Subscribe
Share
Share
Embed
Chaos Lever examines emerging trends and new technology for the enterprise and beyond. Hosts Ned Bellavance and Chris Hayner examine the tech landscape through a skeptical lens based on over 40 combined years in the industry. Are we all doomed? Yes. Will the apocalypse be streamed on TikTok? Probably. Does Joni still love Chachi? Decidedly not.
[00:00:00.27]
Ned: Our dog has a fixed schedule in her mind, and she will brook no alteration of that schedule. She does not give a shit that you got three hours of sleep last night and you just want another half hour, please. There is whining, there is begging, there is bawling at you eventually.
[00:00:22.08]
Chris: And that's just you trying to get the dog to be quiet.
[00:00:29.28]
Ned: Hello, alleged human. Welcome to the Chaos Lever podcast. My name is Ned, and I'm definitely not a robot. I am a real human person that requires an adequate amount of sleep. And no, five hours does not quite cut it. With me is Chris, who is possibly more rested than me. I don't know, and I don't care. Hi, Chris.
[00:00:58.09]
Chris: Oh, Oh, shit. What you're saying is you should have had a five-hour energy, and then you would have had the equivalent of 10 hours of sleep.
[00:01:10.06]
Ned: That's how math works. Yeah. Absolutely.
[00:01:12.06]
Chris: That is how it works.
[00:01:14.00]
Ned: Completely adds up to me.
[00:01:16.00]
Chris: I mean, 5-hour energy is not a complete scam. That's really just a blast of B vitamins. It gives you the illusion of energy for a short period of time.
[00:01:23.04]
Ned: Oh, it's got caffeine. Don't forget about the caffeine.
[00:01:26.07]
Chris: Caffeine doesn't give you energy, Ned. Caffeine stops you from feeling tired.
[00:01:32.17]
Ned: That's better. It's different. It combines together, what you're saying, is I should go chug some caffeine and B vitamins.
[00:01:43.03]
Chris: It would save you approximately 99 100%, yes.
[00:01:46.14]
Ned: All right, I'll be back. I don't have any vitamins. I don't even know where I would get those. Vitamin store? Vitamin Shop? Not a GNC.
[00:01:56.27]
Chris: That is a real place.
[00:01:58.15]
Ned: No, it is. I don't think GNC sells anything that's real.
[00:02:03.06]
Chris: Creatine. I think that's about it.
[00:02:05.05]
Ned: It's a real thing. I don't think it'll really help you. I don't know. Apparently, this is now a fitness podcast, and we're going to rag on vitamin shops and GNC for the next hour. So welcome to our new podcast. Now, we're going to talk about something else. It's our predictions review. This is a thing that we do annually, where we made predictions at the beginning of the year, and now we review those predictions and see how we did. I did great. I was right about everything. Chris, bless your little heart for trying.
[00:02:46.08]
Chris: I've never not been wrong, or haven't I? True. Okay.
[00:02:54.06]
Ned: I don't know how to refute that, so I'm just going to move right along.
[00:02:58.04]
Chris: That's the spirit.
[00:02:59.20]
Ned: So I figured what we would do is we would trade back and forth. I think that's what we did during the predictions episode is we did one of my predictions and then one of yours. We'll do the same thing here. I'm going to summarize my prediction and then see how it did. Sound good?
[00:03:17.26]
Chris: Actually, I will read your summary of your prediction, and then you respond to it.
[00:03:22.12]
Ned: How dare you? Okay, fine. Go ahead. Okay.
[00:03:26.29]
Chris: I'm going to do my best net impression, and we're going to start with We'll just start at the top. We'll start with number one. Okay. Broadcom has no interest in catering to the SMBs of the world or sustaining product lines that don't directly contribute to the cash cow that is V-Spear.
[00:03:44.25]
Ned: I think you're a little bit off.
[00:03:48.01]
Chris: I'll work on it.
[00:03:49.12]
Ned: You should probably lead with what the actual prediction was, which is that Nutanix will finally be purchased by a responsible adult.
[00:03:57.01]
Chris: Shit, I missed the line.
[00:03:59.03]
Ned: It's okay. Well, we absolutely won't fix it in post. How did I do? Well, the first thing that I was saying is basically, Broadcom is not going to cater to SMBs, and so someone's going to buy Nutanix and market Nutanics to those SMVs. Well, I was mostly right about the fact that Broadcom doesn't care about the smaller customers. Although as As we covered on Tech News of the Week, I think last week, this attitude seems to be softening a little bit as Broadcom realizes that large customers might actually leave and smaller customers, customers, Custards. Custards? Smaller customers might actually matter a little bit. So they are engaging with partners again and introducing a SKU for SMBs. Then you What?
[00:05:00.25]
Chris: Oh, you're not done? I thought you weren't done.
[00:05:04.12]
Ned: Well, okay. So the next thing that I predicted, go ahead.
[00:05:09.26]
Chris: I won't get it. I only have one voice, and it gets very annoying even for me. Okay. The next thing you said was that you would not depend on any major product improvements over the next 3-5 years, which technically broke the rules.
[00:05:25.06]
Ned: I suppose. But okay. What I was positing is that Broadcom has no interest in vastly improving the product. They're in rent extraction. For sure. Vsphere 8 is the most current version, major version of the software and that was released in 2022. Now, that by itself is not an indictment. It's been two years, and the fact they haven't released a new major version, usually they take between three and five years between their major releases, if you look at the history. What about update releases? Because those are the minor patches. Update 3 came out in June of 2024, and it brings with it, ESXI Live Ratching, enhanced AI support, dual DPU support, and an IAAS control plane for developer self-service.
[00:06:24.20]
Chris: Which sounds dirty.
[00:06:27.05]
Ned: If that tickles your fancy, I mean, great. But none of this is what I would call an Earth-shattering introduction of a feature. This is no NSX, this is no vMotion. It's just incremental improvements, mostly based off of hardware.
[00:06:46.03]
Chris: Then, blissfully, blessedly, we actually get to the prediction itself.
[00:06:54.11]
Ned: Okay.
[00:06:55.14]
Chris: Wherein you said that you think some large company will scoop up Them.
[00:07:01.29]
Ned: Nutanix.
[00:07:03.04]
Chris: Right. And aggressively promote them as a VMware alternative. Who might do such a thing? Cisco. Now you got in trouble because you said it's going to be Cisco, probably Q2 or Q3.
[00:07:21.22]
Ned: Well, I was wrong. As of Q4 of 2024, Nutanix is still their own company, and their stock is up 40% versus this time last year.Wow.It's.
[00:07:36.25]
Chris: Pretty good.
[00:07:37.12]
Ned: Yeah. Is this due to the VMware alternative campaign they've been running? Yeah, probably. Cisco and Nutanix have continued their global partnership, and Nutanix has also struck a new partnership with Dell in May of this year to sell a package solution through Dell's channels. Dell and VMware's relationship relationship has become strained in the last few years, and this is a further signal that there is distance between the two. All that being said, I still think that someone will buy Nutanix. I just got the year wrong, and now they're going to pay a hefty premium.
[00:08:19.01]
Chris: Yeah. Even with all of the benefits that Nutanix reaped, not necessarily of their own design, I still feel like they missed it a little bit. I still feel like they could have It's done better. Nutanix on its own is still expensive. It's not like a bargain basement product. It's cheaper, but if it were cheaper-er, they could have done better-er. I went to business school.
[00:08:49.13]
Ned: Yes, you did. As did we all. Hooray. All right, let's move on to one of your predictions, which was AI-related. Surprising. Ai is in deep shit for copyright infringement, aka Disney Smash. So what happened? Well, there was a release of info from a New York Times-based lawsuit that showed OpenAI products returning results that were verbatim copies from the New York Times and other sources. That was bad. There has also been a spate of people posting online, showing how even generic prompts are returning, shall we say, recognizable images from DALL-E. So You said, I bet we're going to be finding out. The prediction is said companies will be raining legal hellfire on OpenAI and other AI companies until this problem goes away. What happened?
[00:09:49.01]
Chris: Well, this is America. In America, the arc of the courts is slow but always bends toward... God, I can't even get through a fake without feeling sad. The issue with saying whether this is a win or a loss in the predictions category is really pending, no pun intended, because these cases are real and they are still going. There was a big win for the plaintiffs in one of the cases in August, however, as a judge rejected basically all of the major AI company's attempts to have the case thrown We won big, said artist Carla Ortiz. In the American justice system, that means that the case, which began in January, was determined 10 months later to instill It will indeed still be a case. That's how far we've gotten in 300 days. Now, the other case, the one that I talked about more specifically about the New York Times and stuff. That's where a bunch of the newspapers are suing OpenAI. That opened in December of last year and is also still ongoing. Openai is aggressively defending themselves in that one. In a little bit of Now, for the publicity stunt/slap in the face, OpenAI did allow plaintiffs to do searches of OpenAI's training data on these really specialized, isolated virtual machines.
[00:11:28.05]
Chris: Then they accidentally deleted that data. Meaning that all of the plaintiffs work to try to prove where this stuff came from, what they're using, whether it was copyright infringing. Yeah, they're just going to have to do that again.
[00:11:41.00]
Ned: Wow.
[00:11:42.11]
Chris: This shows either shocking ineptitude or outright malice on the part of OpenAI. I can't tell which one. Either way, I'm sure they build $500 an hour for it.
[00:11:52.09]
Ned: Absolutely.
[00:11:53.15]
Chris: In court, OpenAI actually blamed the plaintiffs for the screw up. In either In the case, the court did nothing about it, which just reinforces the integrity of our feelings about the judicial system, yet again, for the 100th time.
[00:12:11.19]
Ned: It's been a good year for the judicial system.
[00:12:14.28]
Chris: Not a good year for this prediction, although it's not my fault that it's not resolved. So I don't know, man. I think I win this one.
[00:12:25.11]
Ned: All right, fine. I'll give it to you. All right. Next up, I made a prediction about public cloud vendor growth and a rebound in 2024.
[00:12:38.21]
Chris: Oh, you always have to throw one in that's boring. Well, last This year, you predicted that cloud vendors would see a slowing in growth, and they actually did. This year, you thought we would see growth numbers return back to 2022 or 2021 rates as market for AI booms.
[00:13:03.28]
Ned: Okay. And? Well, did we see a return of growth for Azure and AWS? Do the financials. I wish we had a cool sound effect for that. With the spinning Newspaper. Okay, fine. Anyway. Here's a brief summary. Aws had revenue increases of 17% in Q1 and 19% in Q2 and Q3. Azure had revenue increases of 31% in Q1, 29 in Q2, and 33 in Q3. And Google Cloud had revenue increases of 28, 29, and 35, respectively. How does that compare with 2023 and 2022? Aws had a big dip in 2023, dropping their increases from 20% down to the low teens. They have definitely bounced back. And by all accounts from re:Invent, AWS is out there culling some of the crappier services that they farted out and focusing on delivering primitives for people to build solutions, which is what they're really good at. Microsoft had a similar dip in 2023 and a big bounce back in 2024. It can't compare to the 40 and 50% growth numbers of 2021, but we're talking about the pandemic and the law of big numbers, both contributing to that.
[00:14:26.24]
Chris: Isn't it the law of large numbers?
[00:14:29.00]
Ned: Big, large, whatever. Google Cloud has also bounced back from a slight dip in '23 to finally break the $10 billion revenue line for a single quarter.
[00:14:42.17]
Chris: They did it.
[00:14:43.26]
Ned: Hey. He 2023 saw their biggest percentage growth since 2022 when they were booking about half of the revenue that they are now. In summary, public cloud is doing quite well, thank you very much, and has made up for dip we saw in 2023. So I guess people are AIing and stuff. Yeah, I think that's what it was.
[00:15:08.18]
Chris: I give you one quarter credit.
[00:15:13.08]
Ned: Harsh. You know what? There. What was your next thing?
[00:15:20.18]
Chris: Well, you buried... Well, wait, no, not buried the lead. Jumped in front of the head, did the cart before the tail of the- Look, man, you put these in this order.
[00:15:29.29]
Ned: I talked about AI. Of course you did. You said AI will continue its inexorable march forward toward industry domination. After a year of experimentation and wild speculation, you believe that businesses are going to get serious about implementing AI as widely as possible. You do not believe that is going to be the end all for anyone, and you also think business leaders are finally seeing that. But as an assistant, AI is going to have a meteorite rise and probably one that will have a measurable impact on things like part-time work or overseas call centers or entry-level positions. I don't really see a prediction in there, but sure. What'd you find out?
[00:16:25.13]
Chris: I don't even feel like we need to discuss this one. As predictions go, this is admittedly a bit of a layup.
[00:16:33.16]
Ned: Fair.
[00:16:34.25]
Chris: But in terms of what actually happened, it was more like a 360 slam dunk from behind the net, which you could think is only possible in space jam.
[00:16:45.09]
Ned: Yes, it is.
[00:16:47.20]
Chris: Just smile and nod, Ned. Just smile and nod.
[00:16:50.26]
Ned: I support.
[00:16:53.24]
Chris: I mean, if you think about it, this is absolutely what happened. I win, I'm the best. Microsoft has copiloted absolutely everything.
[00:17:02.29]
Ned: Yes.
[00:17:03.28]
Chris: As have all of the major search engines and even some of the awesome minor search engines like Caggy, which is something that everybody should absolutely look into. We've talked about that before. Caggy is a search engine that you actually pay for, so it's not supported by ads.
[00:17:22.27]
Ned: Actually, it sounds better.
[00:17:23.20]
Chris: It has search results that are equal to or superior to Google's. Not a big deal. Also not sponsored yet. Anyway, every single enterprise-level app is being AI washed in some way or another. I could go through the list, but honestly, it's everything. If you were listening to Tech News of the Week last time around, no, two times around ago, sorry. Google and OpenAI have announced a whole ton of new features that are intended to help you with whatever dumb meme thing that you're doing, or I guess work. Anyway, I win again.
[00:18:09.01]
Ned: You also had a bonus prediction that we are going to start seeing some move towards AI-free creations and/or certifications. You said it's getting hard on first glance to blow off photos as obviously AI made, which I absolutely agree with. And now we also have to contend with Sorra.
[00:18:30.26]
Chris: Yeah. Video is in case you didn't listen to said Tech News of the Week, which also, shame, shame, shame, shame. They're like four minutes long.
[00:18:44.27]
Ned: At two double speed, absolutely.
[00:18:47.28]
Chris: The idea here was that people were recognizing that it's getting harder to figure out what was AI created and what was not. So maybe AI creations should have some type of system or some type of rock solid way figuring out what was created by AI. It's happening, but it turns out that doing it is actually way harder than expected. Much like in the world of academia, where there's a bunch of tools out there that are supposed to help you find AI-based plagiarism, and they all stink.
[00:19:23.13]
Ned: Yeah, they do.
[00:19:25.25]
Chris: 70% failure rate, it's really, really bad out there. But anyway, they are trying. Literally this week, Facebook released a, Deep fight, Deep fight?
[00:19:39.18]
Ned: Deep fake?
[00:19:40.14]
Chris: D, that's the one. Deep fake, Fighting AI video watermarking tool, which they naturally claim is the best tool around. They are also creating a competition of sorts similar to what they do with supercomputers. There's a ranking board and a scan of the same types of tests and whatnot to see who's doing it better, faster, greater, etc. The watermarking tools can fight it out. I am glad that I used the weasel word or the weasel phrase going to start seeing because that's exactly what happened. I win again.
[00:20:22.23]
Ned: Fine. All right. Well, my next prediction was another acquisition that Microsoft will buy Firmion for an undisclosed amount.
[00:20:36.18]
Chris: I was confused about what the hell Firmion was then. I have done no learning whatsoever since. Excellent. But apparently, Firmion is at the forefront of WebAssembly tech with their spin product, and their founders are largely folks from Microsoft. Ned was betting that rather than Microsoft developing its own WASM as a service in Azure, they'll just go and buy Firmian and offer it as Azure Spin.
[00:21:06.07]
Ned: Once again, I am incorrect. At least I'm consistent about these acquisition predictions. I'm two for two. Firmion is still trucking along solo and just released version 3 of their spin open-source platform for running Wazam-based application. The platform does seem to be maturing now offers cross-language component support for component launches and better support for heterogeneous hardware environments. Microsoft, let's be honest, has been so focused on AI that they really haven't been paying much attention to anything else. That trend looks to continue in 2025. As for Firmion, I think they'll still get acquired, just not this year, and probably not next year, maybe like 2026. We'll find out when we get to the predictions episode.
[00:22:03.28]
Chris: Do you think also it could be that they just weren't shopping themselves? They weren't interested in being acquired at this juncture?
[00:22:11.11]
Ned: They might not be ready to shop themselves around. Now, that's never stopped Microsoft from just walking up and being like, That's a nice company you got there. It's a shame if we just acquired it out of nowhere. Once they back the money truck up, your investors are going to go. People's opinions.
[00:22:29.24]
Chris: Have a habit of changing real quick. Yeah, that's a fair point.
[00:22:32.27]
Ned: Yeah, because all of these are VC-backed. It's not so much up to the people who run the company as the ones who technically own the company. If they're like, You have to sell now, then you're going to sell.
[00:22:46.15]
Chris: Do you ever dance with the devil in the pale sunlight?
[00:22:49.29]
Ned: That of all my prey. Deep cut. All right. Not deep cut, but depending on the age of the people listening, it would be a deep cut. Anyway. I had a metaverse prediction, which is just fucking adorable. The metaverse makes headway thanks to Apple.
[00:23:11.23]
Chris: You can summarize this if you want. I didn't get to that part.
[00:23:15.08]
Ned: Facebook thought 2022 was going to be the year that fully immersive virtual reality worlds were going to be a thing. But as you said, we stand 700 odd days later and now about a thousand odd days Twitter and the Facebook metaverse, super not a thing. Mostly, we can get to that. But Apple did release an AR/VR headset that solves a lot of the problems that metaverse was happening. But it's important to note that headsets have existed for years and years now. What was your actual prediction out of all this, Chris?
[00:23:56.28]
Chris: Long story short, there are two I have two problems with the concept of the metaverse. The first is that everybody hates Mark Zuckerberg and Facebook, and they only use it because they have to. And deep down, you all know it's true.
[00:24:08.26]
Ned: Oh, yeah.
[00:24:10.13]
Chris: But the other part was you couldn't reliably interact with it because the equipment wasn't immersive, and it wasn't good. It wasn't good screens, et cetera. My theory was that the Apple stuff is going to come out and it's going to basically be 4K screens in your eyeballs directly, and that's going to at least solve that secondary problem, leading to either an independent metaverse of some sort, an Apple metaverse of some sort, but the hardware would lead the way was the theory. On the one hand, I was right in the sense that the Apple Vision Pro did come out. It soared through the skies on the back of endless paid shills. I mean, clearly unbiased YouTube tech reviewers. But then promptly crashed to the ground in a sad, non-user repairable pile. Also a lot of returns, like a lot of returns.
[00:25:13.07]
Ned: I bet.
[00:25:14.16]
Chris: It didn't do good as a product is what I'm saying.
[00:25:19.14]
Ned: I've heard virtually nothing about it since June, maybe?
[00:25:26.11]
Chris: No, it's a total failure. That's the only way to look It's not it. I think it's, 20/20 hindsight, there's a couple of reasons for it. Number one, you look like an idiot wearing it. Totally. The idea that people were going to wear that outside was completely preposterous. You require an external battery pack, which is not cool. Most importantly, probably from all counts, especially over long term, and by long term, I mean 15 minutes, it's just not comfortable. It's too heavy, even in the perfect circumstances. Now, what's funny is, by all accounts, the best head strap that you can get for the Vision Pro is not the one created by Apple. It was created by a company that makes masks for CPAP machines. If you think about it, there's a certain logic to this. They've got a lot of experience with very complicated, heavy, awkward material that you have to have comfortably on your head for a long period of time. Maybe that's a miss on Apple's part. Maybe they should have done something crazy, like sought outside opinions from experts who have done something like this before. But that's just crazy. I know.
[00:26:42.03]
Ned: Now, you did make an actual prediction that we would see version 2 of Google Glass in 2024. Did we?
[00:26:53.15]
Chris: I was so close. All right. I should have combined it with the above prediction about AR, VR, alternate worlds and whatnot, because that's what Google ended up doing. They released Project ASTRO last week, and guess what ASTRO requires?
[00:27:15.26]
Ned: That's it?
[00:27:17.10]
Chris: Smart glasses of some kind.
[00:27:22.20]
Ned: Okay.
[00:27:22.27]
Chris: My theory on this is that they're just holding back on this new product until both of them get, A, finalized and be tied together more closely. So I still think this is going to happen. I just had the wrong year.
[00:27:40.00]
Ned: You didn't get it right, is what I'm hearing.
[00:27:43.14]
Chris: Did I mention so close.
[00:27:49.08]
Ned: Okay. There was one other thing that I wanted to mention in this realm, and it's the fact that Facebook, beta, whatever, they did a product demo of their Orion AI glasses that you can't buy and you never will be able to buy, but they're a proof of concept that some people got to try. And by all accounts, those are, while still on the heavier side, much closer to something you would actually wear and use outside of the house.
[00:28:21.02]
Chris: Right.
[00:28:22.29]
Ned: Maybe that's a 2025 prediction we got coming down the pike.
[00:28:26.16]
Chris: I like copy paste. That saves me 10 minutes. Fair.
[00:28:32.14]
Ned: All right, let's double up my next prediction about cloud flair and digital reality and Equinox, because you don't care about revenue. So I'm just going to blow through them real quick.
[00:28:45.08]
Chris: You said that Cloudflare would hit $2 billion in revenue in 2024.
[00:28:52.09]
Ned: They did not. Revenue was projected, at least in 2023, to be $1.3 billion. If their growth continued, it should be $1.77 billion in 2024. But I thought that based off of growth and things that are happening on the edge, they would finish at 2 billion, and they didn't. So far this year, they're at $1.21 billion for the first three quarters. They're definitely going to beat out last year's numbers by about 30%, finishing around $1.65, $1.7 billion. That is a far cry from the 2 billion that I thought they would hit. However, their stock is up 34% year over year. I wish I'd put my money where my mouth was on that one.
[00:29:47.23]
Chris: Yeah, that's a tough one.
[00:29:50.29]
Ned: And I had another revenue prediction.
[00:29:53.15]
Chris: You followed that particular massive failure up with, Digital Realty and Equinox will have 25 revenue. I'm assuming that was increase.
[00:30:03.01]
Ned: Increase year over year, yeah.
[00:30:06.14]
Chris: In case people don't remember, Equinix and Digital Realty are the two biggest vendors in the data center co-location space. Yeah, you thought their revenues were going to jump 25% year over year.
[00:30:17.08]
Ned: Yeah, driven by AI. Companies that are going all in on AI. This really is the parade of me being wrong.
[00:30:27.04]
Chris: Which also, great name for your autobiography.
[00:30:30.08]
Ned: It does sum up my life in so many perfect ways. I am terrible at predicting anything, so I probably should have taken the easy way out and just predicted that AI would continue to grow, Chris, and that NVIDIA would make a whole bunch of money. But I took a big swing. Digital realty had negative growth in the first two quarters of this year. For Q3, they managed to get back in the green with 1.6% revenue growth over last year, but I'm guessing for the full financial year, we're going to be looking at flat or negative growth. Despite all that, their stock is up 40% year over year.
[00:31:14.29]
Chris: What the hell?
[00:31:17.10]
Ned: Negative growth and upf... Whatever. Anyway, what about Equinox? There's a rosier picture over there. Equinox had consistent 6.5% year-over-year revenue growth for the first three quarters of 2024, their stock is up 20%. I'm not sure what's going on here. The closest thing that I can guess is that despite companies maybe wanting to stand up private AI in those data centers, they literally can't get the hardware to do it.
[00:31:50.02]
Chris: That's part of the problem. I think the other part of the problem is that people that are doing heavy-duty AI as a service are not doing it through general products like digital equity or Equinex.
[00:32:01.09]
Ned: Right. They are building their own space and using it if they're trying to do things at that scale. There are a couple of companies that are doing that scattered across the world. Yeah, man, I was just wrong. It's fine. Let's move on to the next thing. Chris, what was your next thing?
[00:32:19.26]
Chris: Why are you asking me? You're supposed to tell me.
[00:32:21.15]
Ned: Oh, that's right. You said that PCI 6 is going to make all of my 2023 CXL predictions come true.
[00:32:32.07]
Chris: Let's see.
[00:32:33.12]
Ned: You said that PCI 6.0 is on the threshold of release, and it comes with 64 GTS. What the hell does that stand for? Giga transfers? That's what it is. For a second of with 64 billion transfer operations per second per lane, as well as new PAM for RAM, which sends multiple bits per transfer. So PCI 6 will be at least twice as fast than PCI 5, which is insane. It also comes with CXL 2.0 ability, which enables things like memory pooling through switching, which means that the RAM could be shared from outside of the server. 6.0 PCI is ready for CXL 3.0, which other devices with lesser protocols weren't at this time last year. You think the rollout of 6.0 6.0 of PCI is going to make all the difference for CXL.
[00:33:35.07]
Chris: How'd that work out? You know, listener, how you're thinking to yourself, what in the hell is CXL?
[00:33:47.04]
Ned: It's not a good sign.
[00:33:48.12]
Chris: No, no. It really just tells you all you need to know. I probably don't have to ramble any further, but I'm going to.
[00:33:56.09]
Ned: Yeah.
[00:33:58.02]
Chris: Because I don't have anything else to do. No. In terms of CXL, I will admit that I did not do a great job of keeping up with it. But unfortunately, in reviewing very quickly, the story seems to be the same old story. Although now, to be clear, the specification in question is not PCIe6, it's PCIe6.1..
[00:34:25.10]
Ned: I know.
[00:34:27.22]
Chris: That came out in January of this year, which is a full two years after the original 6 spec came out. Added a bunch of stuff to make everything fasterer. I could tell you about it, but does it matter?
[00:34:42.21]
Ned: Maybe.
[00:34:43.17]
Chris: Really, it doesn't. Because here's the other thing, did other products come out aside just from the spec itself? The answer to that is,. There's a few. Rambus, who has always been on the forefront of this CXL PCIe 6 thing, has some products out there. And VIAVI has some products that you can use to help test it. Cxl 3, though, doesn't seem to be happening. Because nobody else is doing anything with it at all. That is the problem. One major problem, which we'll tie back to all the other things we've talked about, Nvidia GPUs don't support It's important.
[00:35:31.14]
Ned: It seems important.
[00:35:32.17]
Chris: Other things like the general purpose AMD Mi300A, it has support for CXL, but it's imperfect. According to some of the things that they have been saying publicly, AMD might be done with any inter-node connection that isn't Ethernet-based. That's not good either.
[00:35:53.02]
Ned: No.
[00:35:53.27]
Chris: Because you know who uses AMD chips? Like everybody. Yeah. Tie back to a few episodes ago, the AMD Mi300A is the chip that is used in the El Capitan Supercomputer, which also doesn't use CXL. I mean, the sad truth is that it might just be that CXL didn't come around early enough, and now it just won't go anywhere.
[00:36:27.28]
Ned: Or it's just one of those things that's a little It's all ahead of its time and will grow into.
[00:36:33.16]
Chris: Hard to say. There is 800 gigabyte Ethernet in the data center at this point.
[00:36:41.08]
Ned: True.
[00:36:42.12]
Chris: Why do we need anything else is, I am guessing the thinking.
[00:36:48.03]
Ned: Yeah, and 1.6 terabit Ethernet is already in the works. Why?
[00:36:57.02]
Chris: Why do we need these speeds?
[00:36:59.16]
Ned: Because I had one last prediction, and it's also a hardware thing to close out the year.
[00:37:10.10]
Chris: And blissfully for the audience, it's a quick one. You said Risc-5 would be available in the public cloud by the end of 2024.
[00:37:23.20]
Ned: Yeah, that's what I said. And I was right.
[00:37:27.27]
Chris: It's good for me. I mean, good. Good job. Sorry, I'm not used to that.
[00:37:31.28]
Ned: I know. It feels awkward. Muscle memory. I don't know how to process these feelings, so I'm just going to read what I wrote and we can move on with our lives. There is a company called Scaleway Labs, and they offer Risc-V servers in the cloud. They're a European cloud provider based in France. You've probably never heard of them. But according to their press release in February of 2024, they are the first cloud provider to offer bare metal Risc-V servers at the low, low price of €16 a month, which is actually not bad. No. You get a four-core system with 16 gigs of RAM and a 128 big local drive. You can just do whatever you want with it. If you are interested in testing out Risc-V stuff server class hardware, you can do it for €16 a month. You can also just buy a Risc-V mini-computer, like a a raspberry pie clone for like 50 bucks. Whatever makes sense to you.
[00:38:37.22]
Chris: Okay, so I mean, it wasn't AWS or Azure, but it is public cloud, so technically correct.
[00:38:45.18]
Ned: Go sign up for an account right now if you want to and tell them that we sent you. Maybe they'll give us money. Seems incredibly unlikely, though. Those were our predictions. And overall, Chris, you did great by lobbing yourself some softballs, and I did terribly because I always do. Before we wrap up, I just wanted to mention, dear listener, this is our last show of the year. We will be back in the second week of January, I believe, with our Fictions episode for 2025, but we are taking a week off in the interim. So enjoy not hearing us, right?
[00:39:24.19]
Chris: You've earned it.
[00:39:25.27]
Ned: You have earned it. Oh, You can find more about the show by visiting our LinkedIn page. Just search Chaos Lever, or go to our website, chaoslever. Com, where you'll find show notes, blog posts in general. Tom Foulery. We'll be back next week. No, we won't. We'll be back in two weeks. You just said that. What fresh hell is coming for us in 2025? Ta-ta for now.
[00:40:01.23]
Chris: Eggnog is overrated.
[00:40:04.24]
Ned: There's a video of this guy, Adam Regucia. He's like a YouTube cook, and he aged eggnog for two years, and I cannot bring myself to watch that video.
[00:40:17.04]
Chris: Why do we, as a species, do this to ourselves?