RTX 50 SUPER rumored window narrows: late Q1 to early Q2 2026, not at CES
If you’ve been waiting for NVIDIA’s next wave of Blackwell-based upgrades, the latest chatter points to a spring 2026 launch for the GeForce RTX 50 SUPER family. Multiple reports suggest the cards won’t debut at CES in early January. Instead, the rollout is now rumored for late Q1 to early Q2 2026, with March to May being the most likely window.
As always, treat this as early guidance. Word is that board partners haven’t received formal launch notifications yet. Still, several breadcrumbs align: recent PSU calculator listings mentioning SUPER-branded 50-series models and long-standing insider hints that these SKUs exist, even if their unveiling isn’t tied to CES.
What to expect from RTX 50 SUPER
– Focus on higher VRAM for the mid to upper-mid tier: Expect SUPER refreshes of the 70- and 80-class GPUs, specifically GeForce RTX 5070, RTX 5070 Ti, and RTX 5080.
– Bigger memory configurations: Rumors point to sizable VRAM bumps over the non-SUPER versions. Early listings suggested roughly 50% more memory for some models, with talk of 18 GB for RTX 5070 SUPER and up to 24 GB for RTX 5070 Ti SUPER.
– Don’t expect a 5090 SUPER: The RTX 5090 already ships with 32 GB of GDDR7, making a memory-focused refresh unlikely.
– 60-class remains a question mark: A 5060 SUPER isn’t guaranteed, though the 60 tier arguably benefits most from higher VRAM for modern games and creator workloads.
Preliminary RTX 50 SUPER lineup details (subject to change)
– GeForce RTX 5080 SUPER
– GPU: Blackwell GB203-450, 84 SMs, 10,752 cores
– Memory: 24 GB GDDR7, 256-bit, 32 Gbps, 1,024 GB/s
– Power: 1x 12V-2×6 (16-pin), TBP 400W+
– GeForce RTX 5080 (non-SUPER reference)
– GPU: GB203-400, 84 SMs, 10,752 cores
– Memory: 16 GB GDDR7, 256-bit, 30 Gbps, 960 GB/s
– Clocks: up to ~2.62 GHz
– Power: 1x 12V-2×6, TBP 360W
– Launch/MSRP reference: originally targeted for early 2025 at $999 US
– GeForce RTX 5070 Ti SUPER
– GPU: GB203-350, 70 SMs, 8,960 cores
– Memory: 24 GB GDDR7, 256-bit, 28 Gbps, 896 GB/s
– Power: 1x 12V-2×6, TBP 350W
– GeForce RTX 5070 Ti (non-SUPER reference)
– GPU: GB203-300, 70 SMs, 8,960 cores
– Memory: 16 GB GDDR7, 256-bit, 28 Gbps, 896 GB/s
– Clocks: up to ~2.42 GHz
– Power: 1x 12V-2×6, TBP 300W
– Launch/MSRP reference: originally targeted for early 2025 at $749 US
– GeForce RTX 5070 SUPER
– GPU: GB205-400, 50 SMs, 6,400 cores
– Memory: 18 GB GDDR7, 192-bit, 28 Gbps, 672 GB/s
– Power: 1x 12VHPWR (16-pin), TBP 275W
– GeForce RTX 5070 (non-SUPER reference)
– GPU: GB205-300-A1, 48 SMs, 6,144 cores
– Memory: 12 GB GDDR7, 192-bit, 28 Gbps, 672 GB/s
– Clocks: up to ~2.51 GHz
– Power: 1x 12VHPWR (16-pin), TBP 250W
– Launch/MSRP reference: originally targeted for early 2025 at $549 US
Key takeaways for upgraders
– Timing: Don’t expect a CES surprise; keep an eye on March–May 2026.
– Who benefits most: The biggest wins appear aimed at memory-bound gamers and creators on the 70 and 80 classes, with significant VRAM increases likely.
– Power and thermals: Early specs point to higher TBPs for some SUPER models, so plan your PSU and case airflow accordingly.
– 5090 owners: You’re already sitting on 32 GB of GDDR7; a 5090 SUPER seems unnecessary based on current info.
As with all pre-launch coverage, details can shift. But the emerging picture is clear: NVIDIA’s RTX 50 SUPER refresh looks geared toward more VRAM and broader bandwidth on the mid to upper-mid range, with a launch window squarely after CES and into spring 2026.
