AI Boom Pushes Nvidia and AMD Gaming GPUs Toward 2028

SAN FRANCISCO — Nvidia Corp. and Advanced Micro Devices Inc. are projected to delay major consumer graphics-card launches until 2028 as manufacturers prioritize high-margin enterprise AI hardware. The traditional biennial release cycle for high-end gaming graphics cards, a cornerstone of the personal computer industry for more than a decade, is buckling under the weight of the global artificial intelligence boom.

Hardware leaker Kopite7kimi, a historically accurate source for Nvidia product roadmaps, has revised earlier forecasts that placed Nvidia’s next-generation “Rubin” architecture, expected to market as the GeForce RTX 6000 series, in the second half of 2027. The launch is now projected to occur in 2028. Fellow hardware analyst Kepler_L2 has also indicated that Nvidia’s RTX 6000 series and AMD’s next-generation, RDNA 5-based “Radeon 10000” series have had their launch schedules deferred to 2028.

For years, PC gamers could expect a new GPU architecture every 24 to 26 months. Nvidia’s modern generations moved from Pascal, used in the GTX 10-series, in 2016 to Turing and the RTX 20-series in 2018, Ampere and the RTX 30-series in 2020, and Ada Lovelace and the RTX 40-series in 2022. Approximately 20 months have passed since the company launched its current Blackwell-based GeForce RTX 50-series, led by the flagship RTX 5090 and high-end RTX 5080.

| GPU Generation | Architecture | Release Year | Interval (Months) |
| :— | :— | :— | :— |
| GeForce GTX 10-Series | Pascal | 2016 | — |
| GeForce RTX 20-Series | Turing | 2018 | ~28 months |
| GeForce RTX 30-Series | Ampere | 2020 | ~24 months |
| GeForce RTX 40-Series | Ada Lovelace | 2022 | ~25 months |
| GeForce RTX 50-Series | Blackwell | Early 2025 | ~27 months |
| GeForce RTX 60-Series | Rubin | Projected 2028 | ~41–47 months |

Should the RTX 6000 series remain on shelves until the latter half of 2028, the gap between generations will reach between 41 and 47 months. A single consumer GPU generation would then have a lifespan of nearly four years, marking the longest gap in the history of the modern graphics card market.

The delay is tied to an acute global shortage of advanced memory components. Generative AI has created insatiable demand for high-bandwidth memory (HBM) and next-generation graphics double data rate (GDDR) memory. Taiwan Semiconductor Manufacturing Co. (TSMC) and other leading semiconductor foundries and packaging facilities have allocated a massive portion of their production capacity to enterprise AI accelerators, including Nvidia’s H100, B200, and upcoming Rubin AI platforms.

That industrial pivot has limited consumer-market allocation and pushed retail prices for gaming hardware to historic highs. Nvidia’s flagship consumer card, the GeForce RTX 5090, has reached market prices of between $7,000 and $9,000 in various distribution channels as researchers and enterprise buyers purchase consumer cards for lower-tier AI training and development tasks.

The same supply crunch has stalled mid-generation products. Industry reports from July said Nvidia had completed designs for an updated RTX 5000 “Super” series, but the cards were held back from mass production because of the high cost and limited supply of required 3GB GDDR7 memory modules. Memory manufacturers are prioritizing higher-density and enterprise-grade modules, leaving consumer graphics divisions with limited components.

SK Hynix, one of the world’s primary suppliers of high-speed DRAM and HBM alongside Samsung Electronics and Micron Technology, has issued conservative projections for the memory market. The South Korean memory giant warned that the current supply crunch is likely to intensify through next year, while structural shortages in advanced memory packaging could persist until 2030. Acer Inc. Chairman and CEO Jason Chen has offered a slightly more optimistic view, suggesting that supply-side adjustments could stabilize the market sooner, although retail pricing is still expected to remain sensitive to enterprise demand for the foreseeable future.

AMD is pursuing a different near-term path in the mainstream market. While its enthusiast-grade RDNA 5 architecture is delayed alongside Nvidia’s Rubin, the company’s upcoming “AT2” GPU remains on track for a launch next year. The silicon is designed to compete directly against Nvidia’s existing GeForce RTX 5080. AMD has historically competed with Nvidia in the ultra-enthusiast GPU segment, but a successful AT2 launch next year could give the company a significant window to capture market share in the high-performance gaming segment while both manufacturers hold back their next-generation flagship architectures until 2028.