Beijing Haidian Launches a 2.34 Million m² Physical Home for AI for Science

On August 22, the 2026 Conference on Scientific Intelligence opened at the Zhongguancun International Innovation Center in Beijing. The next day, Haidian District materialized its three-year-in-the-making AI4S innovation cluster—located at Jinyu Science and Technology Park in Xisanqi, coordinating 1.6 million square meters of industrial space plus 740,000 square meters of pilot-testing space, 2.34 million square meters in total.

This is not just an industrial park opening ceremony. It is the concentrated landing of policy accumulated over eight years.

Eight Years in the Making

Around 2018, CAS academician and Peking University professor E Weinan began systematically promoting the AI for Science concept, while mainstream AI attention was still on images, speech, and later large models. His deep potential molecular dynamics method (DPMD, using machine learning to fit first-principles-accuracy interatomic interactions) enabled simultaneous orders-of-magnitude jumps in scale and precision for molecular dynamics simulation.

After AlphaFold2's 2020 breakthrough, AI solving real scientific problems became a global consensus. In 2021, the Beijing Institute for Scientific Intelligence (AISI) was established around E Weinan's team—the world's first new-type research institute dedicated to AI4S. The inaugural Scientific Intelligence Summit was held in Haidian the following year. In 2022, NSFC's Interdisciplinary Sciences Division incorporated AI for Science into its next-generation AI major research plan; in 2023, the Ministry of Science and Technology launched dedicated AI-driven scientific research deployment.

In 2025, Beijing issued the "Beijing Action Plan for Accelerating High-Quality Development of AI-Empowered Scientific Research (2025–2027)"—the country's first dedicated local AI4S policy. By August 2026, the cluster was physically realized. Policy has grown more concrete year by year: from directional documents to专项行动 to physical space.

Key Numbers

• 1.6M + 740K m²: industrial + pilot-testing space (2.34M m² total)
• 2,500: conference attendees
• 50 academicians + 80 young experts
• 2 plenary + 14 parallel sessions + over 100 academic talks
• 16 first-batch Beijing AI4S exemplar cases, 12 of them in Haidian
• 2B + 8B yuan: commercialization fund + tech growth fund (linked AI4S special fund)
• 1,000+: annual target for high-level AI4S talent cultivated
• 10+: annual target for domestic and overseas teams attracted
• 15+: annual target for quality projects landed
• 10–100x: targeted throughput improvement in pilot validation
• ~3–5 years: timeline for building the "national AI4S main platform"

Three-Tier Layout: Core Sourcing + Pilot Validation + Scenario Expansion

Core sourcing zone at Jinyu Zhihui Center: integrated R&D, commercialization, and incubation; dry experiments on general-purpose scientific models and research agents; development of AI4S-specific chips—the innovation source and compute foundation.

Pilot validation at Zhigu Pilot Port: autonomous scientific discovery labs and AI-native materials labs. The goal is a closed loop where real experimental data flows back to models, with unified instrument interfaces, data formats, and evaluation standards, raising experimental throughput 10–100x. This is the most overlooked segment of the chain.

Scenario expansion linking Dongsheng Science Park, Haikai Park and others, focused on AI + new materials, AI + biopharma, and AI + quantum information, with central SOEs and leading enterprises opening real data and scenarios.

Why It Matters

1. The bottleneck has moved from compute to experiment. In his keynote "Science after AI for Science," E Weinan noted that AI can predict tens of thousands of candidate molecules in hours while a traditional lab validates only a few dozen per month; model accuracy records keep falling, but experimentally validated, industrialized results lag far behind paper output. The cluster's core move is not "another AI center" but physically instantiating the dry-wet closed loop—computation and experiment closed-loop within one physical park.

2. China's policy–capital–park trinity for AI4S has taken shape. In July, the White House OSTP released "Science: The New Golden Age," designating the Genesis plan as a flagship—doubling research productivity within 10 years with over $5 billion in federal investment. China follows the Haidian model: new-type research institutes (AISI) for foundational methodology, incubated companies (DP Technology, Zheyuan Technology, Huashen Zhiyao, Liangzhi Kaiwu) for productization, and the cluster physically packaging the whole pipeline.

3. Working industry cases already exist (beyond papers and leaderboards):

• Materials: DP Technology's Piloteye end-to-end automated battery design platform, shortening battery material R&D iteration cycles
• Biopharma: Zheyuan Technology's pancreatic cancer targeted drug went from molecular design to IND approval in under two years; Huashen Zhiyao compressed preclinical candidate development from 3–5 years to 10 months
• Quantum: Liangzhi Kaiwu's "Zhuifeng" ten-thousand-atom fast rearrangement algorithm and "Bianque" AI quantum error-correction decoder

This mainline stands opposite the US Genesis plan: both pursue national AI4S strategy, but via different paths—the US through federal coordination of supercomputing, models, and data infrastructure; China through physical park packaging plus policy funding plus incubated-company coordination. Which paradigm runs faster will be seen over the next 12–24 months in whoever raises experimental throughput more.

One Sentence

Scientific intelligence is not just another AI vertical—it is China's first attempt to fit the four links of "data–model–experiment–industry" into a single 2.34-million-square-meter physical park. Whether experimental throughput can rise 10–100x is the only verifiable metric for the next 24 months.