More aggressive rumors, attributed to leaker Moore's Law Is Dead, push the specifications significantly higher: a flagship "Radeon RX 10090 XT" with a massive 154 CUs and 36 GB of GDDR7 memory on a 384-bit bus, with a 380W TDP . While eye-catching, this scenario is less independently corroborated than the 96 CU model.
New micro-architectural features are also expected. Leaked details consistently mention additions like Universal Compression to improve effective bandwidth, Neural Arrays for AI-accelerated tasks like upscaling, and redesigned Radiance Cores for a significant leap in ray tracing performance .
On the Nvidia side, plans appear more fluid. Leakers including kopite7kimi, one of the most reliable Nvidia sources, have stated the GeForce RTX 60-series, based on the "Rubin GR20x" architecture, will launch in the second half of 2027 . This aligns with other reports suggesting the flagship RTX 6090 may not be announced until CES 2027 for a later release
.
However, conflicting reports have injected significant doubt. Tom's Hardware and a report from The Information have suggested that Nvidia does not plan to launch any new gaming GPUs in 2026, and that the RTX 60-series itself could be pushed back to 2028 . This uncertainty is reportedly driven by a shortage of 3 GB GDDR7 memory modules and Nvidia's overwhelming focus on far more lucrative data-center AI accelerators, which consume similar resources
. The lack of an RTX 50-series Super refresh at CES 2026 has only amplified pessimism for a timely next-generation launch
.
Because of these conflicting signals, the gap between AMD and Nvidia's launch windows is impossible to predict with certainty. Both are nominally targeting 2027, but AMD's mid-year and Nvidia's late-year targets remain speculative.
Understanding the stakes for RDNA 5 requires looking back at the strange market context RDNA 4 has created. AMD launched the RX 9070 XT and RX 9070 in March 2025, deliberately skipping the ultra-high-end class in favor of volume and value at the $600 price point . CEO Lisa Su framed this as a strategic focus on where most gamers actually buy, not on the $1,000+ tier
.
By many commercial metrics, the strategy worked:
And yet, according to market research estimates, all this momentum failed to shift the overall picture. AMD's share of the discrete graphics card market reportedly collapsed to an all-time low of just 8% in Q1 2025, down from 15% at the end of 2024 .
This paradox is explained by Nvidia's overwhelming dominance in the pre-built and OEM desktop and laptop market—a segment where AMD has little presence—and by Nvidia's stranglehold on the data-center GPU market, which fuels its brand perception and economies of scale . RDNA 4 sold incredibly well in the DIY retail channel where it competed, but it fought for a shrinking slice of a much larger pie.
RDNA 5, by all indications, is AMD's plan to break out of that retail-only success story. By targeting a true flagship product to rival the RTX 6090, the company is signaling that it won't cede the high-end to Nvidia for another generation . Whether it can execute on both the performance and supply necessary to not just compete, but actually gain market share, is a question that no pre-release leak can answer. For now, gamers have two architectures targeting 2027, a mountain of rumor, and a long wait.
Comments
0 comments