NVIDIA Drops Biggest Clue Yet Of When It Will Launch The GeForce RTX 5090

We know with reasonable certainty that NVIDIA is planning to launch a new flagship graphics card in the not-too-distant future, and very likely the GeForce RTX 5090. There are too many leaks and rumors to suggest otherwise. What we don't know yet, at least not officially, is when exactly the GeForce RTX 50 series will be announced. That said, NVIDIA may have just tipped us off.

Over on X (the social media site previously known as Twitter), NVIDIA revealed through its official account that company co-founder and chief executive officer (CEO) Jensen Huang is scheduled to deliver at keynote at the Consumer Electronics Show (CES) next month. The keynote will commence on Monday, January 6, at 6:30pm PT (9:30pm ET).

"We are thrilled to welcome Jensen Huang as a keynote speaker at CES 2025," Gary Shapiro, CEO, CTA, said in a statement. "Jensen is a true visionary in the tech industry. His insights and innovations improve the world, enhance the economy, and will inspire our CES audience."

There's no hard and fast blueprint for these kinds of releases, whereas the timing of a new iPhone series is often more predictable. NVIDIA's GPU launches can and do take place at different times, depending on market conditions and other factors. For example, it introduced the GeForce RTX 40 series—starting with the GeForce RTX 4090 and GeForce RTX 4080—in September 2022.

A few months later, NVIDIA unveiled its mobile GeForce RTX 40 series GPUs in early January 2023, just days ahead of CES. And then at this year's CES event (January 2024), NVIDIA launched its GeForce RTX 40 Super series GPUs.

There's not much left to refresh at this point within the Ada Lovelace family, and just as former Patriots coach famously once said, "We're on to Cincinnati," NVIDIA is undoubtedly on to the GeForce RTX 50 series.

Rumor has it that the GeForce RTX 5090 will feature a GB202-300-A1 Blackwell GPU with 21,760 CUDA cores and 32GB of GDDR7 memory on a 512-bit bus. If true, then the next flagship card will sport 33% more CUDA cores than the outbound GeForce RTX 4090.

Perhaps more exciting, however, is the possible memory bandwidth. Assuming NVIDIA runs wild with 28Gbps memory chips, we'd be looking at 1.7TB/s of memory bandwidth, which would represent a substantial 68% increase from the current generation.

The other big question (aside from pricing) is the power consumption. Prominent leaker kopite7kimi is under the impression that the GeForce RTX 5090 will have a 600W TGP, which is 33% higher than the 450W TGP attached to the GeForce RTX 4090. It looks like we'll find out soon.