Nvidia RTX Spark contra Apple Silicon: pot funcionar?

RTX Spark. RTX Spark. Speaking of big, wait a second, that's not in California. Don't you know RTX Spark in California? Okay, so Apple, a number of years ago, 2020, released the M1 series processors. Big deal. Big change. ARM based, change the world. Okay? Been chasing it that high ever since. Everyone's been chasing that high. Microsoft has been chasing that high. Qualcomm's been chasing that high. They can never seem to to do it because Windows just sucks. Unfortunately, it is truly not good because they need backwards compatibility for like 100 years. So Nvidia Multi trillion dollar company. What do they do? Have you heard of them? You might not have heard of them. They're the big bear in the room. They're the biggest bear in the room. What they did was they made a competitor to the M1. It's called the RTX Spark superchip. It's effectively a laptop slash, you know, desktop PC version of the DGX Spark from last year, which was a personal dev kit, AI box that combines cpu, GPU all in one chip. Now they're going to put this thing in laptops, which is pretty crazy. We don't know really any performance, we don't really know any pricing.

We just have numbers. We have a lot of numbers, David. It has all day battery life up to up to all day. We'll see. I mean it is a 3 nanometer process, so maybe it'll be okay with efficiency. But yeah, it's got 20 CPU cores, 6,144 GPU cores. They say it can have up to RTX 5070 equivalent graphics. We don't know if that's the 5070 on the desktop or the 5070 on the laptop, which are very. When they don't specify, you can probably assume it's the lesser. The lesser probably assume up to 128 gigabytes of unified memory. But it's starts at 16 gigs of unified memory all day. Battery life built in partner built in partnership with MediaTek. One petaflop of AI compute 600 gigabytes per second bandwidth. Or maybe that's gigabits. I didn't, I'm not sure if I read it correctly. Gigabits. Jensen talked a lot about this being a platform for agents. He said agents a lot. Okay? He used that word many times. He said there are only a billion people but there are so many more agents. I'm serious about this.

Very carefully. I also don't know why he said only a billion people because I think there's about 8 billion people. Yeah, but yeah. Okay, my take on this, let's go. Based on what I've seen. Yeah. You know, we got this announcement, we got them focusing very heavily on AI and agents and how, you know, with up to 128 gigs of unified memory, you can run all these local large language models and do all this stuff and execute all these tasks and have even these optimizations. They talked about Adobe with like special versions. Obviously you have CUDA cores because it's Nvidia, but special versions of this new Adobe software where you can have it make stuff for you in the creative suite. Crazy agents happening on your computer. My take basically, which is probably based on how little time I spend in this PC world, is the window for how good or bad this could be. Oh, good one enormous. No pun intended. But the window for how high the ceiling could be or how low the floor could be is huge. Yeah, I think how bad it could be is okay. We don't know how bad like the base chip is. Okay. Starting at 16 gigs Unified Memory, you got a bunch of cores and obviously a whole bunch of bandwidth, 3 nanometer process and then it's just, you know, base just replacing the intel chip that you would have had level and that's okay. And maybe it's a little too expensive. The ceiling is like it is as good as the Apple silicon chips and it's better because of all the optimizations and the CUDA cores and it has all this local, you know, processing available, all this AI for all the agents you're going to run. So it's better for that sort of stuff. So the, the window and obviously it runs Windows. So like however into that you are the window for how good or bad this could be is huge to me. So I am, I'm going to wait to actually get my hands on it. They announced a whole bunch of laptop OEMs are going to be making these RTX Spark laptops and they're all very thin and very powerful, very premium looking. They all scream high price.

Yeah. But they haven't announced any specs or prices or benchmarks yet. So I'm just going to wait until we get some in hand and actually use them, benchmark them and see how good they actually are. The DGX Spark last year was announced at $3,000. When it came out it was $4,000. And then within a few months it became $4,600. And then for some reason the Dell version was $6,300. So yeah, they're probably going to be pretty expensive. Again, we don't really know anything about battery life. The GTX spark pulled 140 watts from the wall. Hopefully it doesn't pull that much. It's going to be scaled down a lot. Yeah. Linus is saying something about 150 watts and then the battery life not that less like two hours. Yeah, he was like, the math isn't mathing here, but like all of it. Because this is all at Computex, right? Yeah. And like, well, there was Computex and then there was Nvidia, had their own thing. And then there was also Microsoft build, which happened like right after. Okay. Yeah, yeah. I mean they announced in it the Microsoft Surface Ultra, which is a 15 inch mini LED touchscreen, you know, the largest haptic trackpad in a Surface ever. One thing that was exactly like a

MacBook Pro, by the way. I saw that it also just looks exactly like a Microsoft Surface laptop. It's looked like, except maybe a tad boxier, like thicker and less of a wedge shape. But I just think it's looked like a Surface for like that for a while. The Verge said that like looked exactly like a MacBook. And I was like, I thought it looks just like a. More like a MacBook. But maybe I'm wrong. Surface Ultra is also a funny name. I mean, yeah, huge trackpad, black keyboard, thin bezels. One thing I thought was interesting that Tom Warren said is so it's got USB C hdmi, full size SD headphone jack, but there's an USB C port on the right that he says looks a little bit larger. And when he asked the Microsoft employee, they just smiled and said they'd have more to share later. Which to me sounds like service Connect port that you like a fast charger, I guess. Like, remember how the old Surface devices had that like detachable one? It was essentially always like this weird, like long, I don't know, flat piece that would go into it and be like a, a proprietary fast charger. But you could also charge with USBC in a different port. Yeah, yeah, that was my guess too.

It might be a breakaway USBC like port. Interesting. Yeah. So it could, it could have the magnetic capabilities also be usbc, which would be cool. Yeah, I think it's just charging. Yeah, I don't know. I don't think it's anything that cool. Despite him smiling and laughing and Being like, you'll find out later this year. Yeah, yeah. This is all. You know, this is Windows and ARM again. We've seen it. Surface has done it. With Nvidia though, being at the forefront and worth a pissload of money. Yeah. There is a higher chance for sure. There's a higher chance this works because Nvidia has a lot more to lose than like Microsoft and Qualcomm did. Yeah. The thing that people are really upset at though is that there is not Linux support right now. And many, many developers use Linux and so everyone being forced to use Windows is not making people happy. Yeah, that could be a future thing. I mean I expect them to listen to their audience at some point and maybe expand support. But yeah, the response to Linus seemed very like we are not thinking about that right now. We have a lot of other things to worry about before we're going to worry about Linux.

Totally fair. They definitely had to launch this first and make sure it goes smoothly before doing anything else. Yeah. So there's like a million questions that we're gonna not have answered until we actually get units of these things in house, which is later this year. Like we're a while away from figuring any of this stuff out for sure. Maybe before Siri. We'll see. Place your bets. Before GTA 6. Yeah, before GTA 6 or my street Light Manifesto album. I'm weirdly like excited about this because there's a lot of like advanced high production creative workflows that more and more people are adopting. We were talking about this in the car this morning, David. Adopting these very slim local models to take care of one task in a really complicated production workflow with all the extra ports and the big trackpad on this Microsoft Surface. I could see this being a weirdly useful laptop with the asterisk being if this Nvidia graphics engine has access to all the like Nvidia specific graphics processes. Yeah. Jetson made a big deal about the fact that this chipset supports literally everything Nvidia has ever made.

So the idea like these are like very hyper specific applications. But I could see a lot of like post production workflows or like computers that run concert visuals or like things like that being able to do like really unique powerful things on this laptop. If it's like not which unfortunately like it seems like they all end up being. I don't think this is going to be. I don't think so either. It's supposed to run through Windows though. And as a Windows user we had to do this thing With a Dell monitor for a video we're making where it has a KVM and I'm switching between it. Setting up the Dell drivers for the KVM of this Dell monitor took me longer on the Windows PC than it did on the Mac for just like, it kept failing and then just extracting it and actually downloading it literally took like 10 times as long. And I was messaging Mark as like, I'm losing my mind here. It was a Dell XPS laptop I was using. That was like hurting. Trying to install Dell drivers. Yeah, it was infuriating.

Damn. I want to be optimistic. I want to. I want to see when they come out, hopefully we get these, these benchmarks going. Yeah. Thank you for the big trackpad. Yeah. All laptops need a giant trackpad. I agree, I agree. The question sort of becomes like, Apple has become the de facto kind of AI agent computers because they're like the best bang for buck right now. But if another company comes in and is optimized for that stuff, are people going to start moving over to, you know, agent specific hardware? If it's built for whatever you plan on doing. Yeah, whatever that means. And can they safeguard? Like, I think another reason the Apple platform is so popular for vibe coding and agents and stuff like that is because it's. You can really monkey around a lot in Terminal and with bash commands and stuff and feel confident you're not going to delete some system file that renders your computer completely broken in a way that I feel a lot less confident on a Windows system. Satya did specifically say they're bringing a lot of that stuff to Windows now, though. Like, they're bringing Homebrew to Windows. He announced that.

What? How? He announced that a build. It's a how. How? Ask him yourself. That doesn't make sense. I thought Homebrew was a Mac thing. I thought it was a Mac repository. He said, we're bringing a lot of your favorite Mac stuff to Windows so that you. So that you can. Oh, how the tables have turned, my friends. For my whole life. It was like, oh, this is really sweet, but you can't do it on Mac, so I gotta own a. Oh, I love this program, but it's. It's Windows. Oh, yeah, yeah. Now look at the Windows crowd, baby. Look at them crawl over to our side, the turntables. Well, we won't know about that stuff until later this year, so a whole lot of hoopla, hoopla, hoopla. But while we're waiting for later this year to roll around, there's a Thing that happens every single gosh darn week, babe. Right on time. And it's hoopla. Oh, trivia spelled hoopla. This was another great Ellis question that I'm just gonna steal from him. So while Google didn't stick around, one feature built into Google eventually became a standalone product in 2013. What product is that?

Two Google questions, one episode. I actually say that again. I think I actually know that Google. Say it again. I'm gonna get this one right. While Google did not stick around. Correct. One feature built into Google eventually became a standalone product in 2013. What product was that?