Nvidia reportedly no longer supplying VRAM to its GPU board partners in response to memory crunch — rumor claims vendors will only get the die, forced to source memory on their own

The entire industry is currently navigating a memory crisis, with companies pivoting to favor AI clients over retail markets, leading to a severe drought that’s expected to only worsen. Nvidia, the world’s largest GPU manufacturer and a leading beneficiary in the AI boom, seems to now be affected by a problem it helped exacerbate. According to leaker Golden Pig Upgrade, Nvidia has stopped bundling video memory with its GPUs that it sells to AIBs, leaving partners to source the required VRAM on their own.

Nvidia doesn’t manufacture VRAM; it’s made by Samsung, Micron, or SK Hynix, which is then procured by either large AIBs directly, or Nvidia bundles it together with its GPUs to sell as a package to board partners. Clearly, the ongoing memory shortage has gotten so out of control that even the Green Team doesn’t consider this feasible anymore.

For larger vendors, sourcing GDDR on their own shouldn’t be a problem since it’s already standard practice. They have enough connections and experience to follow Nvidia’s specifications and keep separately-sourced memory aligned with reference designs. It’s the smaller partners that might be affected, squeezing margins even more in an already tight industry.

Best picks for you

• Nvidia’s RTX 5000 Super could be cancelled or get pricier due to AI-induced GDDR7 woes

• Manufacturers reportedly delaying launch of new memory modules due to massive price crunch

• Distributors force unprecedented RAM and motherboard bundle mandate to fight global shortage, report claims

A lot of you may remember how EVGA left the industry, citing growing tensions with Nvidia, alleging mistreatment. Not to speculate too much, but leaving vendors to buy cutting-edge VRAM on their own could put a lot more pressure on low-scale operations, putting them at risk of shutting down, unable to weather the memory drought.