The Global Memory Innovation Forum 2026 (GMIF 2026) is bringing memory manufacturers, storage specialists, AI-infrastructure companies, researchers and investors to Shenzhen on September 22–23, 2026.
Held at the Renaissance Shenzhen Bay Hotel, the fifth edition is co-hosted by the Shenzhen Memory Industry Association and Peking University’s School of Integrated Circuits. Its theme, “The Future Built on AI Memory and Storage,” reflects the increasing importance of storage performance as AI systems move from model training towards large-scale inference and commercial deployment.
The invitation-only AI Industry and Investment Forum takes place on September 22. The main summit, exhibition, awards programme and business-networking activities are scheduled for September 23, according to the official GMIF 2026 programme.
Samsung, Sandisk and Solidigm join the speaker lineup
GMIF 2026 has announced speakers from more than 18 companies, financial institutions and academic organisations across the semiconductor and AI infrastructure ecosystem.
The confirmed programme includes:
- Kevin Yoon, vice-president and chief technology officer of Samsung Electronics’ Memory Business
- Zhang Dan, senior director of product marketing at Sandisk
- Ni Jinfeng, vice-president of Asia-Pacific sales at Solidigm
- John Xavier Lionel, head of global storage business at Arm
- Zhou Tao, general manager of Lenovo China’s server business
- Huang Shide, assistant vice-president at Silicon Motion
- Chen Lusha, Asia-Pacific general manager at Dify
- Cai Yimao, dean of Peking University’s School of Integrated Circuits
- Yan Zhitian, executive director at Morgan Stanley
Other participating organisations named by the organiser include BIWIN, Paratera, NEXWISE, Infplane, SYNCORE, Genstoraige and OKN Technology.
The presence of memory producers, controller developers, server companies, AI-platform providers and investment specialists gives the forum a wider focus than a conventional semiconductor product conference. Discussions will examine how memory and storage requirements are changing across AI computing, enterprise infrastructure and edge applications.
AI inference moves storage higher up the infrastructure agenda
The event’s programme places particular emphasis on AI inference, enterprise storage and coordination between cloud, edge and endpoint systems.
Training a large AI model requires substantial computing power and high-bandwidth memory. Once deployed, however, the model may need to respond to requests from thousands or millions of users. This creates continuing demand for fast access to model data, vector databases, retrieval systems and previously generated information.
GMIF describes this development through the idea of the token economy. In this context, “tokens” are the units of text and other data processed or generated by AI models—not cryptocurrency tokens. The commercial question is how efficiently AI infrastructure can store, retrieve and process the data required to generate them.
The official summit overview identifies KV cache, vector search, retrieval-augmented generation, long-context models and data accumulation as important parts of this emerging storage market.
For technology buyers, the relevant performance measures extend beyond raw capacity. Latency, throughput, energy consumption, endurance, software compatibility and total operating cost can all affect the efficiency of an AI deployment.
Cloud, edge and endpoint storage form a second major theme
GMIF 2026 will also consider how AI workloads are spreading from centralised data centres to industrial equipment, vehicles, personal computers and other endpoint devices.
This cloud-edge-endpoint structure creates different storage requirements. A data centre may prioritise high-capacity enterprise SSDs and rapid access to shared data, while an edge system may require lower power consumption, compact hardware and reliable operation with limited connectivity.
The technologies highlighted in the programme include enterprise SSDs, high-performance memory, CXL, system-on-chip platforms and endpoint storage. Their integration will influence how quickly AI applications can respond, how much data can be processed locally and how often information must be transferred to the cloud.
This makes the forum relevant to server manufacturers, data-centre operators and storage buyers as well as companies developing intelligent vehicles, industrial systems, robotics and AI-enabled consumer devices.
Exhibition covers the memory and storage supply chain
Alongside the conference, the organiser expects more than 39 exhibitors to present over 300 products and innovations.
The announced exhibition scope includes wafer-based integrated device manufacturers, controller developers, storage-module companies, outsourced semiconductor assembly and testing providers, equipment suppliers, AI-computing companies, server manufacturers and application developers.
Products are expected to cover areas such as enterprise solid-state drives, server memory, embedded storage, controllers, packaging and testing equipment, AI servers and related infrastructure.
For buyers, the relatively specialised format may make it easier to discuss detailed performance and integration requirements than at a large general electronics exhibition. Visitors should nevertheless request product-level specifications, compatibility information, qualification results and supply commitments before evaluating potential partners.
Companies interested in the wider semiconductor manufacturing and packaging ecosystem can also review ChinaExpoHub’s guide to ICES 2026 at CIIF Shanghai, which covers IC design, wafer fabrication, advanced packaging, equipment and industrial AI applications.
Supply-chain resilience and investment complete the agenda
GMIF’s fourth programme theme concerns the changing global semiconductor market. Discussions will address diversified sourcing, regional cooperation and the development of more resilient memory and storage supply chains.
The September 22 investment forum is intended to connect technology companies with industrial investors and financial institutions. The main programme will then combine technical presentations with supply-and-demand matchmaking, product displays, an awards ceremony, the “Greater Bay Area Night” networking event and a global livestream.
For international companies, the most useful conversations are likely to centre on supplier diversification, AI-storage performance, enterprise SSD development, controller compatibility and partnerships linking hardware with AI software.
GMIF 2026 provides a focused view of how the rapid expansion of AI inference is reshaping the memory and storage market. Its central message is that storage is no longer a secondary component of AI infrastructure: it increasingly determines how quickly, efficiently and economically AI systems can operate at scale.
