MSI's AI Flagship RTX Spark Launches in Q4 with Sub-$3000 Price Tag — First Batch Sold Out

PC brand MSI (2377-TW) held a strategic partnership press conference today (16th), where MSI Product Director Peng Jen-fang revealed that the company's flagship on-premise AI device, RTX Spark — developed in collaboration with NVIDIA (NVDA-US) over three years — will officially launch in the fourth quarter. Despite its high specifications, the device is expected to be priced below $3,000, and the first production batch has already sold out due to strong channel demand and aggressive restocking efforts.

Peng revealed that as the market's first flagship AI device to pioneer a new edge computing business model, global acceptance has far exceeded expectations. Top-tier distributors in the U.S. and China have already proactively placed follow-up orders.

To meet overwhelming demand, MSI is actively requesting compliant chip allocations from NVIDIA and simultaneously negotiating with the world's three major memory manufacturers to secure competitively priced core components and stabilize future supply.

In terms of hardware performance and power efficiency, the new RTX Spark breaks through traditional power consumption bottlenecks. While maintaining the same 1,000 TOPS of high-efficiency AI computing power as top-tier gaming GPUs, the desktop version reduces power consumption to 140 watts, while the laptop version precisely controls it at 80 watts to balance performance and battery life.

The sold-out first batch consisted entirely of high-spec models with a maximum 128GB memory configuration, capable of smoothly running 31B-parameter dense small language models, with future ambitions to handle 70B models. Real-world testing shows the device can process one hour of audio and generate an accurate meeting transcript in just three minutes — compared to nearly 30 minutes on a typical consumer laptop running at full load and overheating — significantly saving time.

The router software co-developed by MSI and the Institute for Information Industry (III) will release a trial version in August. By optimizing local resource allocation, it is expected to help heavy users save 30% to 40% on cloud API costs, allowing enterprises to shorten their equipment payback period while maintaining information security.

MSI expects the sales momentum of the high-spec version launching in Q4 to surpass that of the budget version, which won't launch until Q1 next year. Following NVIDIA's historical product release rhythm, the company plans to maintain a major refresh cycle every two years, continuously capturing edge computing opportunities.