NVIDIA RTX 5090D: The upcoming NVIDIA RTX 5090D, once positioned as a high-end GPU for the Chinese market, is reportedly getting a major performance downgrade. Due to tightened U.S. export restrictions, NVIDIA is allegedly being forced to nerf the card — slashing its CUDA cores and memory specs, leaving enthusiasts wondering whether it will retain any flagship DNA at all.
At the same time, NVIDIA is rumored to launch a new RTX 5080 Ti or RTX 5080 Super by the end of 2025 — aiming to bridge the performance gap while adhering to global regulations.
Table of Contents
NVIDIA RTX 5090D Rumored to Be Downgraded to 14,080 CUDA Cores, RTX 5080 Ti/Super Launch Expected in Late 2025
U.S. Export Rules Force a Toned-Down RTX 5090D
Originally meant to provide flagship-level performance in regions like China, the RTX 5090D now faces significant architectural compromises:
- CUDA cores slashed from 21,760 to just 14,080
- Memory bandwidth limited to 1.3 TB/s, staying under the U.S. 1.4 TB/s cap
- 384-bit memory bus with 24 GB GDDR7 VRAM (down from 32 GB)
- Total memory + I/O bandwidth capped under 1.7 TB/s
These reductions align the RTX 5090D more closely with the specs of the RTX PRO 5000, effectively stripping it of its top-tier gaming and professional performance capabilities.
Will the New RTX 5090D Still Compete?
If the rumored specs hold true, the RTX 5090D will fall well below the original RTX 5090 in terms of:
- Gaming performance
- Ray tracing capabilities
- AI and productivity workloads
Some sources even suggest this version might not be marketed as a flagship GPU anymore. Instead, it may be repurposed or renamed depending on how NVIDIA chooses to position it — possibly even replacing or merging with the upcoming 5080-class cards.
RTX 5080 Ti / Super: Coming Late 2025
While the downgraded 5090D garners attention, a more powerful 80-class GPU is quietly brewing in the background. Expected to arrive by Q4 2025, this card — possibly the RTX 5080 Ti or RTX 5080 Super — is rumored to feature:
- 24 GB of GDDR7 VRAM
- Performance that could fill the gap left by the nerfed 5090D
- A chance of being rebranded as the RTX 5090D for restricted markets if specs align
This gives NVIDIA room to navigate geopolitical trade rules while maintaining performance consistency for gamers and creators worldwide.
What This Means for the GPU Market
These developments reflect the growing influence of regulatory policy on tech specs. The RTX 5090D’s downgrade isn’t driven by engineering limits but by policy constraints. As a result:
- China-exclusive GPUs may continue to diverge from global versions
- NVIDIA is forced to segment its product lineup further
- Gamers in restricted regions may pay more for less power, or wait longer for better alternatives
The upcoming RTX 5060, expected to launch on May 20, 2025, will also compete directly with AMD’s RX 9060 XT, adding more heat to the mid-range market as NVIDIA recalibrates its high-end strategy.
As we await official confirmation from NVIDIA, it’s clear the RTX 5090D is no longer the performance powerhouse it was initially shaping up to be. With only 14,080 CUDA cores and reduced memory specs, it’s looking more like a premium mid-range card than a flagship.
However, the GPU war is far from over. The RTX 5080 Ti/Super could redeem NVIDIA’s high-end aspirations for 2025 — if it delivers on both performance and availability.
FAQs
Why is the RTX 5090D being downgraded?
Due to new U.S. export restrictions on GPUs shipped to China, which limit memory bandwidth and processing power.
How many CUDA cores will the RTX 5090D have?
Rumors suggest it will drop to 14,080 CUDA cores, down from 21,760.
Read more : NVIDIA RTX PRO 6000 vs. RTX 5090: 5% Faster in Path Tracing
