NVIDIA will not release a single new gaming graphics processing unit (GPU) in 2026, marking the first full calendar year without a consumer GPU launch since the company's founding in 1999. The decision reflects a fundamental shift in the company's priorities: AI datacenter chips now generate far more revenue than gaming cards, and the company is deliberately rationing scarce memory components to feed its lucrative server business.

Why Is NVIDIA Abandoning the Gaming Market?

The answer comes down to economics and supply constraints. NVIDIA had completed the design for an RTX 5080 Super with 24 gigabytes of memory, and multiple sources confirmed that at least one manufacturing partner had received test units. Then NVIDIA ordered a halt to distribution.

The culprit is GDDR7 memory, the high-speed RAM used in graphics cards. Each 3-gigabyte GDDR7 module costs between $60 and $70, roughly three times the price of older 2-gigabyte modules. Global production capacity is limited, and NVIDIA faces a choice: allocate those expensive memory chips to gaming cards that sell for $800 to $2,500, or use them in datacenter systems that generate billions in revenue.

The math is brutal. A single Vera Rubin datacenter, NVIDIA's next-generation AI platform, costs approximately $47 billion to build and includes 288 gigabytes of HBM4 memory per GPU with 100% liquid cooling. Foxconn is already constructing these facilities. By comparison, releasing an RTX 5080 Super for $900 is, as one analyst put it, "pocket money".

What Was Supposed to Launch in 2026?

NVIDIA's original plan included two major releases. First, the RTX 50 Super refresh, a traditional mid-cycle update that has accompanied every GPU generation since the RTX 20 series. Second, early development work on Rubin, the next-generation architecture already referred to as RTX 60 in industry leaks.

The RTX 5080 Super was designed to address one of the RTX 5090's main criticisms: insufficient VRAM for demanding 4K gaming and local AI workloads. The upgrade from 16 to 24 gigabytes was finalized, and a launch was initially planned for CES 2026 in January, then postponed to summer. Now, nothing.

For the RTX 60 series, the timeline has stretched even further. Reliable leaker kopite7kimi initially predicted a Rubin gaming launch in the second half of 2027. More recent sources suggest the end of 2027 at the earliest, with early 2028 as a more realistic target. This creates a three-year gap between the RTX 50 series launch in early 2025 and the RTX 60 series availability, an unprecedented drought in NVIDIA's gaming lineup.

How to Navigate GPU Availability in 2026

• High-End Gaming: The RTX 5090 remains the only new NVIDIA option, but availability is severely constrained. Official European pricing was €2,349, yet retail prices in July 2026 range from €3,700 to €4,500 for standard models, a 25% premium over launch price. Special editions command even steeper markups, with the ASUS ROG Matrix Platinum reaching €9,539 and limited-edition variants exceeding €10,000 on the secondary market.
• Mid-Range Alternatives: AMD's Radeon RX 9070 XT offers the most practical option for mid-range gaming. Priced between €645 and €870 across European retailers, it competes with the RTX 5070 in 1440p performance and is actually available in stock, unlike most NVIDIA cards.
• Entry-Level Options: Intel's Arc A770 and B580 target budget and mid-range segments, though Intel has no high-end offering. The company is focused on foundry operations and a lithography partnership with NVIDIA rather than discrete gaming GPUs.

When Will NVIDIA Return to Gaming?

The RTX 60 series will use the GR20x chip architecture manufactured at 3 nanometers by TSMC. The flagship RTX 6090 is expected to include 32 gigabytes of memory on a 512-bit bus, with estimated performance gains of 30 to 35 percent in rasterization compared to the RTX 5090.

NVIDIA will likely announce the RTX 60 at CES 2027 in January, but actual availability will lag significantly. Second semester 2027 is optimistic; early 2028 is more realistic. Even then, TSMC's 3-nanometer production capacity is finite, and AI orders take priority. Gamers should expect the same launch chaos that plagued the RTX 5090: stock depletion within minutes, price gouging, and bot-driven shortages.

NVIDIA has not officially canceled the RTX 50 Super refresh, instead describing it as a "pause." If GDDR7 supply improves dramatically, a launch at CES 2027 remains theoretically possible. However, nothing is guaranteed, and industry sources suggest the company is unlikely to reverse course given the vastly superior margins on AI infrastructure.

The broader message from NVIDIA leadership is clear. CEO Jensen Huang stated at the company's GTC conference that NVIDIA's future lies in "AI factories," not graphics cards for gamers. With Vera Rubin and other datacenter platforms generating tens of billions in revenue, the consumer gaming market has become a secondary concern. For the first time in 27 years, gamers will have to look elsewhere.