The Time Series Database Tech Innovation Summit 2025 brought together leading minds from academia, industry, and the open-source community to explore the evolving convergence of databases and AI. Held on July 5, 2025, in Beijing, the summit was co-organized by the School of Software at Tsinghua University and Timecho.
Centered around the theme “Next Frontier: DB + AI”, the event featured over 30 distinguished speakers, including academicians of the Chinese Academy of Engineering, professors from renowned universities such as Tsinghua University, Renmin University of China, University of Science and Technology Beijing, and China University of Petroleum, as well as senior technical leaders from enterprises such as COMAC, XYZ Storage, China Resources Power, and Siemens.
The summit drew more than 400 in-person attendees and over 140,000 online participants from across the industrial internet and infrastructure software community. Focusing on the core technologies and industrial applications of Apache IoTDB*, the summit explored how AI and database integration can accelerate intelligent upgrades in the industrial IoT landscape.
Opening Plenary: Vision from Academia and Industry
Keynote Remarks
The main forum opened with keynote speeches from Prof. Jiaguang Sun, Academician of the Chinese Academy of Engineering, and Prof. Xiaohui Yu, President of the China Academy of Information and Communications Technology. Both emphasized the strategic role of foundational data infrastructure in enabling industrial intelligence.
From Data Collection to Intelligent Utilization
In his talk “The AI Era: From Data Acquisition to Intelligent Utilization,” Prof. Jianmin Wang, Dean of the School of Software at Tsinghua University, described how AI is reshaping the full software stack and driving software engineering into a new 3.0 phase. He noted that as industrial systems become increasingly intelligent and connected, the entire data lifecycle — from collection to analysis — must be rearchitected.
Against this backdrop, Apache IoTDB is positioned not merely as a time series database system, but as a comprehensive “intelligent edge-cloud architecture” purpose-built for industrial digitalization — offering a solid technical foundation to power the next leap in smart manufacturing.
Apache IoTDB’s Breakthroughs from 2023–2025
Dr. Xiangdong Huang, Chair of the Apache IoTDB Project Management Committee (PMC), shared key breakthroughs in four strategic dimensions:
Technology innovation, including engine performance and storage improvements;
International adoption, with IoTDB now used by thousands of organizations;
Ecosystem expansion, with wide integration across platforms and languages;
Benchmark leadership, including record-setting results in TPCx-IoT and benchANT.
These milestones reflect the project’s steady momentum in advancing both the state-of-the-art in time series data technology and its industrial adoption across sectors.
“Great progress comes from steady steps,” he said. “We aim to lead not only in technology, but in real-world impact.”
Unveiling IoTDB 2.0: A Next-Gen Data Intelligence Foundation
Dr. Jialin Qiao, CTO of Timecho and Apache IoTDB PMC member, introduced IoTDB 2.0, built to meet the data management demands of the AI era. The new version includes:
A high-performance time series engine;
A federated query layer for multi-source data;
A native analytics module supporting forecasting and anomaly detection.
These components form a modular, intelligent data foundation designed to support industrial monitoring, equipment maintenance, predictive diagnostics, and more.
Introducing Timer 3.0: A Foundational Time Series Model
Prof. Mingsheng Long, tenured associate professor at the School of Software, Tsinghua University, announced the release of Timer 3.0 (“Sundial”), a large-scale foundation model for time series forecasting. Designed for out-of-the-box usability, it can be directly invoked within IoTDB, while also functioning independently.
Timer 3.0 introduces a hybrid autoregressive-generative architecture, optimized for efficient learning of time series characteristics. It is built on the first trillion-point pretraining dataset in the domain, and features:
Distributed pretraining objectives across broad application types;
Efficient time series feature learning;
Multi-target generative prediction.
It has demonstrated strong performance across several international forecasting benchmarks, enabling wide-ranging industrial and academic use.
Industrial Highlights: COMAC and XYZ Storage
On the industry front, industry leaders presented compelling real-world deployments.
Dr. Can Feng, Chief Flight Test Architect at COMAC Flight Test Center, presented how IoTDB was applied in a device-edge-cloud framework for flight testing. The system reduced airborne data volume by 50%, increased upload speed by over 10×, and supported over 100 concurrent tasks. It also enabled real-time edge-cloud analytics for outfield test scenarios — accelerating the adoption of a new generation of cloud-based flight monitoring systems.
Dr. Yan Wang, Manager of Energy Storage Operations at XYZ Storage, shared how IoTDB powers the AIOPS-2000 intelligent operations platform. The platform manages full-resolution time series data from 100MW-scale storage systems, cutting data collection and storage costs by 90%, and cloud resource consumption by more than 90%. It currently supports 39 power station projects, managing over 5.7 GWh of assets including standalone storage stations, integrated renewable-storage deployments, and hybrid frequency regulation projects.
Panel Discussion: From Real-World Demands to Database Innovation
A high-level panel titled “In the AI Era: Evolving Database Technologies from Real-World Demands” brought together experts from academia and industry:
Academician Prof. Jiaguang Sun, Chinese Academy of Engineering
Prof. Xiaoyong Du, Renmin University of China, Director of the National Key Lab for Data and Knowledge Engineering
Prof. Kaixiang Peng, Vice Dean of the School of Automation, University of Science and Technology Beijing
Weimin Guo, Vice President, China Resources Power Research Institute
Prof. Chunlei Wu, Vice Dean, School of Software, China University of Petroleum
The panelists examined the critical role of time series data in digital transformation across oil & gas, energy, and steel sectors. They called for more intelligent, real-time, and domain-adaptive data management technologies to meet the growing demands of industrial digitization in the age of AI.
From Research to Practice: Pushing the Boundaries of Time Series Technology
Technical Deep Dive Forum: Fusion, Intelligence, and Openness
In this forum, the Timecho R&D team provided in-depth explanations of recent breakthroughs in Apache IoTDB, including:
Tree mode / Table mode fusion: Enabling flexible schema definition via hybrid SQL dialects;
MCP Server: Supporting multi-model, concurrent processing to improve throughput and scheduling;
Stream processing framework: Delivering low-latency analytics across edge-cloud environments;
Cross-platform TsFile: Enabling data interoperability across platforms like Python, Java, Rust;
AINode: Providing LLM-based fine-tuning and integration with pre-trained models;
System tuning & observability: Including I/O monitoring, memory control, and load-aware compaction;
Ecosystem expansion: With integrations into platforms such as Grafana, Prometheus, ThingsBoard, and ctrlX.
These developments reflect IoTDB’s ambition to become not just a database, but a composable infrastructure for AI + time series workloads.
In the live demo zone, IoTDB’s latest capabilities in AI services, time series modeling, and tabular mode were showcased through real-time product demonstrations. Developers had the chance to experience the system’s features firsthand, from visual dashboards to interactive model tuning.
The exhibition area also featured partner booths, interactive installations, and even a game zone, creating a lively and engaging environment that blended technology, collaboration, and fun.
User Case Studies Forum: IoTDB in the Field
This forum highlighted diverse real-world deployments of IoTDB across key verticals such as metallurgy and steel, energy and power, smart manufacturing, connected vehicles, and public infrastructure.
Industry experts and technical leaders shared practical insights and engineering best practices, including:
Qiang Zheng, Senior Architect at CISDI Information Technology, discussed large-scale monitoring in steel production environments using IoTDB to process millions of time series points per second.
Mao Zhao, General Manager of the Data Intelligence Center at Merit Data, shared applications in intelligent manufacturing, including equipment health monitoring and root-cause analytics.
Quan Wang, head of storage systems at Shanghai Electric, presented IoTDB's role in energy dispatch systems and predictive maintenance.
Dr. Chao Wang, Director of AI & Edge at Siemens China, discussed how IoTDB integrates into AI inference pipelines at the edge.
Tao Deng, Deputy R&D Director at NeuCloud, introduced multi-tenant platform use cases based on IoTDB.
Stephen Lawrence, Principal Software Engineer at Renesas Electronics, showed how IoTDB was adopted in the Central Data Service Playground (CDSP) of the Connected Vehicle Systems Alliance (COVESA) to support data standardization and interoperability in vehicle edge environments.
Trevor Bloch, Founder & CTO of VROC.AI, highlighted IoTDB’s low resource footprint, simple architecture, and fast data I/O — features that align closely with the demands of scalable, cost-effective AIoT platforms.
Academic Research Forum: Frontiers of AI + Time Series
This forum gathered a group of distinguished scholars to share breakthroughs in time series research and AI-enhanced data management. Presenters included:
Prof. Shangguang Wang (BUPT): Time series data compression and communication-aware modeling;
Prof. Li Wang (TYUT): Deep learning architectures for noisy industrial data streams;
Dr. Xiaoou Ding (HIT): Trustworthiness and uncertainty estimation in time series prediction;
Dr. Ruiyuan Li (Chongqing University): Large-scale forecasting with sparse supervision;
Dr. Xiaoqi Duan (Guizhou University): High-dimensional time series quality governance;
Dr. Yanning Sun (Shanghai University): Spatiotemporal dynamics modeling;
Dr. Li Lin (Southeast University): Dynamic sensing systems and real-time edge learning.
These discussions reflected growing alignment between academic exploration and real-world data system challenges.
Looking Ahead: Unlocking the Future of DB + AI
From frontier research to industrial deployment, and from open-source innovation to cross-domain collaboration, the Time Series Database Tech Innovation Summit 2025 offered a comprehensive look at the technologies shaping the future of data infrastructure.
Looking forward, the community will continue to build upon Apache IoTDB, driving deeper integration between time series data systems and AI-native architectures — enabling faster analytics, smarter automation, and more responsive industrial systems.
The next chapter — where time series data and AI deeply converge — promises to unlock a smarter, more agile future for industrial intelligence.
* Apache IoTDB, Apache IoTDB Logo, Apache TsFile, and Apache TsFile Logo are registered trademarks of The Apache Software Foundation.
