The Seeway.ai booth at Auto China 2026. The slogan 'Zhijian Xinjing'  hangs beneath glowing ring structures suspended from the ceiling.


Seeway.ai defies easy description as a map vendor. The company started with maps and localization, but today it is attempting to bundle together multiple layers of vehicle intelligence- data compliance, ADAS, cockpit, chips, and global deployment support. What made this booth at Auto China 2026 worth paying attention to was what it revealed: that China's SDV ecosystem is preparing the invisible infrastructure that makes real-world mass production and overseas expansion possible, well beyond the feature wars. This article is a record of what one company - Seeway.ai - tells us about what a Chinese-style new-generation Tier 1 actually looks like.

By Sang min Han _ han@autoelectronics.co.kr
한글로보기







A massive AI sculpture hung from the ceiling of the Seeway.ai booth at Auto China 2026. Glowing circular ring structures. Below them, a slogan: 'Zhijian Xinjing' roughly translated, "AI opens new frontiers." An impressive phrase. But it raises a question. Opens with what, and for what purpose?
Stepping inside the booth only deepened the confusion. Cockpit, NOA, localization, data compliance, AI agent, embodied AI - these words coexisted everywhere. This no longer looked like the booth of a company that sells maps.
Seeway.ai was once NavInfo - one of the earliest companies to build commercial digital maps for vehicles in China, and one of the first to receive a national mapping and surveying license from the Chinese government in 2002. But today the company calls itself something different. And what it actually sells is no longer simple map data. What this company sells are the invisible preconditions required for vehicle AI to function as a real service in China.
Seeway.ai describes this shift not as a "transition" but as "recognition." For the past two decades, the company has been continuously building what the industry now calls the foundational infrastructure of intelligent vehicles. The way the market perceives them has changed not because they started doing something entirely different, but because what was always there has finally become visible.



Compliance First

A giant unmanned mining truck filled the touchscreen inside the booth. Next to it, the text: 'Wúrén Kuàngk? Hégu? Ch?ngj?ng' - "Unmanned Mining Truck Compliance Scenario." Mining vehicles at a car show? It seemed odd at first. But once the context became clear, this was precisely the thread running through the entire booth.
In China, autonomous driving systems that handle camera, LiDAR, and location data must pass national data regulations. Collected road data must be de-identified, geographic information must be reconstructed according to Chinese standard coordinate systems, and all of this must be managed within an auditable framework.
Seeway.ai explains it this way: "In China, any vehicle equipped with highway NOA or autonomous driving functions must ensure that its camera and LiDAR data complies with Chinese regulations. Data must first be de-identified, and geographic information must be restructured to meet Chinese standards. Our current market share in data compliance exceeds 40 percent."
That 40 percent figure means more than market share. It is the product of map licenses, government networks, and regulatory experience accumulated since 2002. In China, any OEM that wants to apply NOA functions to a production vehicle must clear this regulatory hurdle. And Seeway.ai is positioning itself as the most efficient player to help them do it.
Autonomous driving in China is no longer simply a question of algorithm accuracy or TOPS competition. Every piece of data a vehicle collects - camera, LiDAR, location, maps- must be processed within a regulatory framework. Data export controls, de-identification, geographic information management, and regulatory compliance have become the first real barriers encountered at the point of actual service deployment.
But what Seeway.ai describes as the real difficulty is somewhat different from what people expect. Technically, each problem has an individual solution. Sign a new map contract here, hire a GDPR consultant there, attach a localization vendor for each region. But when a Chinese OEM tries to run all of this simultaneously across eight markets at launch speed, the failure mode is not a single broken link. It is the complexity of the entire system continuously amplifying timelines and costs until competitiveness collapses. What is actually needed is not individual component suppliers, but a single partner capable of owning the entire end-to-end stack. That is where Seeway.ai begins.



The three-tier xD Map structure and industry evolution path.
The convergence from HD high-precision maps toward SD+ enhanced maps is mapped out on a single screen.



Maps Get Lighter. Meaning Gets Deeper

Another screen inside the booth displayed a concept called "xD Map." The structure took the form of a three-tier pyramid. At the base: Road Layer. Above it: RTTI (real-time traffic information). At the top: Semantic Lane Layer. The diagram next to it was even more interesting - HD High-Precision Map → Lightweight HD Map → SD Standard Map → SD+ Enhanced Map. A clear movement from heavy HD Map structures toward progressively lighter forms. At the end of this progression, a line of text:
"The optimal balance between industry consensus, cost, and user experience."
Seeway.ai is not rejecting HD Maps. It is re-engineering them to fit the era of mass-market NOA. Maps are still necessary. But maintaining everything in ultra-high-precision form becomes prohibitively expensive to build and operate. Industry direction is moving toward Light HD Map, SD+ Enhanced Map, and semantic road layers.
Seeway.ai lays out the logic: in highway environments, road geometry is relatively stable, so SD+ Enhanced Map can deliver sufficient performance at substantially lower cost and faster update cycles. Meanwhile, maintaining full HD precision across every road is economically unviable for the mass-market city NOA segment.
And city NOA's core requirement, in Seeway.ai's view, is not more geometric precision. It is semantic depth. Not "where is the lane?" but "what does this lane mean?" - intersection topology, signal phase context, permitted actions at each junction. This is not a map representation for humans to read. It is a road representation for AI to understand and reason about. Not an abandonment of precision, but a rebalancing of the precision-to-cost ratio.
A company that spent twenty years building HD Maps is now talking about the era after HD Maps.
HD Maps will not disappear. In high-value domains like robotaxi operations and premium highway NOA, they will remain. But at the scale of the mass market, the direction of the industry has already shifted. Seeway.ai claims a significant role in forming that consensus.




The partner logo display.
Global OEMs, Tier 1 suppliers, and cloud vendors including Qualcomm, Huawei, BYD, Toyota, DENSO, Volkswagen, and Microsoft.



The Single Window Strategy

The partner logo screen inside the booth was dense: Qualcomm, Huawei, BYD, Toyota, DENSO, Valeo, Volkswagen, Microsoft, and dozens of Chinese AI startups packed into a single display.
Seeway.ai describes its role this way: "When Chinese OEMs export to global markets, they have to negotiate separate contracts with different map service providers in every region. Our solution integrates all of these into one. On the infotainment side, OEMs only need to deal with CVAI."
What Seeway.ai is attempting goes further than that. When Chinese OEMs move into global markets, they face a fragmented landscape of map providers, GDPR requirements, local regulations, and localized HMI implementations - all across different layers. Instead of OEMs contracting separately with map providers in each country, Seeway.ai aims to become a single integration window. This mirrors a pattern appearing repeatedly across China's SDV ecosystem: the way Huawei expanded its cockpit ecosystem through HiCar, the way Horizon Robotics built an algorithm ecosystem around Journey SoC - Seeway.ai is absorbing the map, compliance, and global deployment layers into its own structure.
On one side of the booth, a Japanese navigation UI, a Middle East regional screen, and a North American interface were displayed side by side. Above them: "Defining the Best-Fit Navigation Experience For Every Model, Every Market."
This felt less like a map company and more like a global software gateway for Chinese OEMs.




Actual ECUs and heatsinks inside acrylic cases - not renderings.
Hardware making the argument directly: we are a supplier capable of mass production.



The Heatsink Speaks

Inside acrylic cases, instead of rendered images: actual ECU boxes, domain controllers, cockpit modules. Heatsinks and connector architecture fully exposed.
This scene repeated itself throughout Auto China 2026. Horizon Robotics, Autolink, Seeway.ai - Chinese autonomous driving companies were actively putting PCB designs, thermal structures, and connector architectures front and center. The message: we are no longer demo startups. We are actual automotive suppliers capable of mass production.
Seeway.ai was emphatic about its production references. Voyah, Dongfeng's premium brand, was the first OEM to apply Seeway.ai's highway NOA solution in a production vehicle. Mass production began in August 2024. The vehicle has since become one of the best-selling MPVs in the Chinese market.
Their autonomous driving stack is built on Horizon Robotics' Journey 6 series. Journey 6B handles basic L2 ADAS, lane centering, automated parking, and remote parking. Journey 6E covers highway NOA, ramp entry and exit, and automated lane changes. Journey 6P supports city NOA at up to 500 TOPS. Different SoCs are combined depending on vehicle segment and feature tier, with Seeway.ai integrating perception, driving, and parking algorithms along with cockpit functionality on top.
The relationship between Seeway.ai and Horizon Robotics runs deeper than a typical SoC procurement arrangement. Seeway.ai describes it as "systems integration centered on a specific silicon architecture." Horizon's Journey architecture is relatively open - algorithms, toolchain, and inference runtime are accessible to integrators - and it is precisely that openness that allows Seeway.ai to build an application-layer stack optimized around its own map data, localization engine, and compliance framework. The core point: this architecture allows Seeway.ai to claim the value layer above the silicon.
PhiGo Entry, the integrated drive-and-park solution, has already secured over three million design wins. PhiGo Max targets end-to-end city NOA at 1000+ TOPS. Seeway.ai describes itself as currently the largest customer of Horizon Robotics in China.
The chip business is expanding rapidly as well. Subsidiary AutoChips has shipped cumulative SoC and MCU volumes already reaching hundreds of millions of units, deployed across cockpit, body control, and infotainment applications, with support for CarPlay, Android Auto, and Huawei HMS ecosystem integration. In the industry, these low-cost automotive SoCs are sometimes called “Chuhai Shenqi” - "magic weapons for going overseas" - tools purpose-built to help Chinese OEMs expand into global markets.




The Voyah MPV on display- the reference production vehicle carrying Seeway.ai's highway NOA solution.



Where the Boundaries Used to Be

Three displays sat side by side next to the showcase vehicle at the center of the booth. Driving information on the left. Main cockpit interface in the center. 3D avatar UI on the right. Maps, music, and conversation - connected in a single flow.
What Seeway.ai calls “Cangbo Yiti” - cockpit-parking integration - runs cockpit and parking functions simultaneously on a single Qualcomm 8155 chip. The AI agent operating on top handles Starbucks reservations, destination setting, and music playback. The system does not rely solely on the cloud - it operates at 8 TOPS of on-terminal compute, with AI-generated images and models running directly on local hardware. In the live demo, a substantial portion of camera recognition and voice interaction was processed entirely within the vehicle. The key is an AI agent that operates independently inside the car.
One of the fastest-moving developments in China's SDV ecosystem is the disappearance of the boundary between cockpit, autonomous driving, and AI assistant. Huawei, XPENG, and Li Auto are all moving in the same direction. As vehicles evolve into moving intelligent entities, the cockpit becomes the point of contact between that intelligence and the user. Seeway.ai is targeting exactly that position - attempting to connect data compliance, localization, ADAS algorithms, overseas navigation integration, and cockpit AI interface into a single continuous layer.
And the company does not intend to stop at automobiles. Seeway.ai notes that embodied AI systems and robots also require MCUs and control technology. The localization and edge AI capabilities built in automotive are now extending into physical AI domains like robotics.




The global navigation solution display. Market-specific interfaces for Europe, Japan, the Middle East, and North America, all on one table.



When Regulation Becomes an Export Weapon

Seeway.ai is not a company built for the Chinese domestic market. Seat orientation adaptation, certification support across 168 countries, regional map integration, GDPR compliance, overseas cockpit solutions - all of it was on display inside the booth.
Seeway.ai describes the current state of China's intelligent driving as a "frontier" - a market where suppliers are delivering the world's most advanced functions. And that frontier is no longer a story told only within China. It includes expansion into Korea, Japan, and global markets. The means is not algorithms alone. It is the packaging of regulatory capability and regional integration competency into a single solution.
There is a paradox embedded here. The company that understands China's regulatory infrastructure most deeply is using exactly that understanding as the basis for its global market offensive.
Seeway.ai CEO Patrick Cheng frames this as a new Scaling Law for the Physical AI era:
Capability = Data Quality × Expression Paradigm × Data Volume
The old Scaling Law is no longer sufficient in the physical world. What matters now is not raw data volume alone, but the quality of spatial data, a unified expression framework, and a compliant data channel. These three things are precisely what Seeway.ai has been building for twenty years.
The Beijing Robot Marathon, held just before Auto China, offered a concrete demonstration of this direction. Seeway.ai provided high-precision maps, precise positioning, and data compliance support to multiple participating teams, who then executed fully autonomous driving in complex real-world environments. Seeway.ai describes this not as a concept demo but as proof of capability transfer. Its assessment: automotive and robotics stacks will converge first at the perception and localization layer, then expand into decision-making and control.
"Revenue used to be around $100 million. Now it's $2.5 billion."
The goal has moved from "making vehicles understand roads" to "making robots understand the world." The Seeway.ai booth at Auto China 2026 was a window into where that transition stands today.


 

AEM(오토모티브일렉트로닉스매거진)