"Platform competition has begun, but platform dominance remains undecided." At the recently held 2026 China Automotive Supply Chain Conference, Professor Wang Chao from Beijing University of Technology stated this in an interview with media outlets including Gasgoo.
Today, China's cockpit market accounts for approximately 36% of the global total. It holds clear advantages in areas such as displays, HUDs, voice interaction, and domain controller integration, with multiple segments ranking first domestically. This is a proud asset.
However, a harsher reality exists: overseas giants still hold over 70% of the domestic cockpit domain controller chip market, and underlying software such as real-time operating systems (RTOS) and virtualization are also dominated by foreign suppliers.
In Wang Chao's view, 2026-2027 is the critical window to fortify the foundation. "A few years earlier, conditions were immature; a few years later, positions would have been taken. Why are these two years so crucial? Where exactly are the weaknesses? And how can they be addressed?
Why Is the Window 2026-2027?
Why specifically these two years? According to Wang Chao's assessment, three forces align simultaneously during this period.
First, consider the architecture. Data from Gasgoo Research Institute shows that in 2025, pre-installed cockpit domain controllers reached 9.851 million units, with a penetration rate of 42.9%. In the same year, single-chip platforms integrating cockpit and driving functions began mass production. The cockpit domain controller is evolving from "a box connecting several screens" into a computing node supporting cockpit-driving fusion.
Wang Chao repeatedly emphasizes a temporal logic: platforms selected today must support the next two or three generations of vehicle models. "Once interfaces are chosen, the scope for reuse is locked in." Every architectural decision made today defines the ecological boundaries for subsequent platforms.
Behind this lies a shift in market growth logic.
According to CIC Consulting data, the global passenger vehicle smart cockpit solution market size was CNY 366.8 billion in 2024, projected to reach CNY 762.8 billion by 2029, representing a compound annual growth rate (CAGR) of 15.8%. Notably, the Chinese market's CAGR for the same period is 18.4%, outpacing the global average.
The pie continues to grow, but regional divisions are becoming distinct: "Asia-Pacific drives scale, while Europe and the US lead high-end markets." China dictates mass production rhythms and cost curves; Japan and South Korea maintain long-term depth in areas like instrument clusters and HUDs; and Europe and the US control key entry points for high-end products and rule evolution.
A more intuitive shift is occurring in displays. Global automotive display panel shipments reached 232 million units in 2024, but growth slowed to 5.1% in the first half of 2025. As screen quantity expansion decelerates, the value center shifts from hardware to software-defined platforms. This context accelerates architectural upgrades at this specific time.
AI represents another driving force. Approximately 1.67 million mass-produced vehicles featuring cockpit-driving fusion were shipped in 2025, a 43% year-over-year increase. The pre-installation rate of AI cockpits has also reached 12%. This year is marked as the peak release year for automotive agents, with AI transitioning from "conversational" to "executive."
When intelligent agents no longer just chat or answer questions but can invoke tools, orchestrate vehicle controls, and link services, the requirements for underlying computing power, operating systems, and data architectures are fundamentally rewritten. The leap in AI capabilities is forcing a reconstruction of the interaction foundation, and the window for this reconstruction is now.
The hardest constraint comes from regulations. Regulatory logic has shifted from "encouraging feature implementation" to "non-compliance means no sales." Starting July 1, 2026, two mandatory national standards on automotive information security and software upgrades will apply to new vehicle type approvals. Starting January 1, 2027, GB 47955—2026 imposes strict human-machine interface safety requirements for combined driving assistance, mandating hand-off and gaze-away detection, escalating warnings, and eventual disablement as part of product definitions.
From January 2026 to July 2027, regulations regarding information security, software upgrades, cross-border data transfer, and cockpit intelligence grading will land intensively. Compliance is no longer exclusive to a few high-end models: since 2026, the penetration rate of L2 combined driving assistance has reached 70.5%, and Navigation on Autopilot (NOA) has reached 34.2%. Compliance has become a "pre-mass-production prerequisite" for mainstream models.
Going abroad, the regulatory timeline is equally clear. The EU's Attention Warning Driver Distraction (ADDW) regulation covers new cars within its scope starting July 7, 2026. Design obligations under the Data Act apply to newly placed connected products starting September 12, 2026. With China exporting 7.098 million vehicles in 2025, Chinese cockpit companies expanding overseas must cross these thresholds before entering international markets.
Architectural maturity, AI leaps, and regulatory countdowns align in 2026-2027. A few years earlier, domain controller penetration was insufficient, AI had not moved beyond conversation, and mandatory standards were not yet implemented. A few years later, this round of platform selection would be settled, making changes to interfaces and ecosystems extremely costly.
This is the weight of the "window period." It does not belong to any single company but represents an opportunity for the entire industry to reposition itself in the global value chain.
Strong Terminals, Foundation Needs Reinforcement
To understand what needs reinforcement in these two years, one must first distinguish between strengths and weaknesses.
The most notable strengths lie in terminals and integration. The "China Automotive Parts Industry Development Report (2026)," with Wang Chao as executive editor, reveals that in 2025, Chinese panel manufacturers held a 56.6% share of the global automotive display market, doubling from 28.1% in 2019. In the HUD sector, leading Chinese companies rank first domestically with a 25.6% share. In voice interaction, top vendors hold a 41.9% domestic share.
Integration remains strong. Among the top ten domestic cockpit domain controller assembly providers, mainland Chinese enterprises occupy seven spots, accounting for over half the total share, with the leader holding 16.1%. This aligns with the report's summary of "strong terminals, strong integration."
Moving to the middle layer, local platforms are accelerating. A batch of Chinese automotive operating systems and cockpit solutions are transitioning from single-model delivery to reusable platforms, with some local cockpit solutions reaching deployment volumes in the millions (specifically noted as reaching the 9.4 million unit magnitude).
The true weakness lies at the deepest level. In the domestic cockpit domain controller chip market, Qualcomm's share remained as high as 71.4% in the first half of 2026. RTOS and virtualization layers are dominated by QNX, which has a cumulative global deployment exceeding 275 million vehicles (note: this figure represents cumulative deployments across various automotive systems, not solely cockpit OS share). Automotive-grade EDA/IP and certification toolchains represent the "deepest waters" for domestic substitution.
Above the foundation, competition for four control points is fierce. Beyond computing power, interface control is being contested by mobile ecosystems and international Tier 1 suppliers; next-generation smartphone connectivity already covers all driver-facing screens, including instrument clusters. Update control is held by chip platforms and mobile ecosystems; some international systems cover 16 brands and hundreds of models, beginning to push large models directly onto vehicles. Data and security control exhibit the highest concentration, with security foundation and AI middleware companies having deep roots for years.
Wang Chao summarizes a pattern: the more implicit and closer to the foundation a control point is, the higher its concentration and the greater the difficulty of substitution.
Most alarming is that the substitution curve at the foundation level has flattened. Gasgoo Research Institute data shows Qualcomm's share in Chinese cockpit domain controller chips was 72.7% for full-year 2025, 72.1% in Q1 2026, and 71.4% in H1 2026—a decline of only about 1.3 percentage points from full-year 2025 to H1 2026. The combined share of Huawei, SiEngine, and SemiDrive fell slightly from 14.8% to 14.7%, remaining essentially flat.
Wang Chao identifies two hurdles from this. The hurdle of "can it be manufactured, can it be installed" has been crossed. The next challenge is how to continuously win orders in a market characterized by multi-party entry, downward price pressure, and accelerated model iteration. "It is not just a technical issue, but a commercial one—it concerns continuous delivery and customer relationships."
The numbers hide an easily overlooked illusion: domestic assembly rates do not equate to platform capability.
Wang Chao provides an example: a cockpit domain controller delivered by a Chinese Tier 1 supplier may appear 100% domestically sourced based on assembly rates. However, its chips, RTOS, and functional safety assessments may entirely originate from overseas. Calculated by platform capability standards, the depth of substitution might be less than 30%.
He notes that the Blue Book offers another yardstick: Can the same software stack pass certification and ensure long-term maintenance across two or more chip platforms and two or more regional markets? "By this standard, there are likely fewer than five domestic companies capable of achieving this currently."
The difficulty of the foundation lies in the ecosystem, not just the chip itself. Beyond an automotive-grade chip, there are attached toolchains, virtualization, driver adaptation, and the entire certification system. One must pass four gates: "performance usability, ecosystem usability, certification credibility, and long-term maintainability," each taking years.
One of Wang Chao's statements illustrates the problem well: "Scale can compress a software adaptation cycle from 12 months to 6 months, but it cannot compress an automotive chip's 18-month verification cycle to 9 months."
Meanwhile, fragmentation of basic software is the biggest efficiency black hole. In functional safety, Wang Chao emphasizes that "what converges is the computing platform, not safety responsibility." Data compliance requires shifting from "result-based compliance" to "source governance." For cross-border data transfers, providing sensitive personal information of more than 10,000 individuals cumulatively per year generally requires declaring a security assessment, and collection scopes must be limited at the source.
What if these shortcomings are not addressed? Wang Chao's judgment is unambiguous: "We will remain stuck with 'installation volume but no platform authority.' No matter how large the market share, profits and discourse power will flow to those who control the foundation."
Where to Start Fortifying the Foundation?
Given the urgency of the window, concrete actions are required. Wang Chao offers "the single most important thing to do now" for suppliers, OEMs, and government regulators.
For suppliers, the keyword is "dual-chip platform." Wang Chao advocates that companies should not choose between short-term supply stability and long-term autonomy but rather adopt a "dual foundation": adapting the same software stack to both an international mainstream chip and a domestic chip, compressing the migration workload for chip replacement to less than one-quarter of the initial development effort (a discussion target, requiring project-specific calibration). This ensures short-term supply security while retaining a long-term exit strategy.
A more realistic entry point for domestic platforms is not flagship launches but the renewal cycles of mid-range mainstream models. First achieve volume, then move upmarket. Evidence for functional safety and cybersecurity must be prepared before nomination, not after securing orders.
OEMs need to "finalize the next-generation cockpit platform within this year." They should converge towards two hardware sets and one software stack, embedding compliance into supplier admission criteria rather than waiting for pre-mass-production testing. Software architecture, user entry points, data closed loops, and service operations must be retained internally, while other aspects can be boldly opened up.
He uses a vivid analogy to describe the division of labor: "The upper layer can have personality, but the middle layer must speak Mandarin." Interfaces must be standardized, and test results mutually recognized. Otherwise, applications require individual adaptation, and verifications must be redone repeatedly, leaving the root cause of fragmentation unresolved.
In this platform restructuring, suppliers need not fear being marginalized by OEM self-development. Wang Chao believes that what OEMs truly want to retain are the software stack, HMI framework, and user entry points. Meanwhile, chips determine the cost ceiling, basic software determines reuse efficiency, and engineering verification determines entry barriers. These are precisely the parts OEMs are least likely to complete independently and constitute the lifeline of component companies.
For government regulators, the lever is "building public foundations." Common base software and toolchains, shared verification and certification facilities, and interface/data format standards are tasks that single companies cannot undertake alone, nor find economically viable. Policy focus should shift from supporting specific chips or projects for localization to supporting platform capability building.
Wang Chao specifically mentions the need to clarify the status quo of automotive-grade EDA and IP, noting, "There are currently no public statistics." This aligns with the direction set by the nine-department plan for the "15th Five-Year Plan" for Intelligent Connected New Energy Vehicle Industry Development, which lists operating systems and automotive chips as critical weaknesses to overcome.
Another aspect of fortifying the foundation is completing global delivery capabilities. "Overseas competition is not about price, but about systems," Wang Chao points out. To secure high-end projects in Europe and the US, companies must provide complete evidence chains for software process capability, functional safety, and cybersecurity, along with commitments for ten-plus years of maintenance. Preparing these materials typically takes one and a half to two and a half years. "It is too late to start preparing after receiving a nomination." The real threshold is providing third-party verifiable evidence chains to overseas clients, not mere letters of commitment.
There is more than one path to going global. Wang Chao categorizes three approaches observed during research: Some enter global platforms directly with their products, winning bids alongside the platform. Others follow OEMs overseas, establishing local factories to seek global nominations. Still others use joint ventures to channel system capabilities outward. Though paths differ, the prerequisite is identical: delivering verifiable system evidence before nomination.
This systemic capability is already yielding returns: Top cockpit companies achieve overseas gross margins of 27.3%, higher than the 18.4% domestic margin.
It is also essential to recognize the realistic soil for fortifying the foundation. In Q1 2026, the automotive industry's sales profit margin was only 3.2%. In H1 2026, contract prices for memory chips commonly used in cockpits rose by more than double, while domain controllers are the components with the most concentrated memory usage in a vehicle.
Under financial pressure, Wang Chao suggests a "7-2-1" resource allocation strategy: 70% invested in mature projects with clear customers and timelines, 20% in growth-oriented hotspots, and less than 10% in frontier exploration. He also reminds that alternative supply sources must be "verified." Last Q4, a shortage of low-value components forced OEMs to cut production, serving as a cautionary tale.
Moving from configuration competition to platform competition, the answer ultimately condenses into three keys: definition rights, reuse rates, and evidence chains. Vehicles, components, chips, software, standards, and testing must each guard their respective layers, collaborating to strengthen the chain.
The smart cockpit is the most realistic entry point for Chinese automobiles to break through to the mid-to-high end of the global value chain. The window of these two years will determine whether Chinese companies remain stuck in "share substitution" or truly advance toward "platform dominance."