ADATA's TRUSTA brand launched its AI Scaler Extended Memory Solution, a hardware-software integration that breaks GPU memory limitations. The solution includes the AI Scaler Toolkit, extending model deployment beyond GPU-only architectures to system memory and SSDs. This innovation helps enterprises reduce AI deployment costs by over 50% and marks ADATA's evolution from memory provider to comprehensive solution provider.

Running large language models typically requires expensive high-end GPUs, creating deployment barriers. The TRUSTA solution addresses this by allowing model inference optimized for multiple GPUs to run on a single GPU with expanded system memory. It also dynamically allocates computing resources across GPU, DRAM, and SSD for more flexible scaling.

The AI Scaler Toolkit functions as a free, open-source platform supporting mainstream AI models including Llama, Qwen, and Mistral. TRUSTA will showcase its 15.36 TB TD7P51 ECO SSD at COMPUTEX, where the solution received the 2026 Best Choice Award, enabling enterprises to build efficient on-premises AI infrastructure.