Edge-to-Cloud AI Infrastructure Takes Center Stage
Edge-to-cloud AI infrastructure is the combined stack of compute, storage, networking, and software that runs machine learning workloads consistently from local edge devices through data centers and into cloud platforms, allowing organizations to process data near its source while coordinating training and management centrally. At Computex, three major vendors drew a clear picture of how this stack is evolving. Marvell focused on AI data center connectivity and the move from electrical to optical links. Infortrend presented integrated systems that span edge computing platforms, private cloud, and high-performance storage for AI and HPC workloads. QNAP highlighted edge AI NAS and business continuity tools designed to keep services online during failures. Together, their announcements show a strong industry trend toward tightly integrated edge to cloud computing solutions that prioritize both performance and resilience.
Marvell: Connectivity Emerges as the New AI Bottleneck
Marvell’s keynote emphasized that future AI performance gains will depend as much on connectivity as on CPUs, GPUs, or memory. CEO Matt Murphy argued that networking is becoming the main bottleneck in AI data centers, especially as compute performance rises faster than interconnect bandwidth. He highlighted Marvell’s long-term bet on data infrastructure, including acquisitions such as Aquantia, Cavium, Celestial AI for photonics, and XConn for scale-up switching. According to Marvell, they have invested roughly 36 billion dollars in their data infrastructure platform, signaling the scale of their shift toward high-margin data center markets. The company also underlined its closer relationship with NVIDIA, noting its role as an NVLink Fusion partner and NVIDIA’s significant investment. This partnership aims to support next-generation AI data center connectivity, with a clear focus on optical links that can keep pace with large-scale AI clusters.
Infortrend: Integrated AI Infrastructure from Edge to Cloud
Infortrend used Computex to present a lineup designed to simplify AI deployments across distributed environments. Its edge computing platform brings AI inference, real-time analytics, and automation directly to where data is generated, reducing the need to route every workload back to a central data center. The platform is available as bare-metal hardware or as a pre-integrated solution with three deployment modes: Standalone Edge, High Availability Edge, and Advanced Edge clusters, giving organizations flexible options for scale and resilience. Infortrend’s enterprise cloud platform combines high-performance compute, GPU acceleration, and a pre-installed software stack to support AI training, big data analytics, and HPC in private data centers. For storage, the EonStor GS 5000U and GSx 5000 lines deliver high throughput and parallel file access so multiple AI workloads can share large datasets efficiently across the edge to cloud computing pipeline.
QNAP: Edge AI NAS and Self-Defending Infrastructure
QNAP framed its Computex message around “Ready and Recovery,” focusing on data sovereignty, cyber-resilience, and localized intelligence. The new HDP for Business platform centralizes protection for Windows, VMware, Hyper-V, Proxmox VE, and Microsoft 365, and is built on the 3-2-1-1-0 backup rule. It adds Immutable Backup and Airgap+ features to defend against ransomware by isolating backup nodes, while Video Verification records VM boot processes so IT teams can confirm recoverability within minutes. On the infrastructure side, QNAP’s MEGA Scale-out solution and Dual-Active Controller NAS models aim for near-zero recovery time, eliminating single points of failure in data lakes and critical services. In edge AI infrastructure, QNAP is introducing compact 8-bay and 6-bay AI NAS models powered by Intel Core Ultra Series 3 processors, designed to run on-premises LLMs, build private knowledge bases, and automate AI agents while preserving data control.
A Converging Vision for Edge AI and Business Continuity
Taken together, the Computex announcements from Marvell, Infortrend, and QNAP outline a converging vision for future AI infrastructure. Marvell is addressing AI data center connectivity, arguing that optical interconnects will define performance as AI clusters scale. Infortrend is focusing on integrated platforms that bridge edge and private cloud, backed by high-throughput storage tailored for AI and HPC workloads. QNAP is blending edge AI NAS with business continuity tools, aiming to deliver self-defending environments where localized intelligence and rapid recovery go hand in hand. A shared thread is the move toward unified edge AI infrastructure that spans from the intelligent edge to the enterprise cloud, reducing deployment complexity while improving resilience. As organizations scale AI, this integrated edge to cloud computing model is likely to become the default way to deploy, protect, and connect intelligent systems.
