The Spark's Value Gap Narrowed | Generated by AI

Home 2026.05

Question: I researched the DGX Spark and AMD Ryzen AI Max+ 395 mini PCs two months ago, and back then the cost difference between the two was around ¥10,000 (RMB). Now the AMD 395 machines have gone up to ¥25,000 — does that make the DGX Spark the better deal now?

Answer:

You’re essentially correct — the value gap between the two options has narrowed significantly, and here’s why:

What Happened to Prices

NVIDIA DGX Spark also got more expensive. NVIDIA raised the DGX Spark’s MSRP from $3,999 to $4,699 — an 18% increase — citing worldwide constraints in memory supply. So it’s not just AMD machines that went up; the Spark is also suffering from the same LPDDR5X shortage.

AMD Ryzen AI Max+ 395 mini PCs with 128GB have been hit hard. As of March 24, 2026, 128GB AMD 395 machines range from $2,399 (Bosgame M5) to $3,810 (MINIX ER939-AI), with popular options like Framework Desktop at $2,851, Corsair AI Workstation 300 at $2,499, and GMKtec EVO-X2 at $3,000. RAM prices are the main culprit — prices for RAM and storage have been skyrocketing in recent months, which has put these already-expensive computers even further out of reach.

Is DGX Spark Now the Better Value?

Not necessarily — it depends on your use case. Here’s the breakdown:

DGX Spark advantages:

AMD 395 mini PC advantages:

Performance nuance: On heavy models like GPT-OSS 120B, DGX Spark excels at prompt processing (compute-bound) but struggles with token generation (memory-bound), where the LPDDR5X bandwidth of 273 GB/s becomes a bottleneck.

Summary

If the cheapest AMD 395 options (Corsair, Bosgame) at ~$2,400–2,500 are no longer available in your market and prices have genuinely risen to DGX Spark territory, then yes — at ¥25,000 for AMD vs a comparable price for DGX Spark, the Spark starts to look more reasonable because you get the CUDA ecosystem, better AI software support, and a more “turnkey” experience. But if you can still find AMD 395 options at ¥18,000–20,000, the value advantage for AMD remains, especially for flexible, general-purpose AI work.

References:


Back Donate