China Unveils Analogue AI Chip

China has revealed a promising new advancement in artificial intelligence hardware that could dramatically reduce the energy footprint of AI data centres.

Researchers at Peking University and other institutions are working on an analogue AI chip that uses in-memory computing techniques to handle key AI tasks with much lower power consumption than conventional digital processors — potentially cutting electricity use by more than 99 % for certain workloads.

Unlike traditional chips that move data repeatedly between memory and processors (which consumes a lot of energy), the analogue design performs calculations directly inside memory structures. This results in faster processing and dramatically improved efficiency, particularly for tasks such as inference — the part of AI that responds to user queries once a model is trained.

This development comes as global data-centre energy demand continues to rise sharply due to expanding AI workloads. AI datacentres already consume massive amounts of electricity and require advanced cooling systems to dissipate the heat created by thousands of processors running simultaneously.

China’s move also fits into a broader strategy to strengthen its domestic chip ecosystem and reduce reliance on foreign AI processors — a goal underscored by recent government directives banning foreign AI chips from state-funded data-centre projects.

While the analogue chip is still in research and not yet commercially deployed, analysts say it represents a possible future path toward more sustainable AI computing hardware.

If the technology can be scaled and manufactured reliably, it could help cloud service providers lower operational costs, reduce carbon emissions linked to AI operations, and support broader growth in energy-efficient AI systems across industries.

Key Takeaways

• Chinese researchers are developing an analogue AI chip designed to drastically cut data-centre energy use.

• Early research suggests the design could outperform conventional digital chips in speed and power efficiency.

• The breakthrough aligns with China’s push to strengthen domestic chip capabilities and tackles growing energy demand in AI infrastructure.

Adapted From

First Post