Gasgoo has learned that from Jan. 6–9, ThunderSoft took the stage at CES 2026 with the theme "AI-Native Road-Ready," hosting the global premiere of Drip OS 2.0 Pre — a next-generation AI-native vehicle operating system. It also rolled out RazorDCX Sylvania, a new AI cockpit domain controller solution, showcasing full-stack strength across chips, OS and ecosystem.
AquaDrive OS 2.0 Pre: from passive response to proactive empathy
Positioned around three pillars — unified experience, AI-native architecture and faster time to mass production — AquaDrive OS 2.0 Pre reworks the in-car experience through new display layouts and interaction design. Closely coupled with the AIBOX, it aims to scale large AI models efficiently into vehicles, enable scenario-based AI agent applications, and accelerate the spread of AI across the smart-car industry.
On the experience front, AquaDrive OS 2.0 Pre links the center display with a panoramic view setup to balance driving safety with richer information density.
Image credit: ThunderSoft
Its "Zhiqiong Fusion Reality" HMI fuses three previously separate layers — SD Map, SR environmental reality and AVM surround view — to put multi-dimensional data onto a single screen for clearer, faster comprehension. A dual-screen, all-scenario 3D rendering mode spans real-time driving, launch mode, a charging home screen and a lounge view. With real-time lighting effects and models rendered at more than 6.5 million polygons, interface transitions flow in a single, seamless "one-take" sweep for an immersive visual experience.
For interaction, the system debuts a "Dynamic Island" AI real-time feature that proactively surfaces navigation, alerts and other key prompts for one-tap access. It also brings an industry-leading Gaze Aware, gaze-sensing HMI that adapts controls to the user's eye movements, adding nuance and warmth to cockpit interactions.
AI-native design sits at the core of AquaDrive OS 2.0 Pre. Built for the AIDV era, it embeds AI from the system architecture up through the user experience, using three key technology breakthroughs to deliver proactive intelligence across scenarios.
At the architecture level, a mass-production-proven device–edge–cloud AI-native framework enables co-evolution across multimodal perception, contextual memory and cross-domain fusion. It delivers millisecond-grade response for critical functions (≤ 500 ms) to meet real-time driving needs, while tapping cloud compute to extend AI capabilities in complex scenarios — unifying efficient on-device response with cloud-driven intelligence.
At the scenario level, deeper fusion of AI perception and decision-making creates an agent matrix covering the full cycle from start-off and driving to parking. The cockpit, as a result, becomes a "travel companion" that senses environmental changes and anticipates user needs — flipping the model from users hunting for functions to functions seeking out users.
On interaction, an AI UI drives a "borderless interaction" model that breaks the traditional dependence on standalone apps: natural language handles intent recognition, reasoning and execution. Generative HMI further personalizes the experience — shifting from passive responses to proactive sensing, dynamic adaptation and emotional resonance.
For mass-production readiness, AquaDrive OS 2.0 Pre leverages its device–edge–cloud AI-native framework to raise resource utilization and development efficiency. It has been fully adapted to mainstream chip platforms including Qualcomm SA8255P and SA8397, establishing stable baselines across platforms to shorten adaptation and debugging — helping automakers iterate and deploy smart features faster.
The system also supports cross-domain coordination across smart cockpit, cockpit–drive fusion, and cockpit + AIBOX setups, enabling optimized software–hardware allocation and deep system integration. ThunderSoft is working closely with dozens of ecosystem partners — including AISpeech, HERE, Dirac, ETAS, ModelBest, Qwen and Volcengine — to broaden use cases and help OEMs build differentiated smart cars for global markets.
RazorDCX Sylvania: an inclusive path to AI across all scenarios
Unveiled alongside AquaDrive OS 2.0 Pre, RazorDCX Sylvania is the centerpiece of ThunderSoft's push to bring large AI models into cars. Built on the Qualcomm Snapdragon Cockpit Platform Ultimate (SA8397), it complements the AIBOX solution to form a comprehensive, all-scenario technology system.
The platform's core performance stands out: CPU compute reaches SPECint2017rate 80, while dense AI compute hits 320 TOPS. That allows a single chip to run complex large AI models alongside traditional cockpit functions — streamlining production, cutting BOM costs and assembly time. A rich set of interfaces leaves ample headroom for upgrades, supporting long-cycle iterations in premium models.
The AIBOX, for its part, offers plug-and-play upgrades tailored to mid- and entry-level models. Sharing the AquaDrive AIOS core ecosystem, the two solutions together cover lineups from high-end to mass-market.
Image credit: ThunderSoft
In tandem with AquaDrive OS 2.0 Pre, RazorDCX Sylvania provides a full development framework and toolchain so OEMs can customize quickly and cost-effectively. Building on production-grade ThunderSoft AIOS algorithms and proven in-car use cases, it helps cultivate a robust application ecosystem. Through a device–edge–cloud collaborative architecture, AI features such as multimodal interaction respond within milliseconds, with intent recognition accuracy above 95%.
"Competition in smart vehicles ultimately comes down to AI compute and the ability to land real-world scenarios," said Chang Hengsheng, executive president of ThunderSoft and president of its Intelligent Automotive Business Group.
The twin product launches mark ThunderSoft’s core play for high-end smart vehicles, integrating the company's OS and AI algorithm strengths into a coordinated "software ecosystem + hardware platform" model — reinforcing its position in premium cockpits. Next, the company says it will stick to a "technology-first, ecosystem-led" strategy to scale large AI models across vehicles.
