“Having walked through the rain, I now want to hold an umbrella for others.”
Li Liancai, a programmer at iFlytek (SZ002230, ¥47.86/share, market cap ¥110.6B), is also a visually impaired individual navigating the digital world. In April 2025, he developed three AI agents—an Information Accessibility Advisor, a Traditional Chinese Massage Consultant, and a Blind Mentor Assistant—during iFlytek’s Spark AI Agent Challenge, aiming to use AI as a key to digital inclusion. This is just one example of AI for social good.
As AI large language models (LLMs) rapidly advance, how are companies addressing ESG (Environmental, Social, and Governance) concerns? The Daily Economic News Brand Value Research Institute analyzed DeepSeek AI-related stocks and found that 57% (24 out of 42 listed firms) disclosed ESG reports in 2024.
Shi Yicheng, Deputy Director of the Central University of Finance and Economics’ Green Finance International Institute, noted:
“AI companies leveraging green data centers can significantly reduce carbon footprints—but data authenticity, accuracy, and traceability are critical.”
AI Firms Ramp Up Green Energy Adoption
With AI models growing exponentially in size (GPT-4 has ~1.8T parameters vs. GPT-3’s 175B), energy consumption has surged. Stanford AI Research estimates that training GPT-3 once consumed 1,287 MWh—equivalent to 3,000 Tesla EVs driving 200,000 miles each.
Among China’s top 10 DeepSeek AI stocks by market cap, 9 disclosed ESG reports, with 3 reporting Scope 3 emissions (indirect supply chain emissions). Key green initiatives include:
-
iFlytek: Since April 2024, 90% of its data center power comes from wind/solar energy.
-
Kingsoft Office (SH688111, ¥270.75/share, ¥125.4B market cap): Installed 2,640 rooftop solar panels, generating 700,000–800,000 kWh/year, saving 810 MWh annually.
-
Sugon (SH603019, ¥68.49/share, ¥100.2B market cap): Developed liquid-cooled storage systems, cutting 4.5M kWh/year (~1,500 tons of CO₂).
Shi Yicheng warns:
“Green power sourcing must be verifiable—companies must ensure accurate carbon accounting to uphold ESG credibility.”
Compute Power Upgrades: Overcoming AI Bottlenecks
Despite China’s rapid progress in AI infrastructure, challenges remain:
-
High-end chip dependency
-
Energy & cost pressures
-
Software ecosystem gaps
Leading firms are innovating:
-
Sugon’s “3D Computing” strategy integrates hardware, software, and ecosystem development.
-
Inspur (SZ000977, ¥48.81/share, ¥71.86B market cap) launched the YuanBrain Gen8 AI platform, supporting multi-chip architectures and breaking global benchmarks.
Shi Yicheng emphasizes:
“Optimizing algorithms and improving training data quality can reduce compute needs while boosting AI efficiency.”
Bridging the AI Divide: Can Tech Be Truly Inclusive?
AI adoption remains uneven—resource-rich firms gain competitive edges, while others risk marginalization. Companies addressing this gap include:
-
iFlytek: AI-powered education & healthcare solutions, like smart diagnostics for rural clinics.
-
360 Security (SH601360, ¥9.98/share, ¥69.86B market cap): Offers SaaS-based AI tools for SMEs and government agencies.
Shi Yicheng concludes:
“AI must be democratized—lowering costs and barriers ensures fair access, fostering equitable competition.”