weakens lock-in. This is the core mechanism capping the moat.

  Why it is “relatively widening” over the next three to five years: ① the industry has just undergone severe consolidation (Luminar bankruptcy, Cepton acquired by Koito, Ouster reduced to a niche position), forming a Hesai-RoboSense Chinese duopoly, with the report noting Hesai gaining share; ② the Tesla FSD investigation was escalated by NHTSA to 3.20 million vehicles in 2026-03, which instead strengthens the LiDAR safety narrative; ③ L4 / robotics extensions deepen the moat.

  Why it is “absolutely capped”: ① RoboSense is a funded, credible second pole: it achieved its first quarterly profit of RMB103.7M in 2025Q4, with Q4 gross margin rising to 28.5%, so price competition will persist; ② Tesla’s pure-vision path is a long-term structural threat to the mass-market ADAS TAM; ③ the 1260H military-company list + NDAA create a non-operating geopolitical wall that suppresses U.S. market access.

  Ruling: compared with companies such as Cheniere / Siemens Energy that have real contractual pricing-power choke points (Hold 4/5), Hesai lacks pricing power. Compared with a bad business such as Canaan (Avoid 2/5), Hesai is a good company with a real moat. It sits alongside Fabrinet (contract manufacturing with capped thin margins), Harmonic Drive (domestic substitution erosion), and Agnico (zero pricing power cap) at 3/5: the moat is real and widening, but hard constraints cap it. This is the honest Baillie Gifford positioning of “high quality but not top tier,” and the growth narrative should not be inflated to 5/6.

  Jun 10, 2026
• QIf its core business is disrupted, does it have the DNA to reinvent itself? How does it deal with mistakes and bad news?5/10

  A

  Conclusion: Hesai has already completed one textbook reinvention, moving completely from gas detection to LiDAR, proving that it has real DNA for turning around and being reborn when the core business is disrupted. When facing bad news (short reports, litigation, regulatory naming), it has mostly chosen “direct confrontation + rebuttal through performance” rather than concealment. Still, be clear: its “regenerative capacity” is shown more in historical transformation and horizontal extension within the category (automotive-grade → Robotaxi → robotics). It has not yet been tested in a life-or-death scenario where the main lane is comprehensively disrupted by the pure-vision route. Hesai has the foundation for the “reinvention DNA” that Baillie Gifford values, but the biggest exam has not yet been taken.

  Start with the strongest evidence: it really reinvented itself once. The report notes that Hesai originated from co-founder Kai Sun’s Stanford research into laser molecular detection. When the company was founded in 2014, it was making laser methane / gas leak detection products and drone-mounted natural-gas inspection systems. It only shifted to LiDAR around 2016 and launched Pandar40 in 2017-04 to open the autonomous-driving market. This was not a tweak; it changed the entire main lane, from environmental-detection instruments to an autonomous-driving sensor leader. A company that can decisively abandon an existing business early, bet on a market that had not yet formed, and become global number one has the hardest empirical evidence of “reinvention DNA,” stronger than most hardware companies that only defend a single business.

  Horizontal regeneration within the category is also ongoing. Hesai has not nailed itself to the single leg of ADAS: mechanical / semi-solid Pandar/QT/XT serve Robotaxi, while the JT mini-LiDAR series targets robotics / embodied intelligence. Robotics LiDAR delivered 239,000 units in 2025, +425.8%, transferring the same LiDAR capability into lawn-mowing robots and humanoid robots. This “same technology core, multiple end-market fronts” extension is a buffer against disruption in any single market.

  Attitude toward bad news: confront directly, rebut with data, no restatement. This is an important side: ① Blue Orca’s 2025-03 short report alleged 48-67% revenue inflation from 2019-2023, loss of the largest customer, and layoffs of up to 30%. The company denied the allegations entirely, there has been no financial restatement so far, and subsequent FY2025 +45.8% growth and GAAP profitability plus the due diligence for a Hong Kong dual primary listing provide partial counterevidence; ② facing NDAA naming and the 1260H military-company list, the company chose to appeal and respond publicly rather than retreat in silence. The report’s handling is also honest: it does not whitewash the company (the Pacella securities class action partly survived the motion to dismiss in 2026-02 and has entered discovery; it is not over), but it also notes that the short allegations have not been proven so far. This posture of “openly acknowledging bad news and answering with operating results” fits Baillie Gifford’s expectation for how strong management teams handle mistakes and bad news.

  The honest other side: the biggest exam has not been taken. Hesai’s regeneration history is “from instruments to sensors” and “from automotive-grade to Robotaxi / robotics,” both downwind or adjacent extensions within the category. It has never faced an adverse life-or-death test in which the main lane (ADAS LiDAR) is invalidated wholesale by Tesla’s pure-vision route and the company must rebuild its business model again. The report lists Tesla pure vision as a “structural long-term threat to the mass-market ADAS TAM.” If that scenario plays out, Hesai’s reinvention capacity will be truly stress-tested. For now, one can only say it “has the DNA, but it is unverified.”

  Ruling: genuine reinvention history + a healthy culture for handling bad news make this a relatively strong item in the Baillie Gifford framework. But whether it can be reborn a second time if the core is disrupted has no live evidence yet, so it cannot receive full marks. The honest assessment is “positive on resilience, with a reserve.”

  Jun 10, 2026
• QDoes management, especially the founders, have a long-term view, with interests deeply aligned with the company? Are they willing to sacrifice current profits for five to ten years out?6/10

  A

  Conclusion: The three founders have a strong long-term view and are all technical founders who deeply understand the business. Through super-voting rights, they firmly control the company, clearly make long-term decisions, and are willing to spend for the future (the weak quality of reported earnings itself shows they are not optimizing for short-term EPS). “Long-termism” and “founder-led” both stand up. But “deep alignment with shareholders” deserves a discount: through Class A shares with 10 votes each, the founders control about 67.7% of voting power with about 17.7% economic interest, a classic governance risk of “minority economics + majority control.” This means a structural asymmetry of interests between minority shareholders and controllers. Baillie Gifford likes “long-term founder stewardship,” but also worries about “governance imbalance.” On this question, Hesai is “full marks on vision, discounted on alignment.”

  Long-term view and founder leadership: strong. The report notes that all three founders have top technical backgrounds: CEO Yifan Li (UIUC PhD), Chief Scientist Kai Sun (Stanford PhD), and CTO Shaoqing Xiang (Stanford MS), and the three hold almost all Class A shares. This is the standard setup of “technical founders personally steering the company for the long term,” matching Baillie Gifford’s core preference for growth stocks. It is not short-term relay management by professional managers; it is a founder team committing itself for a decade. The company’s actions are also long-term: building the Maxwell factory in-house (capacity from 1 million toward a target of 4 million units), planning the Thailand “Galileo” factory (production in early 2027, both for cost reduction and geopolitical / tariff hedging), and continuing R&D across four generations of self-developed chipization.

  Willingness to sacrifice current profit for the long term: yes, and the evidence is in the financial statements. This point is indirectly proven by Hesai’s “weak earnings quality”: the report breaks down FY2025 GAAP net income of RMB435.9M and shows core operating profit was actually around breakeven (operating profit after excluding other operating income was -RMB12.7M). The company clearly did not sacrifice expansion to make EPS look good: R&D was RMB796.9M in FY2025, 26.3% of revenue, and in Q1'26 it kept investing despite a core operating loss of RMB8.6M, with quarterly net income of RMB18.3M mainly supported by interest. A company willing to trade price for volume, build scale, spend heavily on R&D, and temporarily keep core operations only around breakeven is clearly positioning for years 3-10 rather than serving this quarter’s income statement. That is exactly the “sacrifice today for the long term” Baillie Gifford wants to see.

  Alignment of interests: this is the item that needs an honest deduction. The report states clearly: as of 2026-03-31, all directors and executives together held about 17.7% economic interest and about 67.7% voting power, under a dual-class WVR structure of 10 votes for Class A shares and 1 vote for Class B shares. On the one hand, 17.7% economic interest is not small; the founders are real large shareholders, and their interests are positively tied to company value. This is stronger than companies where the CEO owns only a token stake. On the other hand, “controlling majority votes with minority economic interest” is a classic super-voting governance risk, which the report lists under “governance / short report / litigation risk (medium-high).” It means controllers can still dominate everything through super-voting rights even if their economic ownership is diluted, while minority shareholders lack checks. This is a structural interest asymmetry and cannot be ignored just because the founders are steering the company long term.

  Ruling: Baillie Gifford’s question receives a split answer for Hesai. Long-term view, technical conviction, and willingness to spend for the future are all strong points. But the governance issue of super-voting rights + minority economic ownership control is a real deduction. Compared with companies where founders own heavily and shares are one-share-one-vote, Hesai is “directionally right, with flaws in the alignment structure,” a positive item that must explicitly carry governance risk.

  Jun 10, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its growth sustainable and not dependent on harming society or regulation?5/10

  A

  Conclusion: There are two sides. On “indispensability,” it is moderately strong: if Hesai disappeared tomorrow, 160+ vehicle models and 8-9 of the global top ten Robotaxi customers would feel real pain (automotive-grade certification cycles last several years, and equivalent replacements are hard to find in the short term), but because automakers generally use multiple suppliers and RoboSense is a credible second pole, customers would truly “miss” Hesai, yet it is not “irreplaceable.” On “growth sustainability / not harming society and regulation,” it is clearly better: LiDAR is fundamentally a component that improves driving safety, its growth mode is positive, and it does not rely on harming consumers or exploiting regulatory loopholes. But Hesai has a special “regulatory knife” hanging over it: U.S. geopolitical regulation of Chinese LiDAR. This is not “Hesai harming regulation,” but “regulation targeting it on national-security grounds.” The direction is opposite, yet it still suppresses the growth option. Under Baillie Gifford’s double test, Hesai is “missed by customers, socially positive, but singled out by geopolitical regulation.”

  First layer: indispensability is moderately strong, but not monopoly-grade. The source of customer stickiness is real: the report records 40 automotive brands and 160+ vehicle-model design wins, coverage of China’s top ten automakers, LiDAR standard across Li Auto / Xiaomi hit models, and wins with Mercedes L3 and GAC Toyota bZ3X. On Robotaxi, it supplies 8-9 of the global top ten operators (Baidu Apollo exclusive long-range supply, Pony.ai, WeRide, Zoox, Aurora, Nuro, DiDi, etc.), with about 61% share (Yole 2024). Automotive-grade certification cycles lasting several years + embedding into production-vehicle BOMs mean that if Hesai suddenly disappeared, customers would face real short-term pain from production stoppages / redesigns: they would “miss it a lot.” But be honest: ① automakers’ multi-supplier strategy exists precisely to avoid being held hostage by one supplier, and Hesai’s SEC filing warning that it “faces continuous and substantial price reduction pressure from OEMs” shows customers have bargaining and substitution power; ② RoboSense has already achieved quarterly profitability in 2025Q4 and is a credible second pole, able to absorb some demand. So it is “painful but replaceable,” not utility-like sole-source indispensability.

  Second layer: whether growth is sustainable and not harmful to society and regulation. The social-value side is very positive. LiDAR’s core use is to provide high-precision environmental perception for ADAS / autonomous driving and improve road safety. The report notes that the Tesla FSD investigation was escalated by NHTSA to 3.20 million vehicles in 2026-03, which instead strengthened the safety narrative for LiDAR, while Waymo and Chinese Robotaxi operators continue using LiDAR. Hesai’s growth comes from “putting cheaper safety-redundancy sensors into more cars.” This is a business with positive social effects; it does not rely on addiction, consumer harm, or regulatory arbitrage. That is fundamentally different from businesses whose growth is built on social cost, and Baillie Gifford would award credit here.

  But there is a regulatory knife pointing in the opposite direction, and it must be stated plainly. Hesai’s regulatory issue is not that “it harms regulation,” but that “the U.S. treats Chinese LiDAR as a controlled object on national-security grounds”: the report notes that NDAA Section 164 names Hesai, bans Department of Defense procurement / use, and takes effect at the end of this month on 2026-06-30 (but the Department of Defense is almost not a customer, so direct revenue damage is ≈0); Hesai remains on the 1260H “Chinese military company” list, senators have called for delisting, and the SAFE LiDAR Act (H.R.6576), if enacted, would threaten about 15% of North American revenue and the overseas high-margin growth option. The report honestly states that this is a policy-dependent binary tail risk with highly uncertain direction. The key distinction: this weakens the “geopolitical sustainability of Hesai’s growth path” (overseas market access), not the “social sustainability” of the product itself (which is beneficial). The two must be separated.

  Ruling: customers would miss it (moderately strong, not monopoly), the social character of growth is positive (a plus), but a geopolitical constraint from being singled out by foreign regulation hangs over it (a minus unrelated to the product’s social value). Under Baillie Gifford’s double interrogation, Hesai is a “useful and legitimate business whose overseas imagination is capped by a geopolitical wall.” Its indispensability does not reach the sole-source bucket.

  Jun 10, 2026
• QHow are the unit economics of this business (gross margin, incremental returns)? Do they improve or deteriorate with scale? Where is the cash it earns being spent?5/10

  A

  Conclusion: The unit economics are “among the best in hardware, but still fundamentally manufacturing, with early signs of pressure.” Gross margin around 40% is far ahead of LiDAR peers (RoboSense about 27%, most peers still loss-making), and scale has indeed brought cost declines and GAAP profitability. That is real capability. But the key issue is “low accounting earnings quality, with the core product-selling business only around breakeven.” Incremental returns are eaten by price wars, the money earned is mainly reinvested into R&D and capacity expansion, and free cash flow remains negative. On this Baillie Gifford question, Hesai is “gross-margin leader + scale with positive feedback, but the ceiling on unit economics is manufacturing margins, and they are being tested by ASP decline.” It is much better than a pure cash burner such as lunr, but it does not reach the software 70% gross-margin tier.

  Gross margin: Hesai’s strongest unit-economics advantage. FY2025 gross margin was 41.8%, compared with RoboSense’s full-year figure of only about 26.5%, Ouster/Aeva still loss-making, and Luminar bankrupt. Hesai is about 13-15pp above RoboSense, reflecting structural cost leadership (self-developed ASIC chips + scale cost reduction at the Maxwell factory). More importantly, the report points to evidence of “positive scale feedback”: in an ASP price war where ASP roughly halved by about 55% in one year, gross margin only eased slightly, and FY2025 stayed broadly flat at about 42%. This means cost declines moved in step with price declines, making it “price for volume” rather than “loss of pricing power.” Scale really has made it the only player in the industry that has converted scale into positive profit.

  Incremental returns and whether scale makes the business better or worse: better, but early pressure is emerging. Positive side: R&D expense ratio has been significantly diluted with scale, one of the main reasons for GAAP profitability. The report notes that R&D fell from 41.2% of revenue in FY2024 to 26.3% in FY2025. That is classic operating leverage: the larger the scale, the more fixed cost is spread. But the honest warning sign is that Q1'26 gross margin had already fallen to 39.1% (YoY -2.6pp), which the report calls an “early pressure signal.” So “larger scale → better economics” currently holds, but automaker price pressure is the counterforce. Whether gross margin can hold above 38% is the lifeline of this business’s unit economics (the report lists “gross margin below 38% for two consecutive quarters” as a downgrade trigger).

  The most important cut: low earnings quality, with core operations only breakeven. This is what Baillie Gifford would watch closely. The report breaks down FY2025 GAAP net income of RMB435.9M: after excluding “net other operating income of RMB181.4M” (government subsidies + Ouster arbitration compensation + patent transfer, the last two one-off), core operating profit was -RMB12.7M; about 71% of GAAP net income came from other operating income and interest on a roughly RMB7.2 billion cash pool. Q1'26 GAAP operating income was still a RMB8.6M loss, and quarterly net income of RMB18.3M was even below quarterly interest income of RMB56.8M. In other words, looking only at the “selling LiDAR” business itself, current unit economics are merely breakeven. This is the fundamental gap between Hesai and platform businesses with “software 70% gross margin, where scale prints cash.” The ceiling is manufacturing profitability.

  Where the money is spent: R&D + capacity expansion, with free cash flow still negative. The report records positive operating cash flow for three consecutive years (FY2025 RMB117M), but capital expenditure of RMB309M > OCF, leaving free cash flow ≈ -RMB192M. “Cash flow turning positive” applies only at the operating-cash-flow level. The money mainly goes to: ① R&D (FY2025 RMB796.9M to maintain technology leadership); ② capacity (Maxwell expanding to 4 million units + Thailand Galileo factory). Net cash on the balance sheet is about $941M (about 33% of market cap), but the report honestly notes that this came largely from the 2025-09 Hong Kong equity raise rather than pure operating cash generation.

  Ruling: unit economics are top tier within hardware (gross-margin leadership, positive scale feedback, real R&D dilution), far better than pure cash-burning peers. But four constraints cap it: “core operations only breakeven + incremental returns eroded by price wars + negative free cash flow + manufacturing-margin ceiling.” It does not reach Baillie Gifford’s highest bucket of “the bigger it gets, the more it earns, with unit economics steadily improving.” It is a good business, but not a cash-printing machine.

  Jun 10, 2026
• QWhat conditions need to hold simultaneously for it to rise fivefold over ten years? Are those conditions realistic? What expectations are embedded in today’s share price?3/10

  A

  Conclusion: A fivefold rise over ten years (Baillie Gifford’s passing line, let alone tenfold) is “possible but demanding” for Hesai. It requires four to five conditions to hold at the same time: revenue growing 5-6 times, core operating profitability truly turning positive with margin climbing toward 20%+, the second curves (Robotaxi / robotics) becoming pillars, gross margin holding through the price war, and the least controllable condition: the U.S. regulatory tail not detonating + the Chinese ADR delisting threat not materializing. Each condition is reasonable in isolation, but the joint probability of all of them holding is not high, especially because the last two are policy black swans outside the company’s control. Today’s $18.44 share price implies “reasonable and slightly cheap; the market recognizes it as a high-quality leader but gives no premium.” It neither discounts a tenfold blue-sky case nor prices in a regulatory collapse. The pricing is relatively neutral.

  Current price anchor (pin down the base first): HSAI closed at $18.44 on 2026-06-09, with market cap about $2.88B and 156.44M shares; PE 41.65 / forward PE 27.87, TTM revenue $461.45M (+41.9%), TTM GAAP net income $68.39M; net cash about $941M (about 33% of market cap). A fivefold rise means market cap of about $14.4B; tenfold means about $28.8B.

  Conditions that must hold simultaneously for a ten-year fivefold return (Baillie Gifford-style decomposition):

  1. Revenue grows 5-6 times. From about $433M in FY2025 to about $2.2-2.6B, requiring about 18-20% annual compound growth. The report’s bull case already gives 2027E revenue of about $1.0B, so the path is not unrealistic. But near-term growth is downshifting (Q2'26 guidance +20-27%), and volume-price divergence means revenue grows more slowly than shipments. This condition is “realistic but requires continued delivery.”

  2. Core operating profitability truly turns positive, with net margin climbing toward 20%+. This is the most important condition and the report’s first hard indicator for moving from Watch to Buy. Today, core operations are only around breakeven (FY2025 operating profit after excluding other income was -RMB12.7M), and current “profitability” relies on subsidies + one-offs + interest. A fivefold return needs profits to be driven by selling products rather than interest on the cash pool. That is a qualitative shift from low to high earnings quality, and it is not easy.

  3. The second curve becomes a pillar. Robotaxi (2026 inflection, Pony.ai Q1'26 revenue +395%) and robotics (2025 deliveries +425.8%) must grow from today’s single-digit revenue share into real high-ASP / high-margin pillars that offset the ADAS price war. This is the main source of upside elasticity and the largest source of imagination.

  4. Gross margin holds. During continued ASP decline, gross margin must stay around 40% (Q1'26 has already fallen to 39.1%) and not converge toward RoboSense’s about 27%. This is the precondition for the durability of its scale advantage.

  5. Policy black swans do not detonate (least controllable). The SAFE LiDAR Act does not become law, the 1260H appeal is mitigated, and the Chinese ADR delisting threat does not materialize. The report calls this “the largest asymmetric risk is policy rather than operating slowdown.” These conditions are completely outside the company’s control and are the hardest sword hanging over the tenfold narrative.

  Are these conditions realistic? Each individual condition is not absurd, but the real Baillie Gifford question is “joint probability.” Having all five hold simultaneously and persist for ten years is materially difficult. The first four are operational and can plausibly be worked toward; the fifth is geopolitical and purely exogenous. This is the core reason the report gives “Watch (Neutral)” rather than “Buy”: upside is capped by both the business model and the external environment.

  What expectations are embedded in today’s share price? forward PE 27.87 and EV/Sales only 4.2x, down about 40% from the 52-week high of $30.85, with net cash at about 33% of market cap. This implies that “the market recognizes it as the only truly profitable leader and gives it a reasonable multiple, but not a high-growth premium.” The report judges the current price to be at the lower end of the base-case band ($18-24), reasonable and slightly cheap, with an ideal entry at ≤ $15. In other words, today’s price neither prices in a tenfold blue-sky case (which would require forward PE to stay at 30-35x + revenue above $1B), nor prices in a regulatory collapse (the bear case is $12-15). It is neutral pricing for “a high-quality company that is not temptingly cheap, with upside requiring multiple conditions to be delivered.” The honest Baillie Gifford conclusion: a ten-year fivefold return is an open possibility, not a high-probability outcome; buying today means paying a fair price for “optionality,” not picking up a cheap compounding machine.

  Jun 10, 2026
• QWhy has the market not recognized all this yet? Is it because it does not understand, looks down on it, or cannot look far enough? What will become the “narrative inflection point”?3/10

  A

  Conclusion: In Hesai’s case, the market has actually “understood most of it, but three things are suppressing a rerating.” It is not “not understood” (it is a well-covered star stock with 22 institutions at Strong Buy); it is mainly “looked down on + not looked far enough”: ① the geopolitical / delisting discount on Chinese ADRs structurally depresses valuation (looked down on, an identity discount rather than a business discount); ② the truth of “low accounting earnings quality and core operations only breakeven” deters some investors who demand clean profits (the market sees clearly and therefore does not pay a premium); ③ robotics / Robotaxi second curves are treated as distant options, and the market is not yet willing to pay for them (not looking far enough). The most likely narrative inflection points are “core operating profitability truly turning positive” or “clear mitigation of the regulatory tail.” If either happens, the logic suppressing valuation will loosen. On Baillie Gifford’s final question, Hesai is “a company with some perception gap, but the direction is neutral; it is not a severely underappreciated hidden gem.”

  First reject “not understood”: it is actually a stock that has been examined closely. Hesai is not an uncovered obscure name. The report records consensus Strong Buy (22 firms), with an average price target of about $30 (about +60% above the current price), and Goldman Sachs at $36, betting on overseas high-margin growth. It is LiDAR’s “quadruple champion” (number one in revenue, share, patents, and the only full-year GAAP profitable player) and is heavily covered by institutions. So this is not a perception gap where “the market does not understand the business.” Most professional investors understand very well that it is the best industry leader.

  Real suppression 1: “looked down on” through the Chinese ADR identity discount. This is the largest piece. The report notes that the 1260H military-company list still includes Hesai, senators are calling for delisting, and the NDAA names it, restricting U.S. institutional ownership of the ADR. This is a structural current pressure on the “investor base”: some U.S. capital is blocked by law / compliance. It is not that they are bearish; they cannot buy or do not dare to buy. The result is EV/Sales of only 4.2x and PS of 6.3x, in line with loss-making RoboSense and far below loss-making Ouster (about 14x) / Aeva (about 75x). The report concludes that “Hesai trades at a discount, not a premium” and “the market recognizes it only with the one real PE, but does not give it an expensive multiple.” This is classic “looked down on” (identity discount), not “value unseen.”

  Real suppression 2: “seen clearly, so no premium” because of earnings quality. The market has understood this layer too: about 71% of FY2025 GAAP net income came from other operating income and cash interest, while core operations were only around breakeven. Investors who demand “clean profitability” will discount it for this reason. This is not a blind spot; it is restrained pricing after rationally identifying “low earnings quality.” So Hesai is cheap “for a reason,” not a mistaken selloff.

  Real suppression 3: “not looking far enough” because the second curve is not yet priced. Robotics (2025 deliveries +425.8%) and Robotaxi (2026 inflection) currently contribute only a single-digit percentage of revenue and are treated as distant options. The market is not yet willing to pay for a long-term imagination such as “humanoid-robot sensor-component TAM of about 10 billion dollars by 2035.” This is the only perception gap close to “not looking far enough.” If the second curve delivers, upside rerating space exists.

  What will become the narrative inflection point? Based on the report’s downgrade / upgrade triggers, the two most likely positive inflections are: ① sustained positive core operating profit (excluding subsidies / interest). The report explicitly states this is the first hard indicator for moving from Watch to Buy. Once confirmed, the “low earnings quality” suppression logic collapses directly, and valuation can migrate from a “discounted real PE” to a “quality growth premium.” ② clear mitigation of the regulatory tail (SAFE LiDAR Act stalling / being shelved, 1260H appeal victory, delisting threat receding). Once the identity discount is removed, the U.S. capital base can recover, and EV/Sales of 4.2x has room to re-rate toward the upper peer range. Negative inflections would be SAFE LiDAR Act entering markup or becoming law, gross margin below 38% for two consecutive quarters, or adverse progress in the Pacella lawsuit. Any of these would push the narrative bearish.

  Ruling: Baillie Gifford’s favorite situation, “the market is badly wrong and you are uniquely right,” does not really apply to Hesai. The market understands the business quality fairly clearly. The perception gap mainly consists of two neutral pieces: “identity discount (looked down on) + second curve not priced (not looking far enough),” rather than a “dust-covered great growth stock.” The honest answer to this final question is: there is a perception gap, but its direction is not strong; the inflection points are clear and trackable, making it “worth watching for those inflections rather than heavily betting today on underappreciation.”

  Jun 10, 2026