Researchers on the Korea Superior Institute of Science and Know-how (KAIST) have developed energy-efficient NPU know-how that demonstrates substantial efficiency enhancements in laboratory testing.
Their specialised AI chip ran AI fashions 60% quicker whereas utilizing 44% much less electrical energy than the graphics playing cards at the moment powering most AI programs, primarily based on outcomes from managed experiments.
To place it merely, the analysis, led by Professor Jongse Park from KAIST’s College of Computing in collaboration with HyperAccel Inc., addresses one of the vital urgent challenges in fashionable AI infrastructure: the large vitality and {hardware} necessities of large-scale generative AI fashions.
Present programs resembling OpenAI’s ChatGPT-4 and Google’s Gemini 2.5 demand not solely excessive reminiscence bandwidth but additionally substantial reminiscence capability, driving corporations like Microsoft and Google to buy a whole bunch of 1000’s of NVIDIA GPUs.
The reminiscence bottleneck problem
The core innovation lies within the crew’s strategy to fixing reminiscence bottleneck points that plague present AI infrastructure. Their energy-efficient NPU know-how focuses on “light-weight” the inference course of whereas minimising accuracy loss—a essential steadiness that has confirmed difficult for earlier options.
PhD scholar Minsu Kim and Dr Seongmin Hong from HyperAccel Inc., serving as co-first authors, introduced their findings on the 2025 Worldwide Symposium on Pc Structure (ISCA 2025) in Tokyo. The analysis paper, titled “Oaken: Fast and Efficient LLM Serving with Online-Offline Hybrid KV Cache Quantization,” particulars their complete strategy to the issue.
The know-how centres on KV cache quantisation, which the researchers determine as accounting for most reminiscence utilization in generative AI programs. By optimising this part, the crew allows the identical stage of AI infrastructure efficiency utilizing fewer NPU gadgets in comparison with conventional GPU-based programs.
Technical innovation and structure
The KAIST crew’s energy-efficient NPU know-how employs a three-pronged quantisation algorithm: threshold-based online-offline hybrid quantisation, group-shift quantisation, and fused dense-and-sparse encoding. This strategy permits the system to combine with present reminiscence interfaces with out requiring modifications to operational logic in present NPU architectures.
The {hardware} structure incorporates page-level reminiscence administration methods for environment friendly utilisation of restricted reminiscence bandwidth and capability. Moreover, the crew launched new encoding methods particularly optimised for quantised KV cache, addressing the distinctive necessities of their strategy.
“This analysis, via joint work with HyperAccel Inc., discovered an answer in generative AI inference light-weighting algorithms and succeeded in creating a core NPU know-how that may resolve the reminiscence downside,” Professor Park defined.
“By way of this know-how, we applied an NPU with over 60% improved efficiency in comparison with the newest GPUs by combining quantisation methods that scale back reminiscence necessities whereas sustaining inference accuracy.”
Sustainability implications
The environmental influence of AI infrastructure has develop into a rising concern as generative AI adoption accelerates. The energy-efficient NPU know-how developed by KAIST presents a possible path towards extra sustainable AI operations.
With 44% decrease energy consumption in comparison with present GPU options, widespread adoption may considerably scale back the carbon footprint of AI cloud providers. Nonetheless, the know-how’s real-world influence will rely upon a number of elements, together with manufacturing scalability, cost-effectiveness, and trade adoption charges.
The researchers acknowledge that their resolution represents a big step ahead, however widespread implementation would require continued growth and trade collaboration.
Trade context and future outlook
The timing of this energy-efficient NPU know-how breakthrough is especially related as AI corporations face rising strain to steadiness efficiency with sustainability. The present GPU-dominated market has created provide chain constraints and elevated prices, making various options more and more enticing.
Professor Park famous that the know-how “has demonstrated the opportunity of implementing high-performance, low-power infrastructure specialised for generative AI, and is predicted to play a key position not solely in AI cloud knowledge centres but additionally within the AI transformation (AX) setting represented by dynamic, executable AI resembling agentic AI.”
The analysis represents a big step towards extra sustainable AI infrastructure, however its final influence will probably be decided by how successfully it may be scaled and deployed in industrial environments. Because the AI trade continues to grapple with vitality consumption issues, improvements like KAIST’s energy-efficient NPU know-how provide hope for a extra sustainable future in synthetic intelligence computing.
(Photograph by Korea Superior Institute of Science and Know-how)
See additionally: The 6 practices that guarantee extra sustainable knowledge centre operations
Need to be taught extra about cybersecurity and the cloud from trade leaders? Try Cyber Security & Cloud Expo going down in Amsterdam, California, and London.
Discover different upcoming enterprise know-how occasions and webinars powered by TechForge here.
