While the market spotlight still rotates between humanoid robots and AI pharmaceuticals, smart driving—a “veteran theme”—is quietly reaching its commercialization watershed. Over the past two years, the “democratization of smart driving” has lowered the entry threshold for advanced driver assistance from the RMB 300,000 level to the RMB 100,000 level, while the deployment of end-to-end large models has turned “software-defined vehicles” from a slogan into deliverable product strength. Looking forward from Q4 2026, 2027 is likely to become the pivotal year when advanced smart driving moves from “human-machine co-driving” toward “whole-vehicle intelligence,” and the starting point for rebuilding the valuation system of Hong Kong-listed smart vehicle industry chains.
Twin Engines of Policy and Technology: L3’s “Green Light” Draws Closer
On the policy front, since multiple authorities jointly launched pilot programs for intelligent connected vehicle market access and on-road operation, testing of Level 3 conditional autonomous driving has kept expanding, with cities such as Beijing, Shanghai, Shenzhen, and Wuhan rolling out local legislation and demonstration-zone rules. Industry consensus expects that around 2027, conditional autonomous driving will move from “pilot licenses” to “scaled market access,” as frameworks for accident liability, insurance, and data compliance gradually close the loop. This step matters because “autonomous driving” for the first time has a clear commercial premise—liability shifts from humans to systems, paving the way for replicable, billable mobility services.
End-to-End Onboard: The Moat Shifts from “Parameter Tuning” to “Data Loops”
The biggest technological variable is the spread of end-to-end architectures and vision-language-action (VLA) large models. Compared with traditional modular solutions, the end-to-end approach replaces rule stacking with data-driven learning, lifting the algorithm ceiling markedly, but the core competitive resource shifts from engineer headcount to data scale and computing-power reserves. Leading automakers and smart-driving suppliers are waging an arms race around the loop of “data feedback—cloud training—OTA iteration,” with cloud training compute becoming the new focus of capital expenditure. For secondary markets, this means the yardstick for judging smart-driving companies must shift from “orders and design wins” to “data assets and compute moats”—the first key to understanding the sector’s divergence in 2027.
Three Main Lines of Industry-Chain Value Reconstruction
Line One: Automakers—From “Hardware Cycle” to “Software Cycle”
The valuation logic for automakers is switching. In the past, Hong Kong-listed new energy vehicle stocks were anchored on sales volume and gross margin; as smart driving becomes a core factor in purchase decisions, consumers’ willingness to pay for software will open a second growth curve: urban NOA subscriptions, pay-per-use valet parking, and future Robotaxi revenue sharing could reprice automakers from “manufacturing companies” to “mobility service platforms.” Automakers with full-stack in-house R&D that genuinely control the data loop may command valuation premiums similar to consumer electronics leaders in the “hardware + services” era.
Line Two: Smart-Driving Chips and Domain Controllers—Domestic Substitution amid Compute Democratization
In chips, Hong Kong-listed smart-driving chip companies are in a resonant phase of “surging shipments + climbing product tiers”: on one hand, making smart driving standard on entry-level models brings volume beta; on the other, flagship high-compute chips benchmarked against global giants raise the ceiling on unit price and margin. Domain controllers benefit from the evolution of electrical/electronic architectures toward “compute moving upward and functions consolidating,” and suppliers deeply tied to leading automakers are poised to ride through the price-war cycle. A caveat: R&D and capex intensity here is extremely high, and differences in the pace of profit delivery will be the core variable driving stock divergence in 2027.
Line Three: Sensors and Robotaxi—From “Cost Cuts and Volume” to “Operations Turning Positive”
LiDAR unit prices have fallen from the thousands of dollars years ago to the hundreds, with adoption logic shifting from “premium optional” to “standard safety fit,” and shipment elasticity is transmitting upstream along the vehicle chain. Even more promising is Robotaxi: large-scale operation starting in Wuhan has proven the feasibility of driverless mobility, with leading players’ fleet sizes and daily rides per vehicle steadily climbing. If major cities further relax zone and time-slot restrictions on fully driverless commercial operation in 2027, the revenue model of operating platforms will shift from “subsidy-driven” to “unit economics turning positive,” triggering a round of expectation re-rating for Hong Kong-listed vehicle, sensor, and operation names.
How Capital Markets Price This Curve
Hong Kong’s uniqueness lies in that most smart vehicle industry-chain companies sit within the pricing orbit of the Hang Seng Tech Index and southbound capital, where liquidity and sentiment amplify effects. By analogy with AI pharma and humanoid robots on the eve of commercialization, markets often first apply “pipeline-style valuation”—ignoring current profits and discounting the future using penetration curves, per-vehicle content value, and market share. Rather than chasing sentiment peaks, investors should build a trackable indicator system: monthly urban NOA penetration, average compute per vehicle, LiDAR shipments, Robotaxi fleet size and daily rides, and the share of subscription revenue in smart-driving software. Only when these indicators show sustained marginal improvement does the valuation-switch signal truly hold.
Risk Variables Not to Be Ignored
- Prolonged price war: vehicle price-cut pressure may transmit upstream, squeezing chip and sensor suppliers’ margins;
- Data and safety regulation: compliance costs rise for cross-border data flows, map surveying, and algorithm filing;
- Liability rulings lagging expectations: if L3 access slows, the commercialization narrative will be postponed;
- Profit-delivery gaps: with heavy R&D spending, the lag between “telling the story” and “booking profits” could trigger valuation pullbacks.
2027 is not the end point of smart driving but a watershed in its narrative. In the past five years, the market priced “can it be done”; in the next five, it will price “can it make money.” For Hong Kong investors, smart driving is turning from a thematic-trading chip into an industry trend that is trackable, verifiable, and compounding. In the long cycle where AI reshapes mobility, true excess returns belong to those who can ride through volatility with an industry perspective. (This article is a forward-looking trend analysis and does not constitute investment advice.)
