Where data surges, intelligence gathers along the Min River. As the first year of the “15th Five-Year Plan” intertwines with a new journey of Digital China construction, an in-depth dialogue on unlocking the value of data elements unfolds in Fuzhou.

On April 29, the 9th Digital China Summit was held in Fuzhou. At the “Big Data Technology Innovation” Sub-Forum—guided by the Ministry of Industry and Information Technology and organized by the China Center for Information Industry Development， Zhu Jianyong, Vice President of Beijing KeenData Technology Co., Ltd., was invited to deliver a keynote speech entitled “Data Reconstruction: New Infrastructure for the AI Industrialization Era,” in which he deeply analyzed the key bottlenecks and breakthrough paths of AI industrialization.

With the theme “Promoting Big Data Technology Innovation to Empower Industrial Digital and Intelligent Transformation,” the sub-forum gathered elites from government, industry, academia and research to discuss big data technology innovation and industrial integration, pooling wisdom and strength for the development of the big data industry during the 15th Five-Year Plan and accelerating the construction of Digital China.

The core of AI industrialization lies not in simply pursuing stronger computing power or larger models, but in the in-depth synergy of computing power, algorithms and data，with AI data infrastructure serving as the key hub for this synergy.

Zhu Jianyong opened his speech with striking data: global AI investment continues to grow rapidly, yet 70% to 85% of AI projects fail to deliver expected value conversion—with massive investments not translating into effective financial returns. AI industrialization is trapped in a “hot investment, cold landing” dilemma.

He pointed out that the global AI industry is entering a critical phase where high-speed growth and in-depth industrialization go hand in hand. AI competition has shifted from a superficial contest of computing power and algorithms to a deeper race centered on data, the “AI IQ.” Model capability sets AI’s lower limit, while AI data infrastructure defines its upper limit. Zhu emphasized: “Data defines AI’s upper limit, and scenarios determine application value. Without data infrastructure, even the most powerful models remain superficial. Only high-quality, structured, scenario-based and sustainably supplied data can enable AI to truly integrate into business processes, industries and governance.”

Data reconstructing AI means data is no longer a mere “supplement” to models, but the core engine powering intelligence.

To address pain points: including the disconnection between traditional data platforms and AI development, challenges in unified governance of multimodal data, and high costs of building industry agents, Zhu formally proposed the AI-in-Lakehouse innovation concept, reconstructing the data infrastructure system for the AI era. KeenData’s KeenData Lakehouse data intelligence platform breaks the capability limits of traditional data warehouses and data lakes, enabling unified access, governance, annotation and operation of structured, unstructured and multimodal data, while deeply integrating model training, inference, knowledge enhancement and full-process agent development.

This architecture directly responds to the national requirement of “promoting the construction of high-quality industrial datasets,” transforming data from a static resource into high-quality “fuel” that AI can directly access. It seamlessly connects the entire “data–model–intelligence–value” chain, enabling enterprises to build a data foundation without starting from scratch.

As a national key “Little Giant” enterprise (specialized, sophisticated, innovative) and a core independent and controllable vendor in the Xinchuang (IT Application Innovation) sector, KeenData adheres to independent R&D of core technologies. Boasting a code self-development rate of over 97%, more than 200 invention patents, and a leading market share in the advanced technology lakehouse data platform sector, KeenData has earned authoritative recognition from international institutions like IDC and Gartner.

Technology’s value lies in application. On-site, Zhu shared that KeenData has implemented benchmark cases across multiple key industries: in energy, it transforms systems from resource-driven to data-intelligence-driven; in finance, it supports safe, inclusive and intelligent risk control and services; in retail and consumer, it drives digital and intelligent transformation from “selling products” to “understanding users”; in urban governance, it enables resilient governance from decentralized management to comprehensive coordination.

Building AI data infrastructure requires in-depth synergy across the industrial chain. KeenData will deepen ecological collaboration in three key areas: technology co-construction, open ecosystem, and scenario co-creation. Working with industry partners, we will develop demonstration models of “data infrastructure + high-quality datasets + agent applications,” activating new momentum for the digital economy through AI data infrastructure. This will enable artificial intelligence to truly evolve from technology to engineering, from pilot projects to large-scale application, and from demonstration to mass production.

Every small step adds up to a great distance; diligent practice paves the way for far-reaching goals. In the grand blueprint of Digital China construction, every advancement in AI data infrastructure is a solid step toward an intelligent future. As policy guidance aligns with technological implementation and the data foundation resonates with industry scenarios, KeenData, as a leader in the data industry, is writing a new chapter in AI data technology innovation and industrial digital and intelligent transformation.