NVIDIA RTX 60 "Rubin" Leak: Next-Gen GPUs Slated for Late 2027 Release
While the tech world is buzzing at CES 2026, NVIDIA’s roadmap has taken a surprising turn. With rumors of the RTX 50 SUPER series being sidelined due to global memory shortages, attention has shifted to the next frontier: the RTX 60 series, powered by the ambitious "Rubin" architecture.
The Rubin GR20x Era: From AI to Gaming
According to the reputable hardware leaker kopite7kimi on X (formerly Twitter), NVIDIA’s next-generation consumer GPUs will skip the iterative Blackwell refreshes in favor of the Rubin architecture.
The gaming lineup is expected to utilize the GR20x family of chips (including the GR202, GR203, and GR205), succeeding the Blackwell "GB" series. While the Rubin architecture was originally designed to dominate the AI and server markets, NVIDIA is now adapting this "neural rendering" powerhouse for the consumer market.
Monstrous Specs: A Leap Beyond Blackwell
The server-grade Rubin chips already showcase the raw potential of this new tech:
AI Performance: Delivering up to 50 PetaFLOPS in NVFP4, making it roughly five times more powerful than Blackwell for AI-driven tasks.
Cutting-Edge Memory: Utilizing HBM4 memory with capacities of up to 288 GB for enterprise models. While consumer cards won't feature HBM4, they are expected to benefit from the architecture's massive improvements in memory bandwidth and efficiency.
Neural Rendering Focus: CEO Jensen Huang hinted at CES 2026 that the future of the RTX series will rely less on raw rasterization and more on AI-generated pixels.
Why the Wait? 2027 and the Memory Crisis
Despite the excitement, gamers will need patience. The RTX 60 series is not expected to debut until the second half of 2027.
Two major factors are driving this timeline:
Memory Shortages: A severe global shortage of GDDR7 and high-bandwidth memory has spiked production costs, leading NVIDIA to reportedly cancel or postpone the "SUPER" variants of the current RTX 50 cards.
Architecture Shift: Moving to Rubin (likely on TSMC’s 3nm or advanced 2nm processes) is a massive undertaking that requires a longer development cycle to ensure "day-one" stability.
