[CLSTR]
API
< Back to situations

Monitor this situation.

[SITUATION] · [ACTIVE] · [TECHNOLOGY]

2 clusters · 5 sources · 6 days · First seen · Last updated

Semiconductor developments for AI efficiency

Overview

Recent technological advancements focus on addressing power and memory bottlenecks in artificial intelligence applications.

In Taiwan, United Microelectronics Corporation and AP Memory Technology Corporation announced a collaboration to develop power supply technologies for AI and High-Performance Computing. By combining Deep Trench Capacitor manufacturing with silicon capacitor design, the companies aim to place capacitors closer to computing chips to stabilize power delivery and reduce voltage fluctuations.

In Japan, AISIN and Toyota, in partnership with Tohoku University, unveiled a system-on-chip designed to improve energy efficiency in automotive AI. The 16nm FinFET process chip integrates STT-MRAM to store operating systems and AI model weights, reducing reliance on external memory. This integration reportedly allows for kernel startup speeds 38 times faster than traditional systems and utilizes a hybrid architecture to reduce AI model weight data by 53 percent.

Entities

Tohoku University · Arm · United Microelectronics Corporation · AP Memory Technology Corporation · Toyota

Timeline

  1. [TECHNOLOGY] 3 sources
    AISIN and Toyota develop integrated memory chip for AI efficiency

    AISIN and Tohoku University have developed a new system-on-chip with integrated STT-MRAM to accelerate AI processing and reduce power consumption in automotive applications.

  2. [TECHNOLOGY] 2 sources
    UMC and AP Memory collaborate on silicon capacitors for AI

    UMC and AP Memory are collaborating to integrate high-density silicon capacitors into advanced semiconductor packaging to stabilize power delivery for AI and high-performance computing applications.

Sources

aiThority.com · eetimes.jp · global-net.co.jp · stockhead.com.au · theregister.co.uk