AI News of the Month - July 2026

0:05Brandon Whichard: Good morning, good evening, wherever you are. Welcome back to the Enterprise AI Show. I'm your host, Brian Gracely. And today we're going to do another AI and cloud news of the month. Brandon Whichard is going to join us here in a little bit. We're going to dig into kind of a recap of the World Cup. We're going to dig into cloud earnings from the hyperscalers. We're going to dig into AI agents and harnesses that are breaking out and doing all sorts of bad, nefarious Leroy Jenkins things. We're going to dive into chips. We're going to dive into security. We're going to even dive into a little bit of old school cloud stuff and what's going on with VMware.

0:38Brandon Whichard: So lots and lots of stuff this week. Actually, for this Cloud News of the Month, stick around. We'll get to it right after the break.

0:47Brandon Whichard: Today's show is sponsored by Nasuni. There's a growing gap in AI right now between what's possible in theory and what successfully works at scale inside an enterprise. The difference comes down to unstructured file data. Many AI initiatives struggle because the file data they depend on is scattered, unstructured, and disconnected from where and how work actually happens. Nasuni changes that. It brings your unstructured file data into a single, secure foundation so AI, both generative and agentic, can access it with the context, governance, and performance it needs in production. Bring AI to where your unstructured data lives.

1:22Brandon Whichard: See what it takes to activate your data for AI and request a demo at nasuni.com slash AI. And we're back. And folks, it's been a month. So that means that it is time for us to do everybody's favorite AI and cloud news of the month. And we are excited to have world famous, world class, Brandon Wichert back with us. Brandon, how are you, man? How are things in Texas? How have things been over the last month?

1:45Brian Gracely: Things have been fantastic. You know, the only answer when you ask somebody that question in the summer in Texas is hot. It's extremely hot. We're enjoying it. We're burning up. But we're having tons of fun, if you will, following all the AI news, which we have plenty to talk about this month.

1:59Brandon Whichard: Yeah, we do. We do. Before we get into the news, because there's obviously, like you said, lots to talk about, we can do one of two things. Did you pay attention? Were you watching the World Cup? I think you were. Absolutely. Watched it all. Watched as many games or matches, as they say, as possible. Did you feel like you learned any new corruption things that you could take forward until the next four years? I feel like FIFA was like, hey, we were really corrupt. The people who pay attention to soccer know we're corrupt. The US now knows we're corrupt. And now they're trying this new thing that seems to upset everybody, where they're like, we were really corrupt as a quote unquote non-profit, even though everybody who's on the board is making tons of money.

2:39Brandon Whichard: And now they're like, now we want to be a for-profit. They're taking the open AI model. We want to be a for-profit and we want to start taking money from private equity. So I feel like they're just like, you know what, being corrupt in the dark is not good enough. Let's be corrupt out in public because that seems to be perfectly fine these days.

2:57Brian Gracely: Yeah, and I think it was, first of all, I loved watching all the games, all the matches. And I think one of the things I learned about was like, if you were trying to be like evil, kind of Dr. Evil type person, I can't remember, I'm not gonna say his name right. It's like Giovanni, whatever, the head of FIFA. I thought it was amazing how he always wore the black suit black tie and they they framed him on every uh and somehow he went to like every match like he flew between them and they always framed him as like a dr evil looking character like about midway through the first period or first half yeah and then midway through the second half i was like okay obviously he's told them to do that right i mean there's only that's the only reason you get that face time i was like okay you seem like you're dr evil but the tournament was really fun i think it was generally well received in the united states i actually happened to be in mexico a little bit during it, they had a good time.

3:44Brian Gracely: There was a lot, they had some games there and that was really good. So you're like, out of like a pretty excellent tournament, right? What do you do? You immediately go for more, right? So I think that that's the private equity thing. But I think to your point, I mean, maybe this is like what we need to learn about corruption is like, he seems to have gone so far, everybody is mad, right? Every, if you will, regional soccer Association has now voted against this, like Europe, the United States, or the South America, the Americas. So I'm like, either he has like an extra hand to play here, right?

4:20Brian Gracely: He's like, it doesn't matter what you people think. I still have control of this because he has some type of weird power. Or I think he could potentially have gone too far. I think this may be the moment where they're like, you just went too far and you're going to be ousted. And of course he came in under the auspices of like cleaning up because the last group was even more corrupt. So, so it kind of leads me to believe that like, I think he'll be out and probably there'll be a new corrupt leader very shortly. That's my, that's my prediction of what's going to happen. Yeah.

4:51Brandon Whichard: Yeah. No, I think that was, I think that was good. I think, yeah. I mean, I think like you said, once you get to the, to the, like the pure Dr. Evil point, you just have to double down on that. Cause you're like, I mean, yeah, you, that's your brand. You've established it. You You might as well see, you know, just how high you can get in that, in that thing. Cause yeah, if they're going to, I mean, if they're going to, if they're going to, you know, nab you and somehow put you in jail, but I don't think they are like, I mean, there's no like FIFA jail. There's no, I mean, FIFA, I feel like is, don't they have like their own, you know, they're based out of Switzerland.

5:20Brandon Whichard: So they're basically like, they don't, they don't live by any laws.

5:23Brian Gracely: Well, I also think it's funny too. It's like, you had this huge success. It's like, just take a month off, like just go on vacation for a month. Don't do anything, lay it low. Right. But it's like, no, no, no. as soon as this is done, now I'm going to sell this, I'm going to sell this prize piece of a nonprofit to a whole, to like, we won't even get into who it is, but it's, let's just call it polarizing political figures involved. And it's just like, no, people are pissed. They're not going to, they don't want to do it. So we'll see. We'll see if it's corruption, Doug, we'll see if he can continue on or not.

5:52Brian Gracely: So, and in four years, we'll all be watching, or actually even next year, we'll all be watching the world cup with women's world cup. That's right.

5:58Brandon Whichard: That's right. Do you want to, it's been three months now since we did our AI draft. Do you want to do any sort of review of that? Do you want to do any gloating over your picks? Absolutely. I feel like I'm winning,

6:08Brian Gracely: so I feel like we should go over this for the next 40 minutes. So we'll do a quick thing. So your number one pick was Google, mine was Anthropic. I think it's clear. Yeah, let's just

6:17Brandon Whichard: go through. We had eight clicks.

6:18Brian Gracely: Clearly, I'm winning on that one. Would anyone disagree with that? Like Google, I mean, they're doing some good stuff, but they don't have a leading frontier model. They're basically, if you will, had to lease their computing capacity to the Anthropics. So I feel like hands down, I won that one.

6:34Brandon Whichard: Yeah. Even though Google did make an astronomical amount of money. Sure. Sure. They made a lot of money, but that's sort of like,

6:39Brian Gracely: but they got that money from Anthropics. So I'm taking the win on that one. Number two was, you said major AI centric IPO in 2026. I don't know. Do you want to claim SpaceX as that? Or is that?

6:50Brandon Whichard: I mean, I could, but I feel like SpaceX sort of performed like some of the poor rockets. It took off a little bit. They tried to manipulate the hell out of the market. It's now been down. Yeah, I'm not, I don't feel great about that one yet. You're going to wait on that one.

7:04Brian Gracely: We'll check in on that one. You still have Anthropic and OpenAI coming later this year, I think. So, okay, we'll, to be decided on that one. I took NVIDIA. I think that's a solid, still a solid pick. It's fine. I mean, it's fine. It's doing well there. So that one, I'm going to give myself maybe moderate. We're just going to actually, we're not going to grade that one. We'll call that two up.

7:21Brandon Whichard: Yeah, that's fine.

7:22Brian Gracely: All right. What do you think about this one? You picked, third pick was Amazon. I picked Broadcom. How do you score that one?

7:29Brandon Whichard: Well, I mean, I feel like for all of the, you know, kind of negative Amazon AI news, they did still announce that they're making a ton of money on their cloud. They're back to making a lot of money in their cloud, which again, is mostly serving other people's models. You picked Broadcom. I feel like the only news out of Broadcom lately has been that they're losing VMware customers because they're getting sued. So, I mean, I think even if it's just by a nose, I think I win that one. Okay, that's true. That's true. I forgot about it.

7:56Brian Gracely: I almost forget that Broadcom owns VMware when we talk about it from AI. It's like, oh, yeah, that one as well. There's that part, too. Oh, yeah, we'll stop here at four. We'll just say you made the prediction that a company has more agents than employees. Possibly. I don't feel like that's been a dominant theme. Yeah, hard to know on that one. Maybe, maybe not. I took OpenAI as a frontier model. So again, I feel like, again, they're doing well. They're still out there.

8:22Brandon Whichard: I don't know about that one, though. I mean, has there been any good news about OpenAI lately?

8:25Brian Gracely: Well, maybe that's a good one. Maybe that's our first topic this week. So OpenAI essentially, I don't know how to say it. All their executives mostly left for different reasons. For like, some of it was health reasons. Some of it was more time with their family. Some people were fired. There seemed to be a little bit of everything. Everyone reports to Greg Brockman now, I think, other than Sam Altman. So I think while it was very messy, I do think they mostly have the house in order for at least going on IPO. I think because I kind of read this sort of like we're not going to do all these crazy projects that we had before.

9:03Brian Gracely: We're just building an enterprise coding agent and we have to figure out the consumer business. So I don't know if I want to say this was a good strategy. I guess it was sort of like proper course correction. Like, I don't know. How do you think of it?

9:16Brandon Whichard: Yeah, it's still, it's still, it still feels like it's a weird business because on one hand you've got, I mean, what did I think I read? I mean, they've crossed over a billion people now using chat GPT on a weekly or a monthly basis. So you have this massive consumer business, which is incredibly expensive. You're really not monetizing it terribly well. And then your, your other strategy is let's try and be anthropic, right? Let's try and basically just copy Anthropik's model in the enterprise, which, again, being the second is not necessarily a bad thing to be. It's just weird that they have such a big sort of cost, this anchor, which sounds crazy.

9:57Brandon Whichard: They have a billion users of something that they're losing money on, and they don't seem to want to monetize that in any way. I mean, I haven't read an article about their advertising business in forever. I do get the random, you know, ads in ChatGPT when I use it. But yeah, it's a weird business. It's definitely going through a, you know, we've taken, I mean, they have to be the weirdest, you know, we've taken nearly$200 billion and we're not really sure we have minimum viable product other than the product that we don't really want to monetize.

10:28Brian Gracely: So I just want to read you as you were going over the advertising business in the show notes here, there's an article, everyone can read the whole thing if they want, but I'm you a little bit more data on your advertising says, it says, quote, open and open AI is on pace to undershoot its own five year ad revenue projections by a whopping 90%. So that's, that's bad, Brian. That's really bad. That's real bad. So, I mean, I kind of feel like the way to say this is there the consumer business that, and I think this goes back to the leadership changes. Many of the leaders that departed for a variety of reasons had a more consumer based ad background.

11:04Brian Gracely: I don't think that's breaking any news. So the fact that they're missing it by 90%, I think it's clear. It's not even so much I would say those leaders were bad. It was like, this market doesn't exist in the way that they thought of it was. I mean, there's no getting around this. You don't miss by 90 % and be like, you know what we need is another sales team. It's like, no, you don't have anything. I mean, that's just, that gulf is so massive. So to your point, and it is the ultimate jujitsu move that Anthropic has pulled and it's mostly by luck. It's like, yeah, it turns out having the giant consumer business is what you were just saying before.

11:38Brian Gracely: It's a huge drag because you have to pay for it and it's not really profitable. Here, Anthropic and Mistral and all these other are mostly focused on the enterprise. And it seems like that's where the money is. So as someone that took Anthropic in the draft, I hope they turn it around here. And I need Greg Brockman to be like, yeah, we got to stop working on this ads. We got to work on stuff that's profitable, which I think is what they're trying to do. I mean, remains to be seen if they'd be successful. But I think that's what's happening.

12:03Brandon Whichard: Right, right. Yeah. I mean, it wouldn't surprise me if a year from now, you know, we were reading the Gartner Magic Quadrant of, you know, Codex versus Claude, you know, kind of like whatever that bucket falls into of like AI for productivity or something like that. And those two are, you know, up and to the right. That wouldn't surprise me at all. It's just, yeah, how does OpenAI not keep doing other businesses? Like, for example, they announced, you know, ChatGPT for health. And they're like, just upload all your health data to us, which, sure. Lots of people will do that. Well, and I'm sure there was a certain amount of them.

12:36Brandon Whichard: Like, I'm sure the person who is like ChatGPT for, you know, counseling, essentially, or, you know, like, you know, ChatGPT for therapists, they're like, I don't think you guys realize, like, ChatGPT is used for therapy, cheating on homework, you know, asking if I can avoid using a financial planner, you know, stuff like that, be a trip planner. and they're like, why wouldn't we do health? We already, people are already telling us all kinds of stuff. And somebody was probably like, okay, I'll let you try that for like a month. And then we're gonna look at the bill and see how much we lose. And then we'll decide if we're gonna keep it or not.

13:11Brandon Whichard: Yeah, that's probably true. Like, I feel like there's probably still a certain amount of like the old SAM Y Combinator that's still floating around inside of OpenAI because it's like, guys, we have a billion people of worth of data coming in here. Like, figure out which applications should come out of this.

13:28Brian Gracely: Well, but, you know, this is something you've mentioned on several shows, right? I think you often say something like, you know, the quote unquote first agentic application hasn't been, if you will, discovered, right? Something that could only advance. So kind of think of like Facebook. No one was predicting Facebook at the beginning of the internet. That's sort of like evolved or Uber or something along those lines. So you could, to make a very diplomatic way to say is that, you know, Sam Altman is, he's with you, Brian. He believes, he's like, yeah, he's like, we got to keep trying to figure out what the native agentic application is going to be.

14:03Brian Gracely: Because that's like, you know, worth untold trillions if we can hit it, right? Meanwhile, you know, Anthropic and all these other companies are like, you know what we're going to do? We're just going to make the enterprise a little bit more efficient and people are going to keep buying those tokens. So you could probably say that. And of course, you know, it's been announced and widely, I don't know, I would say pan, but I think people have rightfully have questions. Is there, it seems like their first hardware product, of course, they're coming out with Johnny Hive is the, the speaker. They're going to have the home speaker, you know, the Alexa or whatever you want to call it.

14:31Brian Gracely: And I think it seems underwhelming. I don't feel like anyone is like, what I need is another smart speaker for my house, but maybe we're wrong, you know? And I, and so of course, Sam Altman's like FOMO, right? It's like, he's like, it's coming. It's going to be, It's going to be super great. And of course, what's even more fun about that is Apple's now suing like 400 of their employees for all like leaving. So which is another, if you're Sam Allman, I guess you just get up in the morning, like which problem am I going to work on? We either have to invent what's going to be the new application for all of AI, or we have to hire 500 lawyers to represent all of our employees who are being sued by Apple for trade secrets.

15:05Brandon Whichard: Well, it's one of those things where like, let's say you're sitting there and you're like, okay, we want to be in the consumer space. If we're going to be in the consumer space, we've got to have some sort of hardware or probably not going to build a phone or maybe we're going to build a phone. I don't know. And somebody goes like, hey, why don't we start with a speaker? And you're like, okay, well, that's not a radical idea. It's a pretty easy hardware form factor and so forth. And then you kind of wonder if somebody was in the room going like, hey, do you know how we create stickiness for people using our stuff?

15:32Brandon Whichard: It's we keep track of everything they're doing. Well, in the home, you don't ask it that many things. Or at least like it's not your natural behavior to be like, hey, OpenAI, you know, or hey, ChatGPT, hey, ChatGPT, you know. But like when you're doing work and you're just jamming data down its throat and it's building up memory and then you're like, hey, turn that into a skill. Okay, cool. Now I've got more memory. Like you would think that unless the thing is listening to you all the time, which they probably don't want to do, maybe they're already doing, but you don't want it to think that it's doing that.

16:02Brandon Whichard: you know you sort of are like you're not going to get as much data to create a memory to create a personalization as you think you are and you would have thought after you know 10 years of alexa and 15 years of surrey people would be like oh i already know how much people use that stuff like there's tons of information about that and none of those none of those have been like people be like oh man that thing knows me it's not at all in fact quite the opposite it's like i've been

16:24Brian Gracely: talking this thing for 10 years and it still doesn't understand like the simplest words so So I don't know. I guess they, unfortunately or fortunately for them, they just don't have, this is probably something that needs to be quietly built in a lab and tried out in a quiet way to see if it works. But unfortunately, everyone's going to be watching. And of course, Sam Altman and Johnny, I brought this on themselves. Of course, they had the famous coffee shop video in San Francisco that goes, I mean, honestly, you're like, pair to your reality. You're like, I don't know. Is that an episode of Silicon Valley or did that really happen?

16:50Brian Gracely: So, I mean, they kind of brought it themselves. It'll be fun to watch though. It'll be something for us fun to talk about later this year. Well, one of us will have to buy one and we'll have to give it a review. Does it work? That's right.

17:01Brandon Whichard: Or they can just send us one. They should probably send you to us one.

17:04Brian Gracely: We're always down for that.

17:06Brandon Whichard: What's your take on this thing that's been going on the last, I don't know, feels like few weeks, but maybe it's been going on since Mythos came out. The idea of like, oh, our new model is so powerful. It's so scary. So first it was Mythos and like, hey, it's so powerful and scary for security. Only a few people can have it. now it seems to be the new thing is, oh my gosh, our agents or our model, we were doing testing and it broke out of the sandbox and started hacking into stuff. Are you buying into any of this that it really is scary stuff and it just sort of has Leroy Jenkins kind of DNA? Or do you look at these as entirely marketing events just to sort of scare enough people to be like, okay, what's the anecdote?

17:48Brandon Whichard: What's the thing that I can use to protect myself against you big bad guys?

17:53Brian Gracely: Yeah, I guess I look at it kind of in the middle. I don't think they're purely marketing events. I don't think there's anyone like thinking like, oh, I really want the open AI model to break out and like try to get answers from hugging faces. A good thing. I think it's just more like, I don't know what to call it, like research or just this is how it goes with technology. It's like you put it out there and you see what happens. And of course, like when something happens, people, everyone can just draw a million different opinions from the same event. They can either say, this is the end of the world.

18:21Brian Gracely: This is the beginning of Skynet. It's all going to take over. Or people will say, no, this is just an example of like how insecure things always were. Right. And this is how it always was. Like, you know, I, a friend of mine runs a group of CISOs and he was telling me about, and I was like, what are they, what are they using? What do they care about AI? He says, all they care about is if you will, making sure there are guardrails around there. So it doesn't get to data that is not, that it shouldn't get to. And then he said, he's like, yeah, like it turns out a lot of these AIs are like finding all this corporate information.

18:50Brian Gracely: like everybody's salaries like available on some SharePoint or something. And I was telling him, I was like, well, that's not new. Like it was always there. Like the, you know, the only thing that's new here was just the fact that like there was some, something smart to figure out what it was, but like maybe one person knew where that was before. And it's so, so it's like, to me, it's like, so there's like, Oh, it was changing everything. We have to change everything. It's like, no, not really. It's like, you just had a lot of insecure data before that you didn't know about. Now you are starting to know about it.

19:17Brian Gracely: So why don't you go put out? And again, like one answer could be a better AI, you know, certainly plenty of security vendors would say like, well, we can use AI to combat it. So I just kind of look at it like, this is just where it is. It's like, and of course, like, and I didn't think anyone that predicts like, you know, when they know if it's going to get so smart that it's like smarter than all of us and doing something bad. It's like, it's almost like an impossible question to answer. It's like, all I know is like, it's like, sure. Maybe if we were like, if we were truly a command and control society, we could probably be like, Hey, let's slow down and think about it more.

19:51Brian Gracely: But it's just like, when has humanity ever done that? It's like, there's no way it's anything's going to stop. So it's like, we just, every week we just have to see what happens and then course correct as appropriate. What do you think?

20:02Brandon Whichard: Right. Yeah. It's, it's interesting to me because, you know, I mean, there, there have been things in the, I mean, this feels like it's one of these things where it's like, we're going going to get very close to a line of, is this thing, does it sort of fit in the domain of like, hey, we should just let it see what happens in the marketplace. And there might be some things that fall down, but there's going to be a lot of good out of it. Or does it cross enough over the line to be like, oh no, this is really kind of the equivalent of like plutonium. And we really don't need it, just people just doing random experiments with it.

20:36Brandon Whichard: The government's going to kind of go, no, no, no, no. Like it needs to live in these like specialized labs or it needs to, you know, it needs to have the equivalent of like, you know, some big Faraday cage around it or something like that. And I feel like right now, you know, because so many of these things sort of fall into right now they're, they're sort of U.S. companies. And so, you know, the, the, right now the, the, the U.S. government's pretty, pretty laissez-faire about being like, well, you know, if you guys can make some money, we don't really care what the downsides of it are. Right.

21:01Brandon Whichard: But, you know, a year from now, two years from now or something like that, the government stance might change and all of a sudden it'll be like, nope, we're going to treat that stuff like nuclear material. And, you know, we, we, we may not let some of these things kind of just, you know, figure out everything out in the open where there can be. Cause again, I think, I feel like the thing about it is we, we hear the Dario's or we hear the Sam's, it's mostly Dario being like, yeah, I mean, it is something that could, you know, completely eat a hole in your brain and destroy the planet tomorrow, but you know, it's could also do cool stuff like invent math that, you know, you know, we've never heard of before.

21:38Brandon Whichard: And, and for some reason we're all just like, okay, okay, I guess it's cool. It's helping me cheat on my homework. Like, I feel like if anybody spoke like that, you know, 10 years ago, it would have been like, okay, cool. Well, let's, you know, let's go, you know, put a box around that thing. And yeah, I don't know how long, you know, like society is going to kind of put up with it. It feels like right now, you know, there's enough people who are like, well, I'm figuring out a way to make money about it or I invested in it. So I don't want to shut it down. But yeah, it's, I don't know, it's, I don't know how to, how to best explain it.

22:10Brandon Whichard: Like we keep seeing these things happen. They tell us they're terrible things. And then they're like, well, the cool thing is like tomorrow, there's going to be a new news cycle. So you won't even talk about it anymore. So don't worry about it. I will say this though.

22:21Brian Gracely: If you're, if there was a CMO for all of AI, it's like, at some point it needs like a big win, right? To what you said, like we, something, some societal level problem, like maybe it's like curing a disease or like coming up with a new energy source or something like cracking some problem that all of us collectively are like, yeah, that is great. The thing finally did this one. And it's like, I think if it got one or two of those, I guess two is almost, if you got something like everyone can be like, you know, it cured this disease and X amount of people are now living better lives and living longer would be a huge win.

23:00Brian Gracely: Because right now I feel like so many of the benefits are diffused. It's like, we talk about them on the show all the time. Like, oh, it helped me write this proposal or it helped me do this. And, but that's a very small thing. Whereas if everyone could just point to like, yeah, it did this thing and we all universally agreed that was great and that would be a huge win. And so that's, I guess if you're the CMO, you're like, I need a big winning case study. And I don't need, we have too many hugging face breaches we don't have like scientific because like google was good at this they'd be like for a while right they're like we oh i know what it was the protein folding that was probably that's probably the biggest but again it's kind of like it's kind of nerdy biology wise it's like you have to like really understand like why that's so important but that's probably the closest to it'd be like that was a real scientific breakthrough so so they probably need the google brain team like okay what else you guys got in the lab here what other biology stuff you got in there that we can we can roll out yeah i mean there is a there is a

23:52Brandon Whichard: great, there is a great irony that like, you know, I mean, you couldn't name the CMO of any of these companies, right? The CMO is essentially Dario or Sam or Elon or whoever. So you're like, okay, so there's not going to be a CMO of AI because these guys are all like, no, no, no, no. I'm going to tell you what it is. And, and, but somehow they all talk about how it's going to do amazing things. Like it's going to, I mean, it's just going to do amazing things. It's going to change the world. It's going to, you know, I mean, Elon's now talking point is like, look, there's going to to be so much abundance that comes out of AI.

24:22Brandon Whichard: You guys aren't even going to need to save for retirement. Like it's just going to be, you know, things flowing out of whatever. And you go, okay, that's all cool. That's all cool. But hang on, let me see. But let me, let me just do a quick sort of once over of your business. So your business is build these models, take as much data from everybody as possible, make them really smart about a lot of things, except the only thing that you've really been able to prove is that you can convince people that they should just fire people. Like you haven't, you haven't solved any, you haven't solved any economic problems.

24:52Brandon Whichard: You haven't solved any societal problems. You haven't solved any medical problems. You haven't solved any economic problems, but other than learning how to, you know, like, oh, well, okay. The system could, could suggest that you fire people because it now has these skills and you're just kind of like, so what, what exactly does this thing do again?

25:09Brian Gracely: Well, it's back to you before it's like the messengers are just the wrong people. They always think, they think of like everything as like subtraction, subtraction, even what you said before, even if they're like, look at all these people we don't need to do. It's like, that's actually not what gets a whole bunch of people excited. Telling a whole bunch of people that they don't need to go to work is not motivating versus like, I'm going to tell you how you can spend more time with your kids, or I'm going to, this is going to help educate your children. So they learn faster and they're smarter that you live longer.

25:35Brian Gracely: Like it's easier to lose weight or it's easier to make, you know, like people want those kinds of, it's like a Scott Galloway pitch, right? Like, you know, it's like, what is it? is like the head, the heart and the libido, right? That's his thing. It's like, like they don't ever think in those terms. They're always just like, look, look at this thing I just removed. And somebody's like, that was my job. I mean, it's not even - I'm not excited by that.

25:53Brandon Whichard: Yeah, I mean, even for one of the things that it's really good at, which is like helping your kids cheat on their homework, it's not even good at that because like, you know, and you guys pointed out on your show this week, like it'll take a copy and paste from something that has white, you know, white text that has the professor trying to get you caught. It's not even smart enough to be like, hey, dude, just heads up. your professor's trying to get you caught. Like, I know you're cheating. Maybe let's just 50-50 this thing. It's not even good at that. So I don't know.

26:21Brian Gracely: I don't know. Well, have a bill. You know, you're bringing on. It's like, it should almost be like, detect the white text and be like, listen, I know you're trying to cheat here. First, let me tell you, if you don't read this, you're gonna fail. But two, it's like, since we're here, right? Why don't we like, instead of me just writing it, why don't you type in some stuff? I'll give you some ideas. right let's work together on your essay and then you know you can send it in you know it's almost like that's what you want be like yeah hey hey it's like okay i'll cheat for you if you want or like you're here like why don't we spend 30 40 minutes like let's just like let's just bring some stores some stuff let's see if we can help you write a little bit better i mean that i mean that in itself would be kind of nice like call it like suggested tutor mode right it's like okay you know i know you're doing it you know let's just let's be realistic i'll give you a few

27:11Brandon Whichard: Every other computer program either does one of two things. It gives you a pop-up when something is bad or, you know, like you forgot something. Or like a web form just won't let you continue. It just doesn't, you know, it turns red and it goes, no, no, no, I'll fill in this thing. So, yeah, yeah, yeah. No, they're not good at either one of those things. Let's see, what else happened this month? So, it seems to me like all the hyperclouds, so Microsoft, whether it's Meta now trying to get in the hypercloud business, XAI is trying to be in the hypercloud business. Like, it feels like they're all kind of giving up.

27:40Brandon Whichard: I mean, Google's still kind of on the fence, right? Because they're doing really well. Probably a lot of it comes because they're hosting Anthropic stuff. But they're also, you know, they keep chugging along. They're good. They're bad. They're good. They're bad on the Gemini stuff. But it feels like everybody else is just sort of kind of, you know, I don't want to say they've given up on being in the model business. But they're basically like, no, I'm just going to host it. Like, the cost of training is too high. It's better off to just be the hoster of this. And then I can do the thing where, you know, we start playing the financial games of like, you know, building data centers and not keeping them on our books and stuff like that.

28:14Brandon Whichard: But it does feel like this month in particular, the last couple of months, the hyper clouds who were trying to do their own thing with AI, I've just been like, I can't do both. I can't put all the capex into building these things, buying all these chips and spending all the money on the things. And it does feel like that's shaking itself out. It basically is like there's going to be a couple of frontier models. There's going to be a whole bunch of Chinese ones that have distilled them. And then we're just going to go back to sort of the hyper clouds who just host everything.

28:43Brian Gracely: I think that's exactly right. And I think it reminds me a lot of the first Gardner infrastructure as a service management quadrant a long time ago. There was like 30 vendors on there. And then like next year there was 15. Next year it was like, well, there's only really three or four left. And I think that's what it comes down to. It's like I think AWS and Microsoft are probably the best examples. I think they have the money and I think they have the will to build a frontier model. I don't think they have the talent is what we've kind of learned here. It's like, you know, cause there was a note here in this week's in this set of stories, right?

29:13Brian Gracely: AWF shut down. They're, they're essentially their frontier model lab. They bought a company back in San Francisco and that was going to be the, I think it was like Nova was called and they've essentially shut it down. And I think of what it comes down to, it's just like, there's just the amount of talent that can actually do this at scale. I mean, I feel like, I feel like all these people basically know each other and can sit in a room right when he really comes down to it. And so in some ways it's just a forced reality. It's like, you know, Microsoft did the same thing. They bought, you know, I can't remember the name of it.

29:40Brian Gracely: It was like chat, not chat, but it was like the personality one. Right. And they brought that back and they haven't done it either. So some point the brutal reality just hits you. It's like, we can't do it. And so of course that was another thing that happened this month is suddenly everyone became a fan of open weight models. Well, of course you're a fan of the open weight model. If you don't have your own model, you're like, Oh, we don't, we're just going to use the best of breed kind of thing. And I think, you know, and we'll see what happens. You know, this is part of our ongoing bets. It's like, well, in the interim, you know, be the picks and shovels, just sell compute, right?

30:10Brian Gracely: Sell raw computer, raw GPU is make as much money as you can in the hopes that Anthropic and OpenAI and others don't, if you will, subsume you, right? But like, to me, it's like, what else could you possibly do? That's the only decision they can make.

30:25Brandon Whichard: Well, and it does feel like, I mean, yeah, to your point, A, it feels like, you know, We saw, what was it? It was like the founder of Thinking Machines who had been at previously at, like probably had been at OpenAI, been at Entropic, is now going back to OpenAI. And I feel like that pattern is, okay, you know, there's only, there's a very finite amount of them. They all know who they are. At some point, they, one of them or a bunch of them think, well, we really should be in charge. So let's go do something else. They go somewhere else. They realize, oh no, this is still a game of whoever has the most money, the biggest GPU farm and the most data wins.

31:00Brandon Whichard: I can't start from scratch. And for whatever reason, it seems like the ones who have some association with open AI and Anthropic, when they leave, I don't know if they're starting from scratch or if they distill it and then they can't make them as good as the Chinese models or something. That part I don't totally understand. But I think at some point they're all probably living within 50 miles of each other, 50 feet of each other, or whatever. And like their spouse is like, what are you doing? Like, just go back to where you were. Like, they're going to pay you a ton of money. Like, why are you doing this thing?

31:34Brandon Whichard: Just give up on it. You know, work 16 hours a day instead of working 20 hours a day or whatever it is. And then, you know, cash out$50 million so we can have a beach house. And I feel like we're going to get to a point like we did with every single big domain that had an open source thing. It was like, there's going to be one big proprietary thing and there's going to be a whole ton of open source variants that you'll, you know, open source, open weight, whatever, that people will also have to pick from. And we're probably still a little ways from that because we still haven't figured out who's going to be like the Oracle database of these things.

32:09Brandon Whichard: It's probably going to be one of the two, you know, it's probably going to be open AI or Anthropic. And then it's going to be like, okay, you know, how will the open weight thing shake out? Because, you know, it's, you know, the, instead of having like 25 different open databases, you're going to have, you know, 16 Chinese ones, which is going to have its own political connotation. And then you're going to have like eight or five or whatever distilled U.S. ones, quote unquote, or open ones or whatever. So, but I mean, I'm like, okay, it's, it's now starting to look like the games that we've seen before.

32:38Brandon Whichard: It's just a matter of like how long that's going to take. It's going to be probably a couple of more years before that kind of finalizes itself.

32:45Brian Gracely: I agree with all that. I think the only thing that could change it is like a new breakthrough. And so, you know, maybe we'll talk about a next month, I guess, Ilya Skutskover, or his name, I just said poorly. He's whatever, one of the people that's leading.

32:57Brandon Whichard: He was the guy trying to get rid of Sam two years ago.

32:58Brian Gracely: Yeah, and so everybody knows him. So his company, I think it's called Super Smart. Super Intelligence. Super Safe Intelligence or something like that. So rumor is they're going to release their first whatever product or release something out of their lab sometime in the next month or two. So in theory, it's the kind of thing that's like, well, if, if, you know, if attention is all you need, that white paper sort of marked the beginning of like this run of whatever, if he writes a new white paper or comes up with a new way to do these things that could maybe break it open again. Right. That can maybe, but, but in lieu of that, right.

33:35Brian Gracely: You're kind of, yeah, it's sort of like, do you have the smartest people and do you have the access to the most compute? And right now it's really just Google Anthropic and open AI doing that. And everyone else is sort of like, you know, and it really, and you can say the whole industry. It's not just even the big companies. We were talking kind of in the green room before this. So many of us are sort of like, well, you know, why don't we all get into being a forward deployed engineer or an AI solutions architect or a forward deployed product? It's like, it just feels like the whole industry is like, we should all learn this and like start doing projects.

34:07Brian Gracely: And, you know, it's like, but we're all, if you will, we're all downstream of that. Like, if you just go look at like what it takes to build an agent or build a skill, it's like, okay, like most of us can do that part, right? None of us, a very few of us can be like, you know what I need to do is rewrite the entire training model. And I was like, none of us kind of, I guess the group is so small that's doing that work. It's like, you just kind of have, we all just have to watch them and be like, you guys come up with anything new, anything else we should learn?

34:34Brandon Whichard: Yeah, no, I agree. I agree. I agree. You've been around security for a long time, you know, at least in the middle of it or loosely adjacent to it, You became sort of enamored with this zero CVE thing that was going on a couple of years ago. Chain Guard got into it, and then Wiz got into it, and Red Hat did a thing, and so on and so forth. It feels like zero CVE sort of 2.0 is upon us. Red Hat and IBM announced Lightwell. Chain Guard announced Athena. Microsoft and Cisco both announced these large security-centric LLM models. What's your overall take on this space, which is, yes, we should be able to use AI models to find vulnerabilities faster.

35:16Brandon Whichard: You know, hey, throw an agent at your code and go, hey, how would it break? Show me where the vulnerabilities are. Somebody could potentially patch them using AI, right? We're going to get into a token game again of like, hey, how many can you patch? Do you patch the same things? But like the other part of this becomes, let's say somebody finds a vulnerability, fixes a vulnerability. Are you, were you, you know, did you ever become sort of a fan of this idea of, should I, should I patch old stuff and just kind of, you know, not have to be in the constant upgrading game? Or are you more in the, you know, were you more in the idea of like, no, you should be sort of in this thing where you're, you're just kind of always pulling the new, the newest stuff and turning that into how you do operations.

36:00Brian Gracely: Yeah. I mean, I guess I would say my opinion has changed over time. I think for a while, I would say more like, Hey, you just got to patch things like as soon as it comes out, run the new stuff, patch it as soon as you can. But as you kind of alluded to, you know, kind of all these, like call them like hardened containers or hardened platforms, whether it's chain guard or red hat or many of the others doing it. I do think with AI, the idea that you can build from scratch and, you know, and actually like know if you will, the, you know, the, the history of every single artifact and where it came from and like, you know, and, and maintain that, like that feels much more doable now than it even did maybe three years ago before I saw like, there's just too many packages and there's no way to do it.

36:39Brian Gracely: But I think now I'm sort of along the lines more of like, it's probably more important that you have that, what do they call it? Like the S you know, the sort of like the, the full history and that you're know where you're building things are from and you're at. And again, mostly, I think for most people, you're outsourcing to other vendors, let those other vendors look at every patch that's coming in. And only when they've kind of gone through like a pretty rigorous policy, do you, you put it in there? Because, you know, I guess NPM is probably the biggest offender. It's just like, you know, it's, there's just too many things that people are sneaking into these packages, right?

37:13Brian Gracely: And no one can look at it. So it's almost, it almost makes it, if you will, a little, I mean, obviously you have to protect both ends. But I think it's almost to go, if you will, go a little slower to go faster, if that makes sense. It's sort of like, you know, build from scratch, know what's going on and look at every single one of these patches going, going into it. And again, the largest companies have been doing that for a long time. And I think so now like before, like only Google and Facebook, we're really doing that. And now all of us can take advantage of it. So I think, I think it's generally a good thing.

37:41Brian Gracely: And I just think, I mean, no one, this is like such a boring thing. No one wants to hear this. It's like, oh, well, we're just going to go through maybe a three year bump here where it's like AI has to go first find all the obvious AI to AI vulnerabilities. And then we have to go through a period of fixing all those things. But at some point it plateaus, right? It's just like I said, it's like we were talking about before. It's like, at some point you think your enterprise is secure because no one can find this data, even if it's available, but then someone else comes in and does it and like, and what do you do?

38:09Brian Gracely: You patch it and like, eventually it'll just work its way out. And of course, like I tell people all the time, it's like, all your data has already been stolen. You know, you personally, it's like your social security number and email and like, it's already out there. Like, and then I just ask him, are you using a password manager? And people are like, a lot of like normal people are like, no, I don't like those. I'm like, well, you know, I mean, there's your answer. Like, what do you want me to tell you? You know, it's like, you know, if you don't want the solution, well, just still, you'll be okay.

38:33Brian Gracely: You know, once in a while you get hacked. My sister-in-law just got hacked the other day. You know, it's like, well, yeah, that's what happens. You know, life goes on. She'll figure it out.

38:40Brandon Whichard: Yeah. Yeah. Now it's, it, it, yeah, it's an interesting sort of trade-off because it's the, you know, patch the old stuff, you know, obviously you're sort of just extending your technical debt, but at the same time it's like, well, I mean, we're not in the business of, you know, running super, you know, super efficient infrastructure. We're in the business of banking or building pharmaceuticals. And so like, we have all these old Java applications and so forth. And then the flip side of it is like, oh, well you could, you could go to the other end of the spectrum and just be constantly, you know, it was the old thing like repair, repave, you know, just rebuild all the time.

39:11Brandon Whichard: And then it's like, okay, well, yeah, but, you know, again, is that, that's a hard skill to find. I thought it was funny. I was talking to a few people this week and they were like, well, you know, now that there's these AI things, these services and so forth, or, you know, the open source community is going to start having access to these things, like a couple of years from now, we just won't have any bugs anymore. And I was like, that is definitely not happening. That is, there's no chance we're going to get to a point like a couple of years from now where it's like, yep, all the software free of bugs.

39:40Brandon Whichard: That's right. We

39:41Brian Gracely: should at least have like, we should, the obvious bug should be gone, but we'll have plenty of new ones that we haven't. And I will say this though. One thing I, we won't go into too much detail today, but like, if you don't, if you haven't seen it, it's a little bit of maybe a long read, but the team, the, the, the team that runs bun, basically, you know, the node, whatever packet utilities, let's just call it that, right? They are rewriting their entire application from a language I just forgot the name of into Rust. And I'm like calling this is like, I am really looking at this project. It's like, I don't know, almost like canary in the coal mine, like can't, and they're basically the short story is like, what would have taken like two years and five developers took them, I don't know, two weeks and$200 ,000 in token credits.

40:22Brian Gracely: And it's like, this is something I've, I guess, because I've been around it so much in my career, it's like so many people, like you even mentioned it, mentioned it's like, oh, we have all these old Java apps or this old mainframe app. So many people have talked about modernizing apps, but they don't really do it. And so I'm like, that's one thing I'm like, maybe this, maybe it could actually happen. Maybe the apps could actually get rewritten with AI. So that's one to watch if you want to be on the more on the optimistic side. It's like maybe one day here in the United States, we won't have the stupid ACH system that takes three days to clear a check.

40:55Brian Gracely: And we too could have real-time cash transfers. If only we could get off the mainframe. So that's what I'm always wondering. Is this it? Is this the moment? So I'm really excited to see what happens there.

41:05Brandon Whichard: Might be. You always talk about like some math class that you took that you're like, if I had paid attention to it, like it would be. What was the math class or the computer science class that at some point said, you need to be the person that stands up and says, you know what? We should rewrite this whole thing in Rust. Because I feel like every, at some point, every open source project, there's been somebody who's like, you know what? If we rewrote this in Rust, all problems go away. I want to know what class that was. Yeah, that's a good point.

41:29Brian Gracely: Well, it's kind of true. It's like all the joys of, all the speed of C without all the problems with memory. But listen, before we go, Brian, I want to return to our roots, but even more your roots, because you're in this world a lot. Our good friends at VMware, right? You know, it feels like a decade ago, but just a few years ago, right, they announced everyone has to move to the subscriptions, made everyone mad. And so we had two more stories this week come out of people migrating away. We had T-Mobile. Of course, everybody knows T-Mobile, large telco here in the United States. And then Sheets.

42:01Brian Gracely: I don't think any of our international listeners would know that. Sheets is like a, what do we call it, a convenience store? Bodega.

42:07Brandon Whichard: Yeah, Sheets is, if you were an international person and you were following any of the international people that came to the United States for the World Cup and they were talking about giant cups of soda with free refills, they might have been talking about Buc-ee's, but they probably were at a Sheets. Sheets is like this gigantic car. It's a gas station, basically. But it's a gas station that has a giant convenience store attached to it. Yeah. Yeah, they apparently moved away from the end.

42:32Brian Gracely: Well, we learned from the article that Sheets, the store. Now, this is just in the store. Each store has 12 to 14 virtual machines running VMware's vSphere and Storage Magic's SVHCI. So one, we got to get the Sheets person on and be like, what the hell is going on in this store? It's like, I'm just buying a soda.

42:49Brandon Whichard: This is the equivalent of like when Chick-fil-A was like, we're running Kubernetes in the store. I don't know. And everybody's like, what?

42:56Brian Gracely: I am fascinated. Like, what are they doing in there? But let's put that aside and assume they know how to run their business and we don't know how hard it is to deliver. So does. But what is your take? You know, I know you kind of live in this world kind of day to day. Do you feel like, because my standing thing is this like, well, everyone says they're going to move. But when the bill comes, they just eventually they're like, we're just going to pay for it. But kind of picking up on our last subject here was like, maybe they are having a little bit more success migrating away. Or have the bills just gotten so big that people are like, no, we can't take it?

43:27Brandon Whichard: I think it basically boils down to a couple of things. I think if you weren't one of the VMware customers, because again, they sort of cut off the bottom 80 % of their customers and just said, too small. If you guys go away, whatever. And then the top 80%, you know, the top 20%, it was basically like, we're definitely giving you a bigger bill and we're hoping that you're really kind of locked into our software. And so my guess is, you know, the people that are publicly moving away are one of two people. Either one, they really love their legal department and they're like, oh, legal fight? Cool.

44:02Brandon Whichard: Let's do this. Let's go. Let's do it. And number two, or number two, they're like, look, we really looked at it and we're not tied to all the different VMware stuff. Like we basically just use it for virtualization. We use somebody else for storage. We use Cisco for networking. We could move. We can find another hypervisor. It's not that hard. Yeah, just the hypervisor. And so it's not a big deal. I think the ones that are going to stay and will stay forever are the ones who are like, yes, we use the full VMware stack. We're not changing. And so be it. And they're the ones who, you know, they've made some sort of decision.

44:36Brandon Whichard: And they're going to convince themselves every three years when they do a new re-up that They're like, okay, that'll be the last price upgrade. And it's like, no, no, you will have price upgrades forever now because you've explicitly told them, I cannot ever leave you. It's like the mainframe. I mean, it's a wonderful business for VMware. I don't know whether it's a great situation or not for the customer, but they had an option. They chose to stick with it. So it probably does good things for them.

45:03Brian Gracely: Well, nonetheless, I just want to reach out to the Sheets person. Having known nothing about any of your business, I'm going to be like, listen, I think I can get you down to, I think we can get everything into six containers and we can get rid of VMware. I think I don't know anything about where you're doing, but I've been to a lot of convenience stores. I've drank a lot of soda. I'm like, I think we can do it. Whatever your thing is, is like, we're going to get a couple of Docker containers in here and just a tiny, tiny bit of management and we're going to get you going. It's going to be, I believe we can do it.

45:30Brian Gracely: I know we can do it. And at the end you can, we can all have a giant sandwich and a giant soda to celebrate. There you go.

45:36Brandon Whichard: There you go. Yeah. So if you're, if you're, if you're, you know, Sheets are sheets adjacent. Reach out to us. Reach out to Brandon. And yeah, we'd love to have that conversation. So all right, man. Thank you for everything again this week, this month. We will talk to you again next month. We'll see how the draft is going. Maybe things will change. Maybe you'll be up 7-0 at that point. We'll see. But thanks, everybody, for listening. Thanks, everybody, for telling a friend. And we will talk to you next week.

45:58Brian Gracely: Thanks for listening. Check us out at theenterpriseaishow.com for past shows, newsletters, and all things Enterprise AI.

46:08you