Nvidia, already known for producing some of the best graphics cards, isn’t resting on its laurels. Although the recently refreshed RTX 40-series is still relatively new, Nvidia is already working on its next generation of GPUs, the RTX 50-series. While a definitive release date for the RTX 50-series is still a few months away, various rumors and leaks offer insights into what we can anticipate.
The latest speculation points to a launch after CES 2025, possibly towards the end of 2024 or early 2025. Nvidia is also rumored to be scaling back production of RTX 40-series cards, hinting at the imminent arrival of the RTX 50-series.
However, reports have been conflicting, with some suggesting that Nvidia might not be ready for a launch until 2025. This could give AMD a significant edge, as they are expected to launch their RDNA 4 GPUs later this year, though potential delays exist.
Despite the uncertainty, Nvidia is reportedly aiming to highlight the efficiency of the RTX 5000 series at CES 2025.
In terms of release strategy, Nvidia might follow a similar approach to the RTX 40-series, launching the RTX 5090 first, followed by the RTX 5080 a few weeks later.
However, other sources suggest that Nvidia might return to its previous strategy, releasing the RTX 5080 first, followed by the RTX 5090. The launch of budget-friendly cards could be delayed until early 2025.
The pricing of the RTX 50-series is purely speculative at this point. Nvidia’s pricing strategy for the current generation has been characterized as ‘expensive,’ with the RTX 4090 exceeding $2,000, despite an initial launch price of $1,600. While the price cut in the RTX 40-series Super refresh offers some hope, the RTX 5090 is expected to be priced in the $1,800 to $2,000 range, leading to potential price increases across the entire lineup.
Nvidia hasn’t officially confirmed any specifications for the RTX 50-series GPUs, but various leaks and rumors offer a glimpse into what we might expect. The successor to the Ada Lovelace architecture is rumored to be named Blackwell, after American mathematician David Blackwell. It’s believed to be manufactured by TSMC using a 3nm process, though the possibility of a custom node or TSMC’s 4NP (4nm Nvidia Performance) node remains.
The lineup is expected to include chips ranging from the high-end GB202 (equivalent to the RTX 4090) to the entry-level GB207. This could mean that the AD104 GPU powering the RTX 4070 might not have a direct successor in the next generation.
The RTX 50-series is anticipated to support DisplayPort 2.1 and HDMI 2.1, features missing from the Lovelace lineup.
The memory interface has been a subject of debate, with conflicting reports about the bus width. While the flagship card is rumored to have a 512-bit memory bus, other sources claim it will stick to 384-bit or even have a 448-bit bus. However, all sources agree that Nvidia will utilize the new GDDR7 memory standard, offering a significant upgrade over the GDDR6x in the RTX 40-series.
The amount of VRAM is another point of speculation, with rumors suggesting up to 36GB for the RTX 5090. However, it’s more likely that the RTX 5090 will have 24GB, similar to the RTX 4090.
The number of streaming multiprocessors (SMs) is another area where information is fragmented. Kopite7kimi, a prominent leaker, suggests that the RTX 5090 could have 192 SMs, marking a 33% increase over the RTX 4090. However, it’s uncertain whether Nvidia will fully utilize all the SMs in the graphics card.
Rumors suggest that the RTX 5090, despite its expected performance, might only have a dual-slot design in the Founders Edition, a departure from the four-slot design of the RTX 4090.
The performance of the RTX 50-series is still shrouded in mystery, but rumors point towards a notable performance uplift over the RTX 40-series.
Some sources claim that the performance gain might not be as dramatic as the leap from Ampere to Ada, while others suggest a significant performance leap, particularly in ray tracing.
Clock speeds are expected to reach over 3GHz for overclocked models, a considerable boost over Ada.
Nvidia’s focus on ray tracing and path tracing is expected to result in significant performance improvements in these areas.
The power draw is another critical factor. While the RTX 4090’s power consumption was initially rumored to be monstrous, it ultimately reached 450W.
Nvidia is reportedly planning to introduce a new 16-pin connector for the RTX 50-series, dedicated to 12V power delivery. However, other sources suggest that this might not happen, with Nvidia possibly sticking with the current 12VHPWR connector.
The RTX 50-series for laptops is still shrouded in secrecy, with limited information available. However, a leak from Clevo, a Taiwanese laptop maker, suggests that Nvidia will launch six mobile GPUs in 2025.
The leak also indicates that Nvidia will not be discontinuing older low-end GPUs, with the RTX 4050, RTX 3050, and even the RTX 2050 continuing to appear in laptops.
Nvidia is still keeping the architecture of the Blackwell chips under wraps, but the company has discussed the architecture for its data center Blackwell GPUs.
The enterprise version of Blackwell is built on TSMC’s 4NP node, a 5nm process. The B200 GPU features a dedicated decompression engine, which could potentially benefit consumer GPUs.
Rumors suggest that the RTX 50-series will introduce significant improvements in path tracing and ray tracing, with potential architectural changes, including a redesign of Nvidia’s SMs.
While Nvidia initially considered a multi-chip module (MCM) design for Blackwell, they ultimately abandoned the plan due to challenges such as high prices and latency between dies.
The B200 data center GPU features reworked Tensor cores that support FP4 and FP6 numerical formats for AI inference natively. This could potentially be incorporated into consumer GPUs.
In conclusion, the RTX 50-series is shaping up to be an exciting generation of GPUs from Nvidia. While details are still scarce, leaks and rumors suggest significant advancements in performance, efficiency, and features.
The release of the RTX 50-series is eagerly anticipated by gamers and tech enthusiasts, who are looking forward to the next level of graphics performance and immersion.
