2026 has turned into a year of surprises in the PC hardware space. As we cross the midpoint of the year, there have been no introductions of new discrete GPU architectures. Retail shelves are still stocked with a Ryzen CPU design that's nearly four years old, while NAND flash memory prices have climbed to staggering levels. More troubling is the situation in the DRAM market, which may be the worst inflation cycle ever recorded for memory components. With the consumer graphics space settling into a lull and no new architectures on the horizon, there’s yet another disappointment: the anticipated Super refresh of Nvidia’s Blackwell architecture isn't coming this year. For users like me who were eagerly waiting for a new value tier in the GeForce RTX lineup, this delay is not just frustrating but deeply disheartening. The truth is, these cards already exist, but they simply can't be released right now due to ongoing supply chain issues.
The Ti Super tier: A smart and future-proof choice
If there’s one thing I’m certain of, it's that the Ti Super tier represents the best value in Nvidia’s current GeForce RTX lineup. When I purchased the RTX 4070 Ti Super in January 2024, I did so knowing exactly what I was getting into. I’ve always taken a measured, analytical approach to building my PC, and the decision to invest in the RTX 4070 Ti Super was no different. This card redefined the upper-midrange segment when it launched. It skipped the hefty premium usually reserved for flagship graphics cards, provided rasterization power within 15–25% of the RTX 4080 above it, and came with a generous 16GB of memory on a wide 256-bit bus. This made it not only a powerful gaming option but also a smart investment for future upgrades.
It wasn’t long before the RTX 4070 Ti Super proved itself to be a dual-purpose powerhouse. What I didn’t foresee at the time was how the rise of local AI inference would turn this graphics card into something much more than a gaming tool. My ability to run quantized AI models without constant interruptions is now a major advantage. Just the other day, a close friend of mine tried to load the same Mistral Small 24B model I use, but he couldn’t get it to work smoothly. He was using the 12GB RTX 4070 Ti, and his card struggled to handle the workload, spilling layers into system RAM. In comparison, my Ti Super ran everything seamlessly, showing just how much 16GB of VRAM can make a difference.
The RTX 4070 Ti Super was always the more versatile and forward-thinking option. It gave me extra headroom and power, which are arguably the most critical factors when choosing a GPU. Two and a half years later, the same logic still holds. Whether I’m keeping up with the VRAM demands of 1440p gaming or managing desktop-level AI models, the Ti Super remains a solid choice. That’s why I expected Nvidia to roll out its successor by now. After all, it's been well over two years since I made that purchase.
RTX 50 Super: The latest victim of the memory shortage
It turns out the RTX 50 Super lineup is the newest casualty of the ongoing DRAM crisis. According to a report from VideoCardz on July 17, at least one of Nvidia’s board partners already has completed Super cards in hand. The hardware has moved beyond the prototype stage and is physically present, ready to be shipped. However, Nvidia has reportedly instructed its add-in board (AIB) partners to hold onto the cards without any clear indication of when they’ll be released. The reasons behind this delay are likely related to the VRAM situation.
The RTX 50 Super series is said to use 3GB GDDR7 modules, a configuration that’s currently in very short supply and commanding high prices. Each 3GB module costs between $60 and $70, while the more standard 2GB version is only about $20. With 6–8 of these modules required for each GPU, this price gap quickly adds up. As reported by Tom's Hardware, the cost of materials alone increases by hundreds of dollars per card. For board partners already under pressure, this makes things much more complicated.
The situation is further complicated by the fact that Nvidia no longer supplies VRAM chips to its partners. This shift happened at the start of the memory crisis, and now AIBs must source GDDR7 chips themselves. This not only increases the financial burden on these partners but also leads to delays as they struggle to secure enough supply to meet production demands.
RTX 50 Super models in the pipeline
The specific models affected by the delay include the RTX 5080 Super and 5070 Ti Super, both equipped with 24GB of GDDR7 VRAM. There's also the 5070 Super, which has 18GB of VRAM, and the RTX 5050 with 9GB. All of these models have reportedly been developed and are ready to ship. Importantly, they maintain the same CUDA core counts as their base models, ensuring consistent performance characteristics.
As a consumer, this delay feels more than just inconvenient. It’s a reevaluation of expectations. The RTX 4070 Ti Super, which I still use regularly, continues to perform well for both gaming and AI tasks. However, with the potential successor now stuck in storage, the idea of an upgrade feels stalled. There's a clear disconnect between what hardware is available and what users are hoping for in terms of advancement.
What the future holds
Until the RTX 50 Super lineup is finally released, users who rely on mid-range cards will need to make the most of what they have. Some may push their current hardware to its limits, while others might look elsewhere for performance improvements. As much as I would like to move up to the next tier, for now, the upgrade path remains uncertain.