This New Memory Solution Breaks AI’s GPU Bottleneck, Slashing Enterprise Deployment Costs by Over 50% by Using DRAM and SSDs

ADATA’s Trusta launched the AI Scaler Extended Memory Solution at COMPUTEX, built around the AI Scaler Toolkit.

The system spreads AI workloads across GPU memory, system DRAM, and high-speed SSDs to ease VRAM bottlenecks.

It is designed to support mainstream models and agent workflows while reducing inference and fine-tuning deployment costs by more than 50%.

The move could help enterprises run on-prem AI more efficiently by relying less on scarce GPU memory.