Bernstein is excited about this company’s ambitions in Physical AI

Hesai reported 46% YoY revenue growth and a 14% net margin in 2025, and guided 2026 LiDAR shipments to 3.0–3.5 million units versus 1.6 million in 2025 (nearly double). Management announced a strategic pivot into 'Physical AI' with two product launches planned, while Bernstein maintained an Outperform and trimmed its price target to $32 from $33. The aggressive shipment guidance could materially expand Hesai's ADAS market share if margins sustain, though the transition to AI products may require incremental R&D that could pressure near-term profitability.

Analysis

A supplier that attempts simultaneous high-volume hardware scaling and a pivot toward software-driven “Physical AI” creates divergent competitive pressures: OEMs and Tier-1s will favor partners who can supply integrated stacks (sensor + perception) and guarantee cost, quality, and recurring update services, while pure-play optics or small LiDAR vendors will face accelerating margin compression and consolidation risk. Expect immediate downstream effects in contract manufacturing, ASIC/FPGA sourcing and testing capacity — these bottlenecks will determine whether gross-margin retention is structural or temporary.

Key risks cluster around execution and certification rather than demand per se. Automotive design cycles and safety validation operate on multi-quarter to multi-year cadences, so near-term product launches are binary technical/certification catalysts; a failed demo, a supplier shortage, or warranty/recall event can reset adoption expectations quickly. Conversely, credible design wins and predictable supply chain throughput will re-rate upside into a multi-year software-monetization story.

From a market-structure viewpoint, the pivot favors vertically integrated players and big compute partners that can internalize perception stack development and cloud/edge updating, creating a two-tier market: low-ASP commodity sensors and higher-ASP integrated systems. This bifurcation creates tradeable dispersion between firms that can defend software economics and those stuck on hardware-only unit economics.

The consensus tone looks optimistic on conversion to software-like margins; the overlooked friction points are integration cost per vehicle, recurring revenue attainment rates, and the cadence of OEM validation. Watch OEM design-win announcements, ASP trajectories, R&D spend as a percent of revenue, and inventory days as the loose/strict signals that will validate or bust the narrative over 3–18 months.

Key Decisions for Investors

• Long HSAI (core position, 6–18 months): allocate 2–4% of book, scale in on 10–20% pullbacks; target 2x upside if multi-OEM design wins or recurring revenue indicators appear within 12 months. Stop-loss at -35% to limit execution/certification risk.
• Pair trade — Long HSAI / Short LAZR (12 months): size 1.5:1 long/short exposure to capture dispersion between an integrated-scaling vendor and a pure-play LiDAR competitor. Expect 20–40% relative alpha if ASP compression favors the vertically integrated name; tighten if either posts explicit multi-year OEM commitments.
• Options strategy (18 months): buy HSAI LEAP calls (12–18m) sized at 0.5–1% of NAV and fund by selling near-term (3m) calls for 30–40% of notional. This asymmetric structure captures long-dated outcome optionality while monetizing near-term volatility; be prepared to roll if product launch timelines slip.
• Hedge / sympathetic longs: add small exposure to NVDA or APTV (3–12 months) as convex hedges to a positive Physical AI outcome — NVDA for compute upside, APTV for Tier-1 integration wins. If execution falters, trim hardware-heavy exposure and rotate into downstream integrators.