HBCI: Jingchu·Xing AI Large Model Empowers Central China's Transportation

Known historically as the "thoroughfare of nine provinces," Hubei Province is a vital integrated transport hub in central China. As the primary entity overseeing provincial highway investment, construction, and operations, Hubei Communications Investment Group (HBCI) manages an extensive network. By the end of 2025, it operated over 6,400 kilometers of highways, with its controlled toll roads accounting for over 70% of the provincial total. This cements its absolute dominance in Hubei's highway sector.

Challenges to Go Digital and Intelligent

In recent years, Hubei's highway network has rapidly expanded. However, the province's complex terrain, high number of bridges and tunnels in mountainous areas, and frequent extreme weather like heavy rain and dense fog have placed immense pressure on road monitoring, emergency handling, and routine maintenance. Imagine a sudden, thick fog rolling over a winding mountain highway: inside the monitoring center, staff are forced to manually sift through hundreds of camera feeds. By the time an incident is detected, its location confirmed, and resources dispatched, over ten minutes have already ticked away—a critical window where every second can mean the difference between life and death.

Traditional traffic management is currently hamstrung by data silos, delayed responses, and limited accuracy. Persistent pain points, such as imprecise incident detection, sluggish emergency command and dispatch, and slow identification of road defects, fall far short of modern traffic governance demands. In order to solve these issues, Hubei has joined China's second batch of pilot provinces for transportation infrastructure digitalization, and China's Ministry of Transport is actively advancing its "One Network, Four Transformations" strategy. Against this backdrop, the province urgently needs an intelligent solution for all scenarios. This solution must seamlessly span construction, management, maintenance, operations, and services, setting a benchmark for AI+ implementation in the road sector.

Digital and Intelligent Foundation Empowers Transportation

In 2025, HBCI launched a strategic plan for comprehensive digital and intelligent transformation to address traffic governance challenges using advanced technologies. As part of this AI initiative, HBCI formed a strategic partnership with Huawei. Together, they used innovative in-house technologies like Atlas, TaiShan, Huawei Cloud Stack, and big data to develop the Jingchu·Xing, an industry-specific AI large model and application foundation.

Independent computing foundation: HBCI has built a foundational platform centered on independent innovation using the "Atlas + TaiShan" architecture. Its supernodes deliver up to 70 PFLOPS of AI computing power. Built on Huawei Cloud ModelArts, the platform supports elastic scaling and intelligent scheduling, dynamically allocating compute resources based on the priorities and demands of training and inference tasks. This minimizes idle capacity and resource conflicts, ensuring stable, high-performance computing power to continuously drive the iteration of transportation AI algorithms.

Data engineering system: HBCI maximized Huawei Cloud ModelArts Studio's data engineering capabilities to perform end-to-end cleansing, precise labeling, and standardized processing of 3.18 billion pieces of multi-source heterogeneous data on highway sensing, tolling, and maintenance. This process produced high-value transportation datasets, establishing a robust data foundation for deploying domain-specific models and agents.

Agent application: HBCI has created more than 20 agents covering core areas such as road network monitoring, engineering construction, road maintenance, public mobility, and financial risk control. These agents anchor the group's "1+5+X" comprehensive digital and intelligent transformation system. Among them is the stand-out AI+ Road Network Situation Prediction Agent. Powered by Huawei's Qitian spatiotemporal engine, it boosts the accuracy of traffic emergency detection by 20% over general-purpose multimodal large models. Additionally, it improves the precision of traffic flow prediction during accidents by 15% compared to traditional time-series models. Another breakthrough is the AI+ Emergency Command and Dispatch Agent. Equipped with intent recognition, intelligent retrieval, Q&A, and reporting, it cuts the acquisition of emergency information from minutes to seconds, slashes the time for data entry by 20%, and achieves a 90% matching accuracy for contingency plans. Furthermore, the Jingchu·Xing model has established an integrated space-air-ground sensing and dispatch hub. It delivers stunning detection of road defects, with a precision of 0.1 millimeters, pushes the accuracy of AI incident recognition above 95%, and compresses the handling times of safety hazards from hours to mere minutes. To date, the model has been successfully validated in real-world scenarios like slope monitoring along the Shanghai-Chongqing Expressway as well as bridge and tunnel construction in deep mountains. It effectively tackles Hubei's toughest traffic governance challenges.

Digital and intelligent product deployment: HBCI has launched two digital products simultaneously. The E-Lu Changtong Jingchu·Xing smart mobility service platform is fully connected to the national E-Lu Changtong system operated by the Ministry of Transport. It aggregates multi-source data streams, including real-time road conditions, charging pile availability, ETC, roadside rescue coordination, and rest area amenities, enabling integrated online processing for all highway-related public services. Qianmotong, a digital management suite for highway construction, combines digital twin and AR handheld terminals to establish an all-domain sensing network. It boosts project management efficiency fivefold, and cuts management costs during construction by 20%. This eases the challenges of managing highway construction in mountainous regions.

Scenario-Technology Integration: Tackling Pain Points

During the R&D and deployment of Jingchu·Xing, HBCI and Huawei developed a "scenario data + core tech" model to drive mutual enablement and joint innovation, in order to better deal with the pain points when implementing transportation large models. On June 30, 2026, they officially launched the Jingchu·Xing AI large model, marking a milestone in their work together.

HBCI opened up real-world scenarios and massive amounts of data: Digital and intelligent construction thrives on high-quality data and real-world validation. HBCI fully unlocked its business scenarios across construction, management, maintenance, operations, and services. Through deep-dive scenario survey and pain point analysis, the group has converged and governed data from multiple business domains, injecting a rich stream of "traffic nutrients" into Jingchu·Xing.

Huawei provided full-stack technical support and shared experience: To tackle transportation pain points and difficulties with multimodal data convergence, Huawei went beyond providing robust computing power and data engineering. Drawing on its proven track record in digital transformation, Huawei integrated technologies like computing power scheduling, model tuning, data engineering, and situation prediction with HBCI's local data and systems. Ultimately, this integration ensures high availability and accuracy for AI large models and scenario-specific agents in complex, real-world environments.

HBCI and Huawei addressed industry pain points and reshaped businesses: Driven by a closed-loop mechanism where business owners identify challenges, tech providers deliver solutions, and both collaborate closely, the two parties have successfully transformed a general-purpose large model into an industry-specific one tailored to transportation and business needs. This integration breaks through long-standing technical hurdles, such as the precise detection of minor road defects and prediction regarding complex road networks. More importantly, it empowers HBCI to pivot its management model from passive response to proactive prevention. This partnership proves a vital truth: bridging the "last mile" of deploying transportation models demands the tight fusion of an advanced technical foundation with abundant scenario-specific data.

Ongoing Evolution: Creating a Model for National Transportation Large Models

Looking ahead, HBCI and Huawei will continue to innovate together to steadily expand the Jingchu·Xing AI agent matrix. Technically, they will use Huawei's digital and intelligent foundation to advance interactions between multimodal large models and transportation infrastructure, enabling data collaboration across provincial road networks. For businesses, they will roll out diverse, integrated services under the "AI + transportation" framework, scaling digital solution suites across Hubei and the whole of China. Fueled by this "scenario + technology" engine, they aim to strengthen Hubei's leadership in digital and intelligent transportation, ensuring that the achievements truly benefit public mobility and power economic growth in the region.