MSI Offers EV Solutions, Edge AI Power, and Server Strength

• Share on X
• Share on LinkedIn

June 25, 2026

Sponsored Blog

COMPUTEX 2026 in Taipei Taiwan was a huge event this year, bigger than ever. There were more than 6000 exhibitors in more than 1500 booths, and one of the highlights for me was at MSI.

MSI had so much on display that it’s hard to identify what my favorite experiences were.

One of the first that jumps to mind is the incredible growth of the company’s EV offerings. I took some time for a look at the  MSI EzGo EV Charging Solutions area, which, to me, was a leap forward from last year, when MSI first introduced its EV solutions to me.

EZgo is designed for a wide range of charging needs: charging at home if permanent installations are not possible, charging away from home at work or hotels or other places without fast charging stations, or even as an emergency backup. The goal for MSI seems to be to offer EV owners flexibility, convenience and control.

It delivers up to 7kW of power via interchangeable plug options (e.g., NEMA 5-15, NEMA 14-50) and compatibility with international plugs and common EV connector standards including SAE J1772 (Type 1), IEC 62196-2 (Type 2), and NACS via adapter. It can be controlled via screen, app or buttons, so user really control what they want to experience, EZgo adapts seamlessly to a variety of vehicles and charging scenarios, making it a flexible and reliable portable solution for modern EV mobility.

Another key offering was the comprehensive and flexible MSI EdgeXpert. One demo that was particularly fun was the Tarot AI robot that offered a custom tarot reading, powered by EdgeXpert.

MSI calls EdgeXpert an AI supercomputer for edge, business, and personal uses. I saw applications and use cases from Legal to Smart Security robots, not to mention the tarot reader robot and its psychic powers.

A particularly interesting application came from Galene.ai, which was using EdgeXpert to power their AI agent that is tailored to regulated industries, so they need high security, accuracy and local data only.

The platform is based on NVIDIA DGX Spark GB10 Platform and NVIDIA Grace Blackwell GPU and Arm 20-core CPU. It has NVIDIA NVLink-C2C CPU-GPU memory interconnect, 128 GB LPDDR5x coherent unified system memory, and 1000 AI TOPS (FP4), MSI said.

Of note was the connect three MSI EdgeXpert in a Ring display that created a mesh topography with no switch required. Three QSFP cables is all it takes to form the ring or mesh directly, so no enterprise switch is needed to gain high bandwidth with low latency. The use case here is for larger models, if desired, and for pipeline parallelism.

On the server side, MSI didn’t disappoint. The company showed off its enterprise server platform scope, which was large. On view were the CX171-S3066, CX271-S3066, CX171-S4056, CX271- S4056 and many more. There were platforms were variously built for compact racks, power, flexibility, and energy efficiency. One highlight was the ORv3 21” 44OU Liquid-Cooled Rack Architecture, which is a high-density liquid cooled rack infrastructure featuring 100kW compute node systems and supports static weight loads of 1600 kilograms and a dynamic load of 1200 kilograms, while also enabling efficient power performance for AI workloads.

For me, the highlight of MSI the server area was the XpertStation WS300 (DLC). This NVIDIA DGX powered workstation is designed for prototyping, fine-tuning, and inferencing AI models right on the desktop. Now that’s edge computing with the power of a datacenter.

All in all, an incredible display of compute and sensitivity to engineering needs and concerns from MSI, and I look forward to what the next year will bring.