From Cell Therapy to Steel AI: Chongqing Announces Top 10 Scientific Breakthroughs for 2025
Chongqing's Academy of Science and Technology for Development released its 2025 Top 10 breakthroughs on Jan. 14, highlighting advances from cell therapy to industrial AI. Since launching in 2023, the list has become a quick read on where research is turning into deployment across medicine, semiconductors, infrastructure, and manufacturing.
The selection signals where regulatory momentum, domestic capability building, and near-term applications are lining up. It also reflects the city's push for applied science, with hubs such as the Mingyue Lake Laboratory in Liangjiang New Area driving lab-to-field progress.
Why this list matters for researchers and R&D leaders
- Regulatory green lights are accelerating, especially in cell therapy and medical diagnostics.
- Domestic tech stacks are maturing in semiconductors, imaging, and industrial AI.
- Data-rich methods (immunomics, vision models) are moving into operational settings.
- Engineering methods are delivering measurable gains in build time, precision, and safety.
China's first pediatric CAR-T approval
Chongqing Precision Biotech's Puzolcabtagene Autoleucel Injection secured market approval in November 2025 for pediatric and adolescent B-cell acute lymphoblastic leukemia. It addresses high costs of imported therapies and limited effectiveness of chemo-radiotherapy for relapsed or refractory cases.
For clinicians and translational teams, this is a clear marker: CAR-T pipelines for younger patients are viable in China, with reimbursement and long-term cohort data likely to follow. For context on CAR-T mechanisms, see NCI's overview.
Immunotherapy response and resistance decoded
A team led by Academician Zhang Zemin at Chongqing Medical University introduced a cross-tissue cellular module framework, mapped five immune microenvironment subtypes in non-small cell lung cancer, and proposed recurrence biomarkers with diagnostic value. Results were published in Cell (March 2025) and Nature (May 2025).
For oncology researchers, the work refines patient stratification and offers a pathway to tighter, biomarker-guided trial design. Expect follow-up on prospective validation and companion diagnostics.
Countering citrus greening (HLB) with resistance genes
Collaborators from the Citrus Research Institute of Southwest University and the Chinese Academy of Sciences identified two key resistance genes, MYC2 and PUB21DN, tied to Huanglongbing (HLB). Given HLB's multi-country spread and annual losses exceeding USD 10 billion, these targets open routes for breeding resistant or tolerant cultivars.
Plant biologists and breeders get concrete genetic levers for cultivar development and field trials. Background on the disease is available via the USDA's citrus greening resource.
Ultra-high-speed image sensor for 7-nm inspection
The 44th Research Institute of China Electronics Technology Group developed a fully homegrown ultra-high-speed image sensor now used in domestic 7-nm semiconductor inspection tools. It supports real-time detection of nanoscale surface defects and contamination, closing a key local capability gap.
For fabs and metrology teams, this means shorter feedback loops, better yield diagnostics, and less reliance on imported inspection systems.
Digital in-situ arch forming for long-span bridges
Chongqing Jiaotong University's team, led by Professor Zhou Jianting, built a digital system for in-situ arch forming on long-span arch bridges. It reduces pre-assembly site area by 30%, halves high-altitude control time, and maintains 2-mm closure accuracy on 500-meter-class spans.
Structural engineers can expect safer operations, tighter tolerances, and reduced disruption at large build sites.
Industrial vision AI for the full steelmaking process
CISDI Information Technology (Chongqing) released the CISIDigital AI "Jinjing" model, a 100-million-parameter industrial vision model spanning the entire steelmaking process. It's built to operate in heat, glare, and heavy water mist, cutting repetitive frontline tasks and improving process stability.
For plant teams, this points to broader computer-vision coverage at scale and better defect detection under harsh conditions. If you're upskilling teams for AI deployment in operations, see curated options here: Latest AI courses.
How the Top 10 were selected
Universities, institutes, and companies submitted over 100 advances. Experts shortlisted 27, followed by a vote from leading scientists, including academicians from the Chinese Academy of Sciences and the Chinese Academy of Engineering.
Final picks were judged on innovation, real-world application, and overall impact-favoring work that is already influencing clinical, industrial, or engineering practice.
What to watch next
- CAR-T: long-term outcomes, safety tracking in younger cohorts, and payer decisions.
- Immunotherapy biomarkers: multi-center validation and integration into trial protocols.
- HLB resistance: greenhouse-to-field translation and trait stacking in commercial varieties.
- Semiconductor inspection: deployment rates across domestic fabs and cross-tool compatibility.
- Bridge systems: replication on additional 500-m class projects and code/spec updates.
- Steel AI: model drift management, dataset governance, and ROI in continuous operations.
The through line is clear: translational focus and operational readiness. For teams planning 2025-2026 roadmaps, these signals are useful for grant targeting, partnerships, and trial design.
Your membership also unlocks:
