Nvidia CES 2025 Keynote live blog: the latest on the RTX 5000 reveal and more

And that wraps up Nvidia’s CES 2025 keynote. Now if you’ll excuse me, I have to bother Nvidia PR to try and track down some spec sheets. Stay tuned for more information about CES 2025 throughout the week, including more details about the Nvidia Blackwell GPUs.

I have to say that if I have to deal with the collapse of the AI ​​model in the product I use, I’d rather not have to deal with that at highway speeds.

Hmmm. Synthetic input data has been shown to quickly degrade the quality of the model you train (model collapse), which I haven’t heard mentioned once. I wonder how Nvidia plans to address this issue.

Nvidia’s autonomous vehicle chip would be destroyed in New York traffic, I can guarantee that, although it could work in many other cities.

Nvidia RTX 5090, RTX 5080 and RTX 5070 Ti laptops will be available from March, while the RTX 5070 laptops will be available from April.

Okay, we just got word about the availability of Blackwell GPUs.

The RTX 5090 and RTX 5080 will go on sale on January 30, 2025 for $1,999 and $999 respectively.

The RTX 5070 Ti and RTX 5070 will be available in February for $749 and $549 respectively. Pricing for the UK and Australia wasn’t given, but we’ve reached out to Nvidia for clarification.

The concern with synthetic data is that you ultimately have a Habsburg AIone that has been effectively inbred to its own data to the point where it becomes a useless abomination. Google the Habsburg monarchs of Europe if you want to see why this is such an apt description of the problem.

Okay, so the end of data for training models is another big bottleneck for AI, and what Jensen is talking about here with Cosmos is generating new data on which to train subsequent models (synthetic data), since these models already have all the necessary have used up data. existing data on which it can be trained.

However.

I wonder how Cosmos will prevent the collapse of the model.

Okay, sorry about that folks, we were dealing with some technical issues, but we’re back to the action, and that action is all about tokens. It’s tokens all the way down.

Okay, the virtual human thing still creates a big uncanny valley, but it’s less severe than it used to be.

Coding assistants are the kiss of death for the junior software developer. So much for ‘learning to code’.

Oh man, I just had a dark thought. Can you imagine training the AI ​​agent that will take your job? That’s grim.

I want to say that I find this whole AI discussion interesting from an academic perspective, but I think there are a lot of expectations for these data centers, and no one mentions that the power requirements for these will pretty much set a limit. about what they can do, since we only have a limited amount of electricity available on the grid at any given time.

So we have fully entered the data center segment of the keynote. While GeForce graphics cards received at least a little more time and attention than Lovelace, it’s clear that these cards aren’t as important to Nvidia as the data center business.

And yeah, that shield piece was a bit… well, it was something.

There are also Nvidia GeForce RTX 5000 series mobile GPUs coming, with the RTX 5070 delivering mobile RTX 4090 performance, although I suspect Jensen means mobile RTX 4090 performance.

The RTX 5000 series will also be available start in January, although we don’t know which will come first.

If the shader cores can also carry the weight of AI workloads, as Jensen said, then we’ll get much better DLSS on these cards.

Okay, so very little about the specs, but I want to know more about this AI management processor.

Okay, RTX 5090 starting at $1,999. RTX 5080 for $999. RTX 5070 Ti for $749. Yes, absolutely. This is what I want to see.

Don’t get me wrong, these are still expensive graphics cards, but considering the fear of a $1600 RTX 5080, this is a very pleasant surprise.

Okay, RTX 5070 for $549, RTX 4090 performance. Wow.

Okay, so Jensen is holding the RTX 5090.

Nvidia RTX Blackwell is official, and that’s a very nice looking graphics card.

That was a pretty impressive demo.

Okay, first mention of GeForce, so here we go.

That was a very short one Virtua fighter demonstration.

LOL, Jensen’s jacket is enchanted.

Tokens, tokens, tokens. It’s no surprise that we’re diving straight into AI, but yes, it’s remarkable how much Nvidia has transformed almost overnight.

Okay, NOW let’s get started.

Okay, the Nvidia portion of the keynote is about to start, but it’s going to be a while. Nvidia normally launches faster than this.

Okay, I unironically love “Never Will You Give Up.” I used to roll to that song as a kid.

CTA President Gary Shapiro introduces Jensen Huang.

Okay, I DO want an exoskeleton. Those things look cool.

Okay, here we go.

The biggest thing I’m expecting tonight is Nvidia’s GeForce RTX 5090, RTX 5080, and possibly the RTX 5070 Ti.

As the successor to Nvidia’s Lovelace GPUs, the Blackwell-based RTX 5000 series is expected to be significantly more powerful, with the rumor mill putting the RTX 5080 around 10% faster than the RTX 4090, currently the best graphics card on the consumer market.

Of course, that would put the RTX 5090 in a class all by itself, and there’s no telling where its performance will ultimately end up. That said, if speculation is present, it should feature 32GB of GDDR7 VRAM with a memory bandwidth of 1.52TB/s on a 512-bit memory bus, making it truly the world’s first 8K gaming graphics card.

Alright folks, we’re arriving at the 15 minute mark from the start of Nvidia’s CES 2025 keynote, where CEO Jensen Huang will take over the state at Madalay Bay’s Michelob arena. We’re expecting big news tonight, so for those who have been waiting for Nvidia’s latest generation of graphics cards, you won’t have much longer to wait.

And if you really want to hear all about data center AI and Omniverse stuff, I’m sure Jensen will get to that too.