# The REAL Story from NVIDIA GTC This Week!

## Метаданные

- **Канал:** Matt Wolfe
- **YouTube:** https://www.youtube.com/watch?v=JyWdgDcsmbk
- **Дата:** 17.03.2026
- **Длительность:** 10:11
- **Просмотры:** 26,379

## Описание

Sharing what I found interesting from Day 1 of NVIDIA GTC.

Discover More:
🛠️ Explore AI Tools & News: https://futuretools.io/
📰 Weekly Newsletter: https://futuretools.io/newsletter
🎙️ The Next Wave Podcast: https://youtube.com/@TheNextWavePod

Socials:
❌ Twiter/X: https://x.com/mreflow
🖼️ Instagram: https://instagram.com/mr.eflow
🧵 Threads: https://www.threads.net/@mr.eflow
🟦 LinkedIn: https://www.linkedin.com/in/matt-wolfe-30841712/
👍 Facebook: https://www.facebook.com/mattrwolfe

Let’s work together!
- Brand, sponsorship & business inquiries: mattwolfe@smoothmedia.co

#AINews #AITools #ArtificialIntelligence

## Содержание

### [0:00](https://www.youtube.com/watch?v=JyWdgDcsmbk) Segment 1 (00:00 - 05:00)

So, I'm out here in San Jose, California. This week is all about Nvidia. Nvidia's GTC conference is happening right now. This is where they make some big announcements about what they've been up to. They have this whole expo hall where all the companies that work with Nvidia come out and show off what they've been up to. I've heard it described as the Super Bowl for AI. I've also heard it described as the Burning Man for AI. And I wanted to make a quick breakdown of the stuff that I think you'd probably care about. Now, a lot of what happens during GTC is really focused on like enterprise and companies building data centers and for the most part not really for the common folk, but there are a handful of things that they did announce and a few like general themes that sort of bubbled up that I wanted to share and talk about real quick. So, this is really the things that I think normal people that aren't trying to build their own data centers would probably care about. And let's start with OpenClaw. If you've been paying any attention to the AI world over the last like two months, OpenClaw is everywhere. You can't get away from it. If you're on X, if you're on YouTube, if you're on pretty much any social platform, people are talking about OpenClaw. If you're not familiar, it's basically like an open- source project that allows you to turn any AI models into agents that have memory, that can use tools, that can go off and act like actual human workers on your behalf. And during Jensen's big keynote at the beginning of this event, he dedicated an enormous amount of time to talking about OpenClaw and how it was such a pivotal moment. Now, the reason they brought up OpenClaw was because they actually released something called Nemo Claw. And what Nemoclaw is a super simple setup process to install OpenClaw yourself. You type like one line of code into your terminal and OpenClaw just installs. Now, the big fear that people have around OpenClaw is there's a ton of security issues, right? You're giving it access to all sorts of things and people are worried that it could leak some of the access to your API keys or passwords or bank details or things like that. Well, this Nemo Claw, one of the things that they rolled out with Nemo Claw is an additional security layer that supposedly secures a lot of that stuff. So, if you install OpenClaw using Nemo Claw, not only is it like a super instant, super easy setup, but it also adds that additional security layer. So, anybody who's been worried about actually installing Open Claw because of the security issues, supposedly this solves that. So, if you're somebody that's been sort of on the fence about getting on board with OpenClaw, this potentially could ease some of the fears that you might have about finally testing it. Now, as a side note, they had a little like offshoot event called Build-A-Claw where you can actually go and have professionals help you set up OpenClaw. I actually brought my DGX Spark Box out with me to this event because I still haven't set up OpenClaw on this device yet. So, I took it over to the Build-A-Claw and they actually installed OpenClaw on this thing in about 2 minutes. It was insane. They typed like two lines of code and it was just like on here and working. They also installed their new Neotron 3 Super 120 billion parameter model on this. So the AI that runs on here can actually run completely on device. I don't even have to send my prompts to OpenAI or Anthropic or any of those companies. I can actually have a really smart AI running fully on device because of this Neotron 3 Super that they installed on here. And one thing that they did show off in the keynote is that Neotron 3, that open weight model that they installed is actually pretty close to state-of-the-art. You can see that there's like two enthropic models and one OpenAI model that beats it, but it's like right up there with those models and it runs locally on this device. I don't even need to tap into a cloud. That's pretty sweet. Peter Steinberger himself, the guy who actually created OpenClaw, stopped by the Build-A-Claw event and was like immediately mobbed. I would have loved to say hello and chat with him, but like apparently everybody else had that same idea. Now, another thing that I think some people would be interested in is the new DLSS5 that they talked about. This is probably more interesting for gamers, but with this new DLSS5, what it appears to do is actually run an AI upscaler on your game while you're playing it. So, games that already exist, if you're using this DLSS5, from my understanding, now I'm not 100% sure. I haven't dug too deep into this, but from my understanding, if you've seen those AI upscalers that will take, you know, sort of grainy pixy footage and then upscale it so it looks like polished and much more realistic. This apparently does that in real time on your game. So games that already exist can get better graphics than they have now. That seems pretty sweet. Now, I know a lot of people don't really like the DLSS. The DLSS, I think 3. 5, when that one came out, they were adding extra frames in between each real frame. So, let's say a game was 60 frames per second. For every one frame, it would add three AI generated frames, effectively making the game look like it was running at 240 frames per second. But some of those extra frames would kind of hallucinate sometimes and add extra artifacts. And a lot of people didn't like that. And I imagine with this new like upscaler in DLSS5 that

### [5:00](https://www.youtube.com/watch?v=JyWdgDcsmbk&t=300s) Segment 2 (05:00 - 10:00)

they're going to put on top of these games, you might get some of that kind of stuff. And some, you know, hardcore gamers are not going to like this, right? It might actually hallucinate some details into the graphics that weren't actually there. I don't know yet. But the people in the room at GTC seemed to love what they were showing off. the sentiment online among gamers seem to all not really be a fan of it. So, we'll see how it plays out once it's more in like wider production and actually being used in games. Now, I was going to go walk the showroom floor and find some of the coolest booths and talk about some of the most interesting things I saw in some of these booths. Uh but yeah, it was an absolute mad house. I could not even walk around on the showroom floor. It was such a cluster. So, I walked around for like 2 minutes and went, I got to get out of here and I bailed. But it was just insane. There there's a lot of booths there, and I am interested in learning more about what some of these companies are doing at the booths, but I need to find a time to go back when there's not a million people all in the same place at the same time. Now, there was another announcement that I thought was pretty cool, but also still feels pretty far off, and that was the Nvidia Space 1 Vera Rubin module. So, they actually showed off what a data center in space could look like, but they also admitted they still haven't quite figured out how to dissipate the heat off of these GPUs. A lot of people are saying it's really cool the idea of putting data centers in space, but because of the heat dissipation issue. I mean, these things get really hot. The sun also creates heat, and they haven't quite figured out a way in the vacuum of space to sort of dissipate the heat off of these GPUs yet. and he even admitted as much. But they did say they have their best engineers working on it. I don't know if we're going to see it, you know, before 2030, but would be really cool if they did. And you know, they've now got a name for it, Nvidia Space 1. So, we'll see how that one plays out, but that's something that I think is generally interesting as well. There was also two major themes that I noticed during the keynote. One of them was how insanely integrated Nvidia is into everything. They work with every single AI cloud provider. You name it, they work with Google. They're working with AWS. They're working with Oracle. They're working with Microsoft. They're working with Coreweave. You name it, they have their GPUs working with those companies in some way, shape, or form. Almost every company you probably work with, any sort of SAS company, any company that has AI baked into it, is probably being touched by Nvidia in some way or another. They're also in like every single vertical now. automotive, financial services, healthcare and life science, industrial, media and entertainment, quantum computing, retail and CPG, consumer package goods, robotics and telco. They are in all of them. And then the final theme that I got from this keynote was really just about the rapid rate at how fast everything is accelerating. They kept on showing different charts of how fast the tokens per second are increasing, how big the token context windows have gotten in such a short amount of time, how costs are coming down, how energy usage is coming down. Everything is scaling so fast. Anyway, this is GTC day one. I am going to see Jensen speak again tomorrow in a little private Q& A. I'm going to get a tour of some of the booths tomorrow. I'm also going to get a ride in a new autonomous vehicle that they want to show me. So, I'll probably be doing another update tomorrow about the cool things I came across in GTC day two, but wanted to share a quick update because I spent the whole day watching Jensen's keynote, which went for like two and a half hours. That guy loves to just talk. He loves to make huge presentations. And then I went over to the build the claw event, and hung out with the people who built OpenClaw and then had them install it on my device for me. I hung out with Ray Fernando and Alex Finn and Matthew Berman and it just all sorts of amazing creators here. I went over to the expo hall and I wandered around there for a little bit but got a little claustrophobic because of how crowded it was and bailed and then went, you know what? I'm going to go back and share a quick recap about my GTC day one and what I think most people would actually be interested in from this event. Not the big data center buildouts, not the big enterprise products, but the stuff that average everyday people might care about. And for me, that was making it really easy and much more secure to use. Open claw. If you're a gamer, DLSS5, that looks pretty interesting, and it might make our games look a little bit better. The Nvidia Space One module, I mean, come on, data centers in space sounds pretty sci-fi and cool to me. And then just the two overarching themes of Nvidia is in everything. They are like the most dominant company in the world for a reason. Every single company seems to touch them in some way. And then finally, just them showing off the insane rate at which all of this is scaling, how fast moving. It is just absolutely mind-blowing. So, that's a quick recap of my day one of GTC. I'm going to do my best to do a day two recap. Can't guarantee it. It might end up being a day two and three recap, but that was day one. And that was the big keynote day where most of the announcements came out and those are the ones I think you'd care about. And if you want to stay looped in on all the latest AI news, all the coolest AI tools, and just be genuinely looped in

### [10:00](https://www.youtube.com/watch?v=JyWdgDcsmbk&t=600s) Segment 3 (10:00 - 10:00)

on everything that's going on, make sure you like this video and subscribe to this channel. I'll make sure more videos like it show up in your feed. Thanks so much for hanging out with me and nerding out with me today. Really appreciate you. Hopefully, I'll see you in the next one. Bye-bye.

---
*Источник: https://ekstraktznaniy.ru/video/15038*