MSI EdgeMesa N AI Mini Workstations (Image © MSI)
The EdgeMesa N AI+ system is equipped with the [NVIDIA RTX Spark N1X chip][1]. The architecture integrates a high-performance GPU and CPU on a single platform and uses a unified memory system to optimize data transfer between the computing units.
- GPU: NVIDIA Blackwell RTX GPU with 6,144 cores.
- CPU: NVIDIA Grace CPU with 20 cores.
- AI Computing Power: The system achieves a peak AI performance of up to 1 petaflop (FP4).
- Memory: Up to 128 GB of LPDDR5X unified memory, enabling the system to process larger AI models that typically exceed the capacity of standard consumer GPUs.
Specifications
- Operating System: Windows 11
- Processor: NVIDIA RTX Spark N1X
- GPU: NVIDIA Blackwell RTX (up to 6,144 cores)
- CPU: NVIDIA Grace (up to 20 cores)
- AI Performance: Up to 1 petaflop (FP4)
- Memory: Up to 128 GB LPDDR5X Unified Memory
- Storage: 1x M.2 NVMe slot
- Network: 10-GbE LAN / Wi-Fi 7 / Bluetooth 5.4
- Rear Ports: 4x USB-C (20 Gbit/s; 3x DP Alt Mode 2.1, 1x Power Delivery), 1x HDMI 2.1, 1x RJ45 (10 GbE)
- Security: TPM 2.0
- Power Supply: 240-W power supply
- Cooling: Vapor chamber with fan cooling
- Weight: 0.55 kg (net) / 1.55 kg (gross)
This configuration is thus similar in design to AMD’s Strix Halo systems and also competes with Apple’s Mac Studio.
To ensure stability during sustained AI workloads, MSI has introduced the “AI Pulse Cooling System.” This cooling solution uses a vapor chamber in combination with two turbo fans to handle a thermal design power (TDP) of 140 watts. The system is housed in a 1.3-liter aluminum chassis. The exterior features a fingerprint-resistant coating, allowing the hardware to maintain a professional appearance in a desktop environment while taking up minimal space.
Connectivity
The EdgeMesa N AI+ is equipped with high-bandwidth interfaces to facilitate the transfer of large data sets and the integration of multiple peripheral devices:
- Network: 10-Gigabit LAN and Wi-Fi 7 (with Bluetooth 5.4) for fast data transfer and network stability.
- External Ports: Four USB Type-C ports with a transfer rate of 20 Gbit/s. Three of these ports support DisplayPort Alt Mode 2.1, while one is designated for power delivery. An HDMI 2.1 port (FRL) is also included.
- Display Connectivity: The system can drive up to four independent displays simultaneously.
For resource allocation, the system features Personal AI Router (PAIR) technology. This allows compatible local devices to form a computing cluster that intelligently distributes AI requests across available hardware to optimize resource utilization.
Among the configurations, there is a 64 GB RAM version, which should be sufficient for some less demanding models; however, the more interesting LLMs require 128 to 512 GB of memory. This means that only the significantly more expensive 128 GB RAM variant is a viable option if you want to purchase such a system. However, for this configuration, there is the AMD Strix Halo platform, which is used, for example, in the [ACEMAGIC M1A PRO Mini-PC (Review)][2] and costs significantly less than the NVIDIA DGX solution.
[1]: https://www.pcmasters.net/server/133716845-nvidia-rtx-spark-dgx-spark-and-dgx-station-these-are-the-alleged-powerhouses.html
[2]: https://www.pcmasters.net/review/133716977-acemagic-m1a-pro-mini-pc-review-featuring-ryzen-ai-max-plus-395-and-128-gb-of-ram.html



