Chris: 0:14

Hello and welcome back to another episode of your monthly Cables to Clouds news update. I'm your host, Chris Miles, and my co-host joining me today. As always, there's never really never been a day that he doesn't join me on this podcast. Tim McConaughey. Not yet. Not yet, anyway. But yeah, so if you're if this is your first time listening, basically Tim and I kind of go over uh a few news articles that we found over the past four weeks that we found interesting and uh get into the details and kind of go over them. Uh this usually is consumed as kind of a uh more bite-sized version of the podcast rather than when we had guests on things like that. Um so with that out of the way, let's hop right in. Um so coming up, um, this was reported on uh 22nd of Jan. Um Amazon is looking to do the second round of their layoffs that they announced last year. So some I think about 14,000 people were laid off. Um, and I say that in terms of white-collar jobs. They have um, I think other rounds of layoffs that they're doing as well, but this is specifically around things related to people, technology, et cetera. Um so in October of last year, they announced they were going to cut about 30,000 jobs, and I think they laid off about 14,000 at that point. And I believe uh coming very soon they plan to do the other 16,000 of those, right? Um, so we're looking at um uh, like I said, a total of 30,000 jobs being taken out, which is uh pretty substantial. Um, although when when this was announced, there was commentary specifically from Andy Jassy saying that um this was not really financially driven and not even really AI driven, uh quote, direct quote from him. Rather, that it's culture and kind of meaning that the company was um facing too much bureaucracy. So that's kind of strange to me. I mean, I'm assuming um Tim and I were having a kind of a bit of a riff about this before we record, but you know, that it sounds like they're doing the whole we're flattening the org uh type thing where they cut out some of the middle management, and then people that used to have maybe five reports now have probably 50 reports or something like that. And um, I'm sure it does great for um the managers of the world within AWS um or within Amazon as a whole, I guess I should say. Um, so uh at this point, this is kind of just par for the course. We know the layoffs are gonna come right around these points of the year, almost every year. Um, so not to kind of you know make light of it or anything, but um, you know, we have a lot of friends over there at at Amazon, so hopefully we uh we don't see anyone close to us affected. Um but yeah, what do you think, Tim? What's new about these ones?

Tim: 3:07

I mean, there's yeah, honestly, it doesn't feel like there's a lot new about it. Like you it just uh what did they say? Uh the full 30,000 jobs would represent a small portion of Amazon's 1.58 million employees. Now, this is Amazon plus AWS. This is not obviously just AWS, clearly. Um and then yeah, so uh, you know, comparatively in 2022, this has I don't know why this article specifically says 2022, because I know it's been I'm pretty sure it's been more since then, but it mentions that you know they did 27,000 in 2022. So between like last year and now 30,000, again, they're flattening the org, which is a the fun way of the fun corporate speak way of saying that you know we're we're laying off middle management. So yeah, nothing new here. Uh still more of the same. I I I'm curious to see what the actual impact will be. Uh, not to the shareholders, I'm sure they'll love this all day, but uh to productivity and and just you know team morale and team culture. Because once you're a you know, if you're a manager with 50 reports, you're you probably hate your life. And if your manager has 50 reports, you probably also are not getting you know more than 15 minutes of the of the manager's time. So good luck on any kind of career advancement or anything like that if you're trying to sort that out. So but I do get it, right? I understand that we're all on the the save money to you know, and I know we I know Jassy said that this isn't about AI, but really everything's about AI, like when it comes to money. Yeah, like it all just because it doesn't just because there's not a s a line on the spreadsheet that's next to the AI line and you know where they move the numbers around, uh it really is, especially with this, with this ridiculous race that everybody's on. It's all about the money and the debt that you can take out to get the AI. So yeah, I don't know. That's it. So speaking of, uh well not speaking of, but but moving on, I should say. Um the next article is up uh talking about now. This is actually kind of good news, and I'm I I have a friend in uh, you know, we have a friend in in uh the Art of Network Engineering Discord who works for Microsoft who who kind of shared this with us. I don't remember if this is the if it was this exact article, but um so Microsoft is uh kind of un unveiling a new uh AI they're saying an inference chip, so Maya 200. So what's interesting about this new chip is like think about you know all of the silicon basically being used to do AI inference and then this chip, I'm trying to find the specific numbers that they were throwing around. A three times better four-bit floating point performance than the third generation Amazon training chip, which is by the way, not a an old chip by any means. They're you know, these so these are these are these these new silicon chips are just rolling out, um, as you know, as well as they can as fast as they can make them and design them. Uh Maya provides a 30% better performance per dollar than the latest generation hardware in our fleet today, is what Microsoft is saying. So what we're talking about now is like, you know, we we were talking about this before we hit record, and I've actually said this a few times, you know, either on Twitter or maybe at some point recently, that you know, as the AI arms race kind of moves into its more mature phases, it debt is harder and harder to secure. So I think the big gains are we're finally getting to the point we knew we would eventually get here, where you know, we've kind of drawn enough water from the well that we have to start making the water, you know, pooling process more efficient because you can't just keep taking debt over and over and over. So now you have to actually go back to the well and build a better pump, as it were, right? So we're talking, we're finally, I think, getting to the point where we're seeing some efficient, some some thinking around efficiency. So, you know, this that that this is great news, right? So it says uh in practical terms, Maya 200 can effortlessly run today's largest models with plenty of headroom. So not only is it more efficient, and this is what you see with micronization of technology over time generally, anyway, right? Is the first tech is is is clunky, it's it's it's not efficient, and then over time, you know, as you've as people figure it out, it becomes more efficient. But yeah, it's it's a really interesting um article from Micro, you know, from Microsoft about this this new chip. It mentions that previous versions were plagued by the design and development challenges that were mostly self-inflicted. So I another good part again with the micronization of technology is over time it just gets better, right? Like as you do better, it gets better. Uh Moore's Law Notwithstanding, right? So it's a pretty cool, pretty cool article. Um yeah, what's your what's your take on this one?

Chris: 8:12

I mean I've if there's one thing that I'm not like uh motivated by or generally interested in, it's chips. Like, I'll be honest, I could give a fuck less about chips. Like I like it's cool, yeah. I'm I'm glad, but that's the thing, is I understand that this should be kind of the status quo. This is they this should be where things head, right? That things should get more efficient and you can do more with less, and and that's how you know kind of that we're able to let this thing stand on its own. Um, otherwise, like you said, we're just gonna keep going into debt. Um, so yeah, it's cool. I I'm glad that this is this is where things are headed. Um, obviously, you know, the the future of AI, I think, is is a very different thing, but at least we know we'll potentially have the hardware to do it. Uh another minor grievance that I have here is I hate how everything's measured in teraflops. Like, flop to me sounds like a bad thing. Like flop is generally can like when something is a flop, it's it's it's a bad connotation, but this is always spoken about in some like, oh, it's it's doing this many teraflops as an impressive feat. And I'm just like, dude, that sounds like shit. That sounds awful.

Tim: 9:23

Like, you're right, like that's a good point. Like I I have to look it up actually, because I don't actually even remember like teraflops versus terabits, like I don't even remember what that means at the end of the day. Like so and well, and they mentioned specifically about floating point math. So maybe it's like a maybe it's math calculations or something. I have to look it up, be honest with you. Um but yeah, I'm right where with you. Yeah, we'll we'll go figure it out. Actually, I'll I'll just go ask Claude and and just complete the circle of what you're gonna do.

Chris: 9:58

We'll put the quote and the show notes. Just kidding that we won't.

Tim: 10:04

All right. And actually, as a follow-up, well, not so much a follow-up corollary. I don't know how you would how you'd call this one. The next article basically is Microsoft uh CEO, Satya Nadella, says it's time to start talking about AI slop and start talking about a theory of the mind that accounts for human beings equipped with these new cognitive amplifier tools, which is a very like feathery, like I don't know what you'd call it, like it's a very, very uh heady way of saying basically like we we should just stop using slop, guys. Just just stop calling it slop and let's start talking about how awesome it's gonna be. So and and this was uh January 2nd. This is when it's been this this particular article came out. So this was before the the recent thing in the WEF or whatever in Davos. I think so it's yeah. But uh the article goes on to oh go ahead.

Chris: 10:59

You were gonna say you have something I think this is actually based on um like a blog post that he does on his like scratch pad thing.

Tim: 11:07

Oh, yeah, yeah, that's right.

Chris: 11:11

That's just before the start of the year. Um, but he kind of goes on to that, you know, that we need to evolve this bicycles of the mind uh uh concept that I don't know if he coined or someone else coined, but um such that uh we think of AI quote as a I think of AI as scaffolding for human potential versus a substitute. And I think that's very easy to say from I don't know, man, like like that makes sense in white collar world, I think, especially specifically in the tech sector as well. Like it's I think it's easy for us to kind of get lost in this like oh like AI is not gonna take your job, you need to use it to empower your job and things like that. There's a lot of people that cannot do that, right? Um or their work, yeah. Exactly. Like if you if you're in you know uh the broader context doing some type of blue-collar job, some like actual hands-on activity, the only option is for it eventually to probably highly augment your job to the point where you're not needed, right? Um so it's I don't know. Uh like I'm uh my mind is growing you know dimmer and dimmer on this topic every month, every day. So um maybe I'm just being next fancy here, but I don't know.

Tim: 12:32

Uh what I what I love about this article is it points out that uh Merriam Webster, the dictionary, their you their word of the year for 2025 was slop, which is beautiful, by the way. I think that's great. Uh I know there are people out there who really wanted to like forget the word slop or say that the word slop is cringe and nobody should use it anymore. Um, but here's the thing, we don't get to decide when a when a word, the etymology of a word or when a word stops being relevant or whatever. But I mean it's it's so perfect, right? Because the year of slop, I mean, yeah, there was this was the year of 2025 was kind of the year of let's just see what happens if we let you know AI Jesus take the wheel, just hand over everything, right? So um hard to argue. Uh it was quote unquote a year of AI humiliation, uh like is what um what was it, PC Gamers Lincoln Carpenter said that the 2025 was a year of AI humiliation. So like there was just so many things that AI just failed at, or and also that it got better at, but really just the idea of just like let's let's just hand over everything to AI and let it figure it out. Like that that that idea failed spectacularly, I think. So yeah, right. Um so Satcha really wants people to stop using the word slop and to start thinking about again, the the the Steve Jobs bicycle of the mind theory, uh, saying, you know, making it make everybody's job easier. And I do think that's great because there are so many zealots out there that are still, even today, saying crap like, you know, we won't even need network engineers in six months, or how long, how many, how many six-month intervals have all these CEOs been saying coding, we won't need coders? Like, you know, code's gonna die and AI is gonna do it all for you six months from now, and it's been six months for like last two years, so and still going.

Chris: 14:26

Imagine what six months from now is gonna be like, it's gonna be.

Tim: 14:29

Oh my god, I can't wait. I can't wait to not have to just not need a job anymore six months from now. So, I mean, the the article just points out that again, slop was the word, and just the idea is let's transform it. Satcha wants us to transform it from hand over the keys to enable and empower. And I can generally get on board with that. I think it is easier said than done, for certainly for some for many uh professions, right? So there will be professions that I think will find this easier, this idea of bicycle the mind. Like, for example, I would think cybersecurity would be really, you know, anything where you're gonna be chomping up ch you know tons of data and pulling insights out of it if you're good at doing that, and then you as the human take that insight and do something of value with it. I think that's gonna be something where we can get it behind this whole bicycle the mind thing. But like 100%. So yeah, so uh for what it's worth, Microsoft is you know, they're leaning in on AI as we do, but uh at least they're bringing us some new chips with some efficiency. So I guess we'll see where it goes from there.

Chris: 15:44

Yeah, maybe maybe in six months RAM will be affordable again. Wouldn't that be crazy?

Tim: 15:48

Oh no no, RAM is like projected out for the next five years to be just shit. Like I'm not even joking when I say that.

Chris: 15:56

That's so funny, dude. All right. All right. Uh next up we have uh something thankfully not AI related. Um we're trying to be better about having more and more and more non-AI related things in the world. As hard as it is these days, it's becoming difficult. Um, but yeah, hopefully we'll have something in the future that will that will assist us with that. Um just a kind of a hint, we might have something better for you in the next few months, so just stay tuned.

Tim: 16:29

It's an it's an AI agent. I'll split up.

Chris: 16:31

It's an AI agent, the Campus Clouds agent. Um goes out and looks for news that is specifically not AI related. That was pretty good, Tim. Nice. Um, all right, uh article here. The Nutanix CFO, um uh I can't even remember how to say his name. Uh Rukmini Savaraman um uh uh basically made some commentary on the fact that um Nutanix stocks have been uh kind of nicked, kind of falling quick uh a little bit because of the um the lack of migration from VMware customers that they that they had expected up to this point. So basically the sentiment of the article sounds like people are not moving away from VMware as fast as as everyone thought they would. Uh obviously, if you're Nutanix, that is a very you know big business opportunity for you. I'd imagine I haven't even tried it, but I guarantee if I go to the the Nutanix website right now, there's gonna be some big link to click that says how to migrate off of VMware or something like that. Um, which it might not be there, but that's what I would expect. But um hints at the, you know, they they've instituted a stock buyback program to kind of help uh pump the share price a bit. Um, but it also makes note that they're sitting on quite a bit of cash for that reason and you know potentially hints at them making some type of acquisition in the near future. Um it there's a note in here that they did acquire a company back in early 2024 called D2IQ, which kind of helped them with their Kubernetes play um in the um in the long term, and which was good, but I'll be honest, like thinking about that, I'm like, what I don't know exactly what Nutanix could acquire that would that would set them apart and one, either get people to move off VMware faster or to kind of put them into a new um kind of a new category altogether.

Tim: 18:34

Yeah.

Chris: 18:34

Um I don't I don't know. I mean, this is why I'm not a CEO, I'm not a CFO, like I'm not smart enough to know this kind of shit, but like I just don't know what would really push that needle for them in that in that particular category.

Tim: 18:48

Yeah, that's a really good question, actually. Like what if you were Nutanix, right, and you were trying to, I mean, let's assume, because we don't actually know, right? Let's assume they were trying to entice VMware customers, right? Like let's assume that was the goal. What would you do? What would you buy that, like would you buy off the market essentially as a as an acquisition? Because I was immediately what I think of is like, how about building a VMware style or a vSphere type of, you know, something something command and control a little bit more friendly to VMware customers? Like, that's what I would think. But yeah, you wouldn't buy that off the that's not something you can get off the market as an acquisition. So yeah, no. I'm with you. I'm kind of wondering what they would acquire.

Chris: 19:32

I mean, it's also like I don't know. I I feel like I don't hear much about it. I I genuinely don't know what the sentiment are is among VMware customers today. Like, are they just waiting it out for for the you know the contracts they extended um and potentially looking at moving off later? Or like is Broadcom putting the effort into developing VMware at this point where people are kind of seeing new features, new things being added on a regular basis, has or has that kind of halted? Like I don't know, I don't know the state of it. Um maybe we should um Yeah.

Tim: 20:15

Is there anybody out there that has that knows yeah no I agree with you. If if we know somebody who's into the who's in who knows more about that, I'd love to dig in on that actually. I mean the article mentions that like the people that are migrating to Nutanix obviously are not doing it quickly. Because you can't nobody could possibly take their you know entire server fleet and just move it over to Nutanix. So it does mention we're getting you know that people are migrating but it's slow, which is what I would expect, right? Yep. Anyway. Uh okay I yeah like I I can cover the last one. I think I don't think we decided but uh I'll talk about the last one here. Oh did we say it was me good then because I'm gonna cover it. Good because now I'm gonna cover it. Okay cool. Uh you can see I I do a lot of prep a lot of preparation for this. So something came out recently not not the product but uh but an article about it. I guess it got kind of gotten to the news a little bit more and this is new it's this new thing called Claudebot and a lot of you know it and I I apologize for going back to the kind of veering back towards AI for a little bit but this one's kind of funny. So Claudebot is a uh personal assistant like a like a Siri type of thing but but it's an AI think of like an AI agent as personal assistant and and this person kind of I don't know if he vibe coded it or just like just built it but um the the the maintainer of the program it's an over it's an open source uh bot essentially uh you can get it on GitHub um recently kind of came under fire from security researchers because they kind of looked at it and remember this is a this is a personal assistant bot so the the goal of this bot is to essentially get access to your data your calendar your email your chats all of this and then run point for you like create calendar invites uh what was it uh restaurant uh reservations and stuff like that so yeah so that means that you have to basically hand over the keys to your data and to your apps to this bot and so a lot of security researchers were you know putting in um you know GitHub issues and like just just basically putting stuff on Twitter about uh how insecure this thing is like apparently this thing is so insecure that we're talking like you know keys in plain text type of insecurity like you know that there's just no security in this thing right now. And the project maintainer was like well this thing hasn't been around very long and it's an open source product. It's basically a hobby of of mine and whatnot. And I I can actually honestly if and I could be I can actually kind of agree with that. Like if you're if if I just made something and I was like hey you guys go kick the tires on this or try it and then people went took that thing and just started like just doing you know things you would probably never want to do with with a hobby project I would probably be like you know righteously concerned as well. But anyway so there's that. So that's this this bot. So on the top of that this Claudebot is recently had to change its name uh very recently like with like today I think actually they changed their name from ClaudeBot to Moltbot. And the reason is that ClaudeBot was originally built uh quote unquote the the project maintainer oh I forgot the name Peter oh I forgot the guy's name um I just lost it it's not in the article um it's Peter something I'll find it we'll put it in the show notes but um he basically said like he created Claudebot like the name for Claudebot came from Claude like the the monster or whatever that Peter stuff that that's a Seinberger um but like that's where he got the name for Claudebot. So so very clearly like Claude Bot was modeled around Claude. So apparently Anthropic who owns Claude did not take particularly kindly to that and reached out and basically said we own Claude which I feel bad for anyone who happens to be named Claude I'm I'm sorry to have to inform you that for you to find out this way but apparently Anthropic owns everything about Claude. So he had to go change and also there's and randomly there's a meme coin called Claude also C-L A W D. So apparently a bunch of you know bit not Bitcoin uh crypto bros or something I'm not sure started giving him shit as well because of the meme coin. So this poor guy got it from both barrels on both different ways and then had to it it does say Peter yeah sorry and then um had to change the name. So short answer is ClaudeBot is now called Moltbot because it was based on a lobster and they decided that a lobster molts when it grows and you know good for him clever. I do think I do think it's just goofy. Could you imagine just being an open source maintainer and like just like your thing take off and everybody just run with it and this and then this this this kind of uh what's the word this comedy of errors if you will you know just like comes into your project and it just kind of became viral um so it's it's kind of interesting. However having said all of that no security I wouldn't definitely not recommend that anyone run out and start running Claudebot on a real like set up a you know set up a a VM and like you know if you want to test it out set up a VM yeah and some dummy accounts or something to play with this thing.

Chris: 25:47

But apparently sandbox if you will a sandbox yeah you could you could do a sandbox with it because it is um there is absolutely no security to it so I and oh the article said something else that was really funny that uh or it was a linked article that you you sent to me Chris that a guy turned on ClaudeBot and it went and created what did it mess with this calendar and then created uh restaurant uh reservations for him and it costs like eighty seven dollars worth of tokens yeah he said uh it was a tweet he said uh his name is Vesuman on Twit on X uh sorry he said holy cow dude I gave Claude bot a task and after four messages back and forth it made a reservation at a restaurant for me this saved me so much time wow after another six texts it sent a calendar invite for me and it only cost me$87 in Opus 4.5 tokens yeah very funny so yeah you can so it's so it makes it honestly uh so you could probably hire a human to to be your your assistant and pay them less than what you will pay for their tokens I don't know the sh the jokes write themselves guys I don't know what to say about this one.

Tim: 27:01

But yeah that's yeah I did anything else you'd like to add to this it's kind of ridiculous. Oh well one more thing I wanted to point out actually before you before you before you take over so U Underground who is we I I personally love quite a lot they actually just dropped a uh a new song um making essentially giving shit to this to this AI uh bot thing because it's so the security is just so bad they they they really turned it around quickly and I remember this morning I saw on Twitter when they when they announced they were changing it from ClaudeBot to Moltbot they responded and were like well fuck because they'd already like recorded the song oh man yeah it's good stuff oh I see yeah they posted this tweet only 16 hours ago I've not heard this one all right well we'll put that in the show notes um and you can take a look at it um and yeah I don't really have much to add there it's uh obviously very comical um I will not be using ClaudeBot probably in the um very near future unless it is near to distant future yeah um but yeah so that's great and with that we'll go ahead and wrap it up for this month.

Chris: 28:17

So if you enjoyed this um well you made it to the end so obviously you enjoyed at least a little bit of it and uh we appreciate you doing that and we hopefully um uh can get you to share this with a friend um post it just do something to get it out in the community because it's the only way we can get this to grow and we uh greatly appreciate that um and with that uh we'll take it away and we will see you next month bye bye yep see ya