Phison Electronics (8299TT), a global leader in NAND flash controllers and storage solutions, today announced a collaboration with Intel to enable AI PCs to deploy larger, more capable AI applications locally. The collaboration combines Intel Core Ultra Series 3 processors with Phison's Pascari aiDAPTIV, a memory extension solution that unblocks memory-constrained systems to support larger Mixture-of-Experts (MoE) AI models, longer-running AI sessions and agentic AI workflows.

Today, AI PCs are moving beyond simple assistant applications toward more advanced local AI use cases. These solutions now support end users and businesses in document analysis, multistep workflow execution and private data protection while reducing reliance on cloud-only AI services. As these workloads can require larger AI models, persistent session state and heavy reliance on memory, many users are looking toward the next generation of client systems with the required memory capacity to support them.

aiDAPTIV addresses this challenge by extending effective AI working memory across system DRAM and high-performance, extreme-endurance NAND flash using Pascari aiDAPTIV Cache Memory. By reducing the DRAM required for certain local AI workloads and supporting runtime features such as KV cache reuse, aiDAPTIV helps larger AI workloads run locally on Intel AI PC platforms. In Phison testing, aiDAPTIV enabled a 26B-parameter model to run on a system with 16 GB of DRAM, compared with 32 GB of DRAM required without aiDAPTIV in the same test environment.

The collaboration focuses on enabling Phison's technology on Intel AI PC platforms powered by Intel Core Ultra processors, including support for the OpenVINO toolkit. Together, Phison and Intel are working to support ISV evaluations, technical demonstrations and optimized workloads for public performance claims.

"AI PCs are evolving into platforms for more sophisticated local AI workloads, including agentic applications and larger MoE models that place increasing demands on memory capacity and responsiveness," said KS Pua, CEO and Founder at Phison Electronics. "Through our collaboration with Intel, aiDAPTIV helps expand the necessary memory available to AI workloads on Intel AI PC platforms, allowing OEMs, developers and end users to run more capable AI applications locally while maintaining privacy and infrastructure efficiency."

At Computex, Phison will showcase aiDAPTIV-enabled demos on Intel AI PC platforms. Phison and Intel will demonstrate a local chat UI running a MoE AI model that would normally exceed available system memory, exhibiting how aiDAPTIV extends effective AI working memory using Pascari aiDAPTIV Cache Memory. Phison will also demonstrate a hybrid LLM routing application built on OpenClaw, an open-source AI agent framework, highlighting how larger MoE AI models can run locally with aiDAPTIV while reducing reliance on cloud token usage, with cloud routing available when needed for more complex requests.

Phison's booth will also feature demonstrations from AI software ecosystem partners - including Ollama, LLMWare, TurinTech, Intel AI Superbuilder and Intel AI Playground - showing aiDAPTIV in action across real-world local AI applications. The ecosystem showcase will also include hardware platform collaborations with ASUS, MSI, and Acer.

"Memory is a limiting factor in running many of the most capable models on client hardware," said Michael Chiang, Co-founder at Ollama. "Phison's aiDAPTIV approach on Intel AI PC platforms could let people run far larger models locally than their hardware normally allows."

"Enterprise GenAI is moving toward practical local workflows, including RAG, agents, and domain-specific models," said Darren Oberst, Co-founder and CTO at LLMWare. "Phison's aiDAPTIV approach is promising because it can help Intel AI PC client systems support larger models and more capable local AI applications while keeping data closer to the user."

"Working with Intel and Phison enables us to bring AI driven code optimization into practical, on device workflows, where performance, privacy, and cost discipline are critical," said Kee-Meng Tan, Chief Operating Officer at TurinTech AI. "By combining Artemis with Intel AI PCs and Phison's alternative memory approach, we can support larger and more capable local AI workloads without demanding ever increasing system memory."

"More users and businesses want to run AI locally—faster, more private and without the cost of sending everything to the cloud," said Jim Johnson, Senior Vice President and General Manager, Client Computing at Intel. "Our collaboration with Phison enables Intel AI PC platforms to support larger local AI workloads with simpler memory configurations, so customers can turn their own data into useful applications and real business value at a lower total cost."