This web page was created programmatically, to learn the article in its authentic location you possibly can go to the hyperlink bellow:
https://www.tweaktown.com/news/113534/nvidias-rtx-5090-cards-are-ending-up-in-ai-servers-in-bulk-instead-of-gaming-pcs/index.html
and if you wish to take away this text from our web site please contact us
Photos shared by HKEPC on X present pallets of brand-new retail GeForce RTX 5090 playing cards being unpacked and slotted into multi-GPU AI servers. These aren’t NVIDIA’s workstation RTX PRO 6000 playing cards constructed for AI servers, however moderately the identical MSI and GIGABYTE RTX 5090 playing cards you’d discover on Newegg below the “gaming graphics card” class, that’s, in the event you might discover them at something resembling an inexpensive value.
The least expensive RTX 5090 listings within the US have crossed $5,000, greater than double the $1,999 launch value. European pricing has already cleared €5,200, and now now we have a clearer image of what is behind these skyrocketing costs.
Interestingly, the playing cards in these photos aren’t even the RTX 5090 D, the China-specific variant, which was intentionally nerfed for AI workloads at 2,375 AI TOPS in comparison with the usual mannequin’s 3,352, whereas maintaining gaming efficiency largely intact.
Back in 2021, when miners swept up RTX 30-series playing cards, NVIDIA responded with Lite Hash Rate variations that reduce mining efficiency whereas leaving gaming efficiency intact. The restriction was ultimately bypassed, however it labored effectively sufficient for lengthy sufficient to matter. This time, NVIDIA has not performed the equal.
NVIDIA is not primarily a gaming firm. The phrase “gaming” seems third within the metadata of NVIDIA’s personal homepage, behind AI and HPC. CEO Jensen Huang not too long ago joined X and shared a letter from NVIDIA and 24 different firms urging US policymakers to not limit open-weight AI.

VIEW GALLERY – 3 IMAGES
Some components of the HKEPC pictures seem to have been edited or altered, so deal with the photographs with some warning. We coated a Vietnamese system builder putting in seven RTX 5090s into customized AI servers again in April 2025. While $5,000 costs would possibly make it unimaginable for even premium rig builders to take a position, it appears to work simply fantastic for the ravenous AI scene, largely as a result of the RTX 5090 is a discount in contrast with an AI workstation GPU.
NVIDIA needs $16,000 for an RTX PRO 6000, and whereas that card packs 96GB of VRAM, many firms scrambling to construct servers could make do with far much less. Some firms would possibly look to the newly debuted RTX PRO 5500 Blackwell workstation GPU with its GDDR7 reminiscence, however it in the end comes down as to if it may well beat that $5,000 mark, which does not appear probably.

Frequently Asked Questions
Open a query for a solution from TweakTown’s protection of this information, or ask your individual beneath.
Question #1
Why are RTX 5090 playing cards being utilized in AI servers as an alternative of NVIDIA’s RTX PRO 6000 playing cards?
RTX 5090 playing cards are ending up in AI servers as a result of they’re much cheaper than NVIDIA’s RTX PRO 6000 playing cards, even on the inflated market costs. The piece says an RTX PRO 6000 prices $16,000 and has 96GB of VRAM, whereas many firms constructing servers could make do with far much less reminiscence, so the RTX 5090 remains to be seen as a discount for that use. It additionally notes that NVIDIA has not added an RTX 30 collection type limiter for this technology, so the patron card might be put to work in AI deployments.
Question #2
What makes the RTX 5090 enticing for AI workloads regardless of being a gaming GPU?
It presents sturdy AI functionality in a shopper card, with the usual RTX 5090 listed at 3,352 AI TOPS, and it’s not the nerfed China-specific RTX 5090 D. It can also be far cheaper than NVIDIA’s RTX PRO 6000, which the piece says prices $16,000, so firms constructing AI servers can use the 5090 as a discount possibility even with a lot much less VRAM.
Question #3
How does the RTX 5090 examine with the China-specific RTX 5090 D for AI use?
The commonplace RTX 5090 is the one exhibiting up in bulk in AI servers on this story, and it’s described as a discount for AI use in contrast with workstation GPUs. The China-specific RTX 5090 D is explicitly totally different as a result of it was intentionally decreased for AI workloads, at 2,375 AI TOPS versus 3,352 for the usual mannequin, whereas maintaining gaming efficiency largely intact. Separate protection in NVIDIA’s new GeForce RTX 5090D for China has restricted AI, crypto-mining efficiency says that China-tuned card is meant to cease AI use and cryptomining.
Question #4
Did NVIDIA do something this technology to restrict the RTX 5090 from getting used for AI prefer it did with Lite Hash Rate playing cards?
NVIDIA didn’t do the identical form of broad shopper restriction for the RTX 5090 that it did with Lite Hash Rate playing cards. This story says there isn’t any RTX 5090 equal of these mining-limited fashions, although the China-only RTX 5090D was intentionally nerfed for AI workloads whereas maintaining gaming efficiency largely intact.
Have a query that is not listed right here? Ask beneath, and TweakBot will reply it.
That stated, the RTX 5080 isn’t protected both. Plenty of AI deployments will fortunately make do with 16GB of VRAM, which successfully pushes the life like ceiling for shopper GPU patrons right down to the RTX 5070 Ti and beneath. You’re nonetheless over $1,000 for these on present market listings. The prime finish of shopper GPU gaming has been quietly cordoned off, and nothing on the horizon means that adjustments anytime quickly.
This web page was created programmatically, to learn the article in its authentic location you possibly can go to the hyperlink bellow:
https://www.tweaktown.com/news/113534/nvidias-rtx-5090-cards-are-ending-up-in-ai-servers-in-bulk-instead-of-gaming-pcs/index.html
and if you wish to take away this text from our web site please contact us

