A joint research project between US-based TetraMem Inc. and South Korean semiconductor giant SK hynix has yielded a new chip design that promises to dramatically cut the energy and heat generated by artificial intelligence workloads. The findings, published in the journal Advanced Intelligent Systems and selected as its cover feature, detail a memristor-based AI System-on-Chip (SoC) that performs computations directly where data is stored, rather than shuttling it between separate processor and memory units.
This approach, known as Analog In-Memory Computing (A-IMC), targets a fundamental inefficiency in modern AI systems. As foundation models grow from billions to trillions of parameters, the constant movement of data between chips and memory has become the largest drain on power, contributing to soaring electricity costs and thermal management challenges in data centres. The new SoC integrates emerging memory devices, circuit design, and AI architecture into a practical platform that reduces this data movement, improving system-level performance and energy efficiency.
For UK households and businesses, the development signals potential long-term relief from rising energy costs tied to AI infrastructure. Data centres already account for a significant share of UK electricity consumption, and with AI adoption accelerating, that burden is expected to grow. Technologies that make AI computing more energy-efficient could help contain operational costs for cloud services and reduce the environmental impact of digital services that millions of Britons rely on daily.
The Bank of England has been monitoring the economic effects of AI adoption, noting that while productivity gains are possible, the associated energy demands could fuel inflation if not managed. The FTSE 100 has seen increasing exposure to semiconductor and AI-related stocks, though UK-listed firms are more heavily weighted toward financials and commodities. A shift toward more efficient AI hardware could nonetheless benefit UK-listed technology companies and data centre operators by lowering their capital expenditure on cooling and power infrastructure.
For UK savers and mortgage holders, the direct impact is indirect but meaningful. If AI energy efficiency improves, it could ease upward pressure on electricity prices, which have been a key driver of household inflation. Investors with exposure to technology funds or UK-listed infrastructure trusts may see longer-term gains from reduced operational costs in the AI supply chain. However, as with all emerging technologies, outcomes remain uncertain and readers should consult a qualified financial adviser before making investment decisions.