AI investment & startup in China Brief — 2026-09-16
Top Stories
1. ByteDance’s AI drug-discovery spin-off Anew Labs raises $290 million at $1.5 billion valuation
- Source: Reuters · September 16, 2026
- Summary: ByteDance’s Shanghai-based AI drug-discovery unit Anew Labs has completed a $290 million external financing round following its spin-off from the parent company. HSG, IDG Capital and Hillhouse led the round, with 5Y Capital as co-lead and participation from several Chinese venture and strategic investors. ByteDance will retain a 56% stake after the financing, while Anew Labs reaches a reported $1.5 billion valuation.
- Why It Matters: The transaction demonstrates continued Chinese investor appetite for AI applied to high-value scientific domains rather than only general-purpose models. The spin-off structure also gives ByteDance a way to finance specialized AI businesses while allowing them greater operational independence.
- URL: https://www.reuters.com/legal/transactional/bytedance-completes-290-million-fundraising-ai-drug-unit-after-its-spin-off-2026-09-16/
2. China’s new recursive-AI startup Apex Intelligence raises nearly RMB 400 million
- Source: China National Radio · September 16, 2026
- Summary: Beijing-based Apex Intelligence, founded in June 2026 by Tsinghua University AI researcher Chen Yongchao, announced nearly RMB 400 million in angel and angel-plus financing. Investors include IDG Capital, Xinglian Capital, XtalPi, Zhongguancun Science City Fund, Shenzhen Capital Group and Shanghai Future Industries Fund. The company is developing foundation models designed around Recursive Self-Improvement (RSI), with research automation as the initial focus.
- Why It Matters: Capital is beginning to move beyond conventional LLM scaling toward systems intended to improve their own research and reasoning processes. The combination of leading research talent, substantial early financing and state-linked capital indicates growing interest in RSI as a potential next-generation AI infrastructure layer.
- URL: https://www.cnr.cn/mspd/msgc/qwfb/20260916/t20260916_527815351.shtml
3. Deep Material raises nearly RMB 100 million to industrialize AI for materials science
- Source: 36Kr · September 16, 2026
- Summary: Suzhou-based Deep Material completed an A+ round of nearly RMB 100 million, led by Source Code Capital and Jinyu Maowu, with participation from Yichen Capital and others. The company combines AI models, proprietary materials data and automated high-throughput experimentation to develop high-performance metal materials. It reports nearly 20 new materials, with several already entering mass production.
- Why It Matters: The financing reflects a shift in AI-for-Science investment from model capability toward closed-loop industrial execution. Proprietary experimental data, automated laboratories and real-world manufacturing feedback can become strategic assets that are harder to replicate than generic AI models.
- URL: https://www.36kr.com/p/3985466314587140
4. Silicon-capacitor startup Senmaru Electronics raises RMB 100 million for AI-chip infrastructure
- Source: 36Kr · September 16, 2026
- Summary: Suzhou-based Senmaru Electronics completed a roughly RMB 100 million Series A led by an optical-module industry investor and Yaotu Capital, with Shunrong Capital participating. The wafer-level passive-component IDM is developing silicon capacitors and integrated passive devices for high-end optical modules and power delivery in AI chips. The new funding will support technology development, capacity expansion and overseas market growth.
- Why It Matters: AI infrastructure investment is extending beyond GPUs into power delivery and high-speed interconnect components. As AI data-center density increases, specialized semiconductor components that address power and networking bottlenecks could become increasingly important investment targets.
- URL: https://eu.36kr.com/zh/p/3985710325038080
5. Apex Intelligence’s rapid financing highlights China’s emerging RSI startup wave
- Source: 36Kr · September 16, 2026
- Summary: Apex Intelligence has reportedly completed two billion-yuan-scale financing rounds within roughly two months of its launch. The company is pursuing a foundation-model approach to Recursive Self-Improvement rather than focusing solely on an agent application layer, using research trajectories as training signals for iterative model improvement. Its near-term plans include building model-training infrastructure, expanding research capabilities and developing a dedicated research-oriented foundation model.
- Why It Matters: The financing provides an early signal that Chinese venture capital is willing to fund highly speculative model-layer bets when they are backed by experienced research teams. If RSI systems can reliably turn research activity into better training data and capabilities, the resulting feedback loop could reshape the economics of AI model development.
- URL: https://eu.36kr.com/zh/p/3985658685052800
6. China’s AI materials sector increasingly competes on proprietary data and experimental infrastructure
- Source: 36Kr · September 16, 2026
- Summary: Deep Material’s latest financing is being directed toward its M-Loop intelligence system, automated M-Lab experimentation platform, proprietary M-Data dataset and distributed autonomous laboratory network. The company says its platform integrates AI reasoning with physical experiments and has accumulated more than 100,000 materials data records. Its business model emphasizes taking AI-generated materials through validation and industrial production rather than stopping at simulation.
- Why It Matters: This points to a broader investment thesis around vertically integrated AI: proprietary data generation, physical experimentation and industrial deployment can create stronger defensibility than model access alone. For AI investors, the scarce asset may increasingly be the feedback loop connecting models to the physical economy.
- URL: https://www.36kr.com/p/3985466314587140
Investment Signals
1. Capital is moving deeper into AI’s physical and scientific stack. Today’s disclosed transactions span drug discovery, materials science and semiconductor infrastructure, showing investment interest extending from foundation models into domains where AI interacts with proprietary data, laboratories and industrial systems.
2. RSI is emerging as a new speculative investment category. Apex Intelligence’s rapid financing suggests that recursive self-improvement is beginning to attract dedicated startup capital in China. The investment case depends heavily on whether autonomous research can generate sufficiently valuable training signals to create a measurable capability flywheel.
3. Proprietary data generation is becoming an investment differentiator. Deep Material illustrates a broader pattern: AI startups increasingly need mechanisms to create high-quality domain data rather than relying exclusively on public datasets or general-purpose foundation models.
4. Strategic and state-linked capital remain important. Today’s deals include participation from major venture firms, corporate investors and government-backed funds, particularly in areas connected to biotechnology, scientific infrastructure and strategic AI capabilities.
Bottom Line
September 16’s disclosed Chinese AI financing activity points toward a market increasingly willing to fund AI systems connected to science, physical infrastructure and recursive improvement. The common thread is a move away from generic AI applications toward companies attempting to own a differentiated combination of models + proprietary data + domain expertise + real-world execution.