Beijing, October 27 (CNR News, Reporter Ji Ailing) — The National Data Group was formally inaugurated recently. It will coordinate the advancement of three major tasks: the development and utilization of data resources, the interconnection of data infrastructure, and support for the development of artificial intelligence. This move reflects the broader trend of Digital China construction advancing from top-level design to deeper implementation.

After data is "managed," how can it truly be "used" and become "intelligent"? Currently, China's supply of high-quality datasets is insufficient, massive amounts of data are scattered across various departments and industries, cross-domain circulation barriers still need to be broken down, and data governance costs remain high. Moreover, raw data can only be transformed into usable data assets after standardized governance and processing.

The reporter learned that currently, multiple research institutions and technology vendors, including Tsinghua University, the China Academy of Information and Communications Technology (CAICT), KeenData, and Infinigence AI, are accelerating their exploration of the construction path for AI data infrastructure.

Taking KeenData as an example, at the product level, KeenData's capability chain covers data governance, trusted circulation, multimodal processing, and high-quality dataset supply, extending to model training and inference as well as the full lifecycle operation of agents. Its core product, the KeenData Agentic Lakehouse Platform, is collaboratively formed by the KeenData Lakehouse multimodal lakehouse platform and the Keen Agentic OS agent development operating system. It unifies multimodal data access and governance, high-quality data production, model training and inference, industry knowledge construction, and agent development and operation into a single foundation, achieving integrated synergy between data supply and intelligent operation.

It is understood that this foundation differs from the one-time governance and delivery model of traditional big data platforms. It is designed for real production environments, supporting data to be continuously generated, processed, updated, and invoked alongside business operations. Models continuously read business knowledge, and agents carry out business processes within a complete security governance and permission system. This drives data to shift from static resource accumulation to dynamic and continuous output of intelligent value, completing the transition from a "data backend" to "underlying support for intelligent production."

Yu Yang, Founder and Chairman of KeenData, pointed out that in the era of AI production, the industry's contradiction is no longer "whether there is data or not," but rather a sustainable capability that continuously transforms raw data resources into high-quality production materials usable by models and businesses. As models and agents put forward requirements for real-time, multimodal, and trusted data supply that far exceed those of traditional data warehouses, the construction approach of data infrastructure itself is undergoing a profound transformation from "building systems" to "building capabilities," and from "single-point supply" to "systematic collaboration."

Yu Yang stated that with the establishment of the National Data Group, the future development and utilization of data resources will enter a larger-scale collaboration—jointly participated by national-level platforms, local data entities, industry institutions, and professional technology enterprises. In this collaborative landscape, the role of third-party AI data infrastructure enterprises is not to replace operating entities, but to embed themselves in the construction and operation of infrastructure as core technology supporters.

This division of labor has already shown an observable practical form in the field of trusted data spaces. Taking the construction of a city-level trusted data space as an example, with the overall positioning of a "one-stop hub" for the national data infrastructure pilot program, it is guided by the provincial data bureau and the municipal data resources bureau, overall constructed and operated by a city data group, with KeenData providing core technical support.

This project has built a trinity construction model of "City Data Exchange + City Trusted Data Space + High-Quality Datasets 'Master Ledger'." It has been selected as one of the first batch of trusted data space innovation and development pilots by the National Data Administration. The platform has attracted 1,250 data element enterprises, released 1,031 data products, gathered 21,065 high-quality datasets, and achieved data space mutual recognition with 18 pilot cities nationwide. KeenData was also selected as a typical unit in the first batch of trusted data space pilots by the National Data Administration.

The establishment of national-level data entities, coupled with the booming development of the AI industry, has jointly opened up a vast industrial space for AI data infrastructure. Facing the needs of urban governance, industrial upgrading, and the digital transformation of large organizations, AI data infrastructure is no longer just a storage and computing tool, but a key foundation for activating data elements and developing new quality productive forces.

As the construction of a national integrated data market accelerates and the linkage system between central and local data groups continues to improve, the industry needs more integrated foundation solutions to connect the entire chain of data governance, assetization, trusted circulation, and AI applications, allowing data elements to truly transform into digital momentum driving Chinese modernization. With the national team setting the direction and the technology sector strengthening the foundation, each playing its role and collaborating effectively, the scale advantage of data resources can be transformed into a lasting winning advantage for the intelligent economy.