Nvidia’s New Budget AI System Comes at a Premium Due to Memory Shortages

Nvidia is rolling out a new lower-memory version of its DGX Spark desktop AI system, but with a surprising twist, it costs more than the original model did at launch. The new 64GB DGX Spark will hit the market on October 23 priced at $4,999, which is $1,000 higher than the original 128GB model’s launch price of $3,999, despite offering half the memory.

The pricing discrepancy becomes even more apparent when comparing the original model’s current cost. Today, the 128GB DGX Spark retails for about $6,950, roughly 75% above its debut price. Partner versions are reported to range from $5,999 to $9,000, depending on storage and configuration. Despite the reduced memory, the 64GB version still features Nvidia’s GB10 Grace Blackwell Superchip and the same AI software stack, with the main trade-off being a reduced capacity to support AI models with up to 100 billion parameters, compared to 200 billion for the higher-memory system.

Nvidia has attributed the rising prices to industry-wide memory supply constraints, a challenge that may persist. Micron has warned that memory shortages could worsen through 2027 and 2028, with no clear timeline for supply and demand to balance.

For investors, the concern extends beyond the DGX Spark’s pricing. The broader question is whether elevated memory costs will impact Nvidia’s pricing strategy, customer budgets, or hardware configurations across its AI product lineup. So far, strong demand has allowed Nvidia to pass on much of the cost to customers. However, if memory prices remain high, customers may become more selective about expanding AI infrastructure, making memory supply a key cost variable to monitor as Nvidia navigates the next phase of AI development.