AI News of the Month - August 2026

0:08Aaron: Good morning, good evening, wherever you are, and welcome back to the Enterprise AI Show. This is your host, Aaron, and this will be a super quick intro this week. It is time for our AI News of the Month. Brian, Brandon, and I discuss the biggest AI news stories from the month of August. Topics include NVIDIA's record quarter, the acquisitions of Hugging Face and OpenRouter, and OpenAI's jalapeno chips, and more. And all of that is coming up right after this break.

0:39Aaron: What does the outside world already know about your company? Cyber criminals could be seeing leaked credentials, compromised session cookies, exposed infrastructure, and even impersonations of your brand and executives. Nord Layer Intelligence by Nordstellar gives security teams visibility into those threats across the deep and dark web. Data breaches, attack surfaces, and brand impersonation all in one platform. Find out what attackers know before they can use it. Visit nordlayer.com slash intelligence slash enterprise AI and use code enterprise10 for 10 % off your Nordstellar plan.

1:27Today'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.

2:01See what it takes to activate your data for AI and request a demo at nasuni.com slash AI.

2:10And we're back. Another AI News of the Month show. Got all the boys here. Brandon's here. Aaron's here. Guys, how are we doing? Fantastic. We are rolling into September, which means, guys, ultimately, no matter what we talk about today, we made it. We made it to football season. The thing that most people really want to talk about. Aside from just general excitement, general, now you can plan for your weekends, any just general thoughts about either the NFL or college football or any other form of football that you're excited about right now? I'll just say I'm enjoying the concluding of the absolute chaos that has been college football, where it's mostly just been lawsuits and more players and coaches arguing.

2:47So it's just, it's almost like covering this, the AI stuff. There's the actual, what's really happening? And then there's the palace intrigue. So I'm looking forward to what I think will actually be the game. So the next, what, 14 weeks before we then return to more legal action, I'm sure in the off season.

3:01Aaron: And based off the latest legal action, it's just going to be one big league. Pretty soon. There's not going to be college or pro because all the players can just go wherever they want. Yeah, we got that. I went to my second cousin, plays for the number one team here in North Carolina. So I went and saw him play a high school game, got all sorts of insider information on NIL at the high school level, which was fun. Yeah, I think we get to stop talking about sort of the money and lawsuits and stuff. And we get to just watch games. We get to watch Pat McAfee 24 hours a day on some channel, which for some people, I guess, is fun.

3:35But yeah, so let's jump into it. We're going to do mostly AI News of the Month. I think we couldn't come up with a worthy cloud news of the month story. But why don't we start with NVIDIA made some money. Brandon or Aaron, any thoughts on how much money they made? Did they make enough? Did they give you enough excitement about what the next quarter's sold out GPUs look like? What did you think about the latest announcement from NVIDIA?

3:54Aaron: First of all, it's never enough. But Brandon, what's your analysis of all of this? I mean, it's just another blowout quarter. I think we always get on here and we talk about, is it a bubble and what's going on? And then NVIDIA comes out and announces record earnings again. and then they will go on to it, then announce and forecast record earnings in the future. So for all the Ed Zitron and all the bubble watchers, at some point you just kind of, I don't know what to do. You throw up your hands. You're just like, okay, maybe the money isn't coming from enough customers or maybe the revenue isn't coming in diversified way.

4:29But at the same time, it's like, we've just never seen a company at this level continue to grow at this level for this long. So saying that it will one day end, I guess this has to be true. When it will end, I have no idea. It just seems to be an endless amount of money. What were your thoughts, Aaron?

4:46Aaron: No, I was going to say it's super interesting to me because your exact point, everyone's predicting bad times ahead for NVIDIA. But yeah, and it might happen eventually, but it doesn't look like it's going to be this quarter. And then they're starting to do a lot of acquisitions. They're starting to help. We just did a podcast on it here recently. They're financing a lot of the stuff in the industry right now. And so they're making record money, but then they're also, I feel like a lot more so than a lot of other companies, whether it's partnerships, whether it's acquiring things, whether it's helping others secure financing.

5:18Aaron: They're definitely putting their money to work. Yeah. Probably more than any others. Brian, what's your thoughts? Yeah, I think their numbers were up. I think they forecasted their numbers up for the next quarter. Yeah, for next quarter, I think they expect to be growing again like 70, 75 % or something like that. I think the two numbers that caught people's attention, free cash flow was way down. So they went from, I think their normal sort of free cash flow was like in the 40 or 50 billion dollars. So their free cash flow was down to like 21 billion dollars, which basically means people like aren't necessarily sending them money.

5:51They're sending orders out the door. But and then the other part of it was they apparently they financed a ton of this stuff for customers. So again, goes back to the thing we talked about a few weeks ago that people are more than happy to say we would love to have more gpus because there is a bunch of usage demand but as far as actually paying you we're starting to get into i gladly pay you tuesday for a gpu today i think they said the numbers the dates on their free cash flow it dropped back in the 50 60 day range people aren't paying them nearly as quickly which means they're going to have to start figuring out how to actually make some money with these gpus that i think is the the one thing to take a look at because we've always said that the challenge with this is outside of Anthropic, who seems to still be announcing numbers that are growing and maybe even growing faster than people believe, the rest of their customers, whether it's the hyper, what do they call them?

6:40The hypercloud, not the hyperclouds, the neoclouds, the hyperclouds, like those folks don't necessarily seem to be equally reporting. The neoclouds, I think. The neoclouds, right? The core weaves of those guys of the world don't seem to be announcing blockbuster numbers, nor do we hearing sort of blockbuster numbers out of open AI. If you look at their biggest customers, they haven't yet quite figured out how do they make money in the same way that NVIDIA makes money. I'm super curious to see, because you would think of the companies that should have the best look into the backlog of who's making money.

7:10You think it would be TSMC and NVIDIA, right? Because they can actually meet with the customer. They can actually have a meeting with Andy Jassy. Be like, okay, show me what you're selling, right? Before we sign on, because it's not even arguing, it's almost like you're not even discussing price anymore. I have what you want. We know what you're going to pay. Now, I want to see what are you actually doing, right? I want to see how quality, because if your revenue looks, if you will, shaky, then maybe I give it to somebody else. So that's the part I'm really curious about. What do they do when they go in there?

7:40Especially TSMC, who's the one that has to actually on the hook for giant fabs that cost tens of billions of dollars. They have to really be looking at that and saying, I can't just have you call and cancel this order because I'm literally spending$20 billion or$30 billion in advance of that. So it would be interesting to see what kind of due diligence that's going on. And maybe I'm naive. Maybe no due diligence goes on. You just take the order. But like I said, with TSMC, I'd be worried if I were them. That's why I think they would probably require more than just an order form. I think they, I think, so if you're TSMC, you don't really care that much about any individual customer, like not making their thing because they're buffered.

8:21It feels like to a certain extent, they're buffered by two things. One is this, this idea of, okay, if it didn't go to AWS, like Anthropix is going to hoover those up or OpenAI is going to hoover those up or whatever they are. Right. I think in the other part of it, it was, I think they probably feel a little bit buffered by the fact that NVIDIA went out and found all this, was it$500 billion worth of funding that they're going to back 25 % of it through all of these financial instruments. Right. So to a certain extent, they're like NVIDIA. I mean, NVIDIA is both sides of the market at this point and there's buffered that.

8:53And I think the other part of it is I think they're all, to a certain extent, they're playing a big game of chicken or they're trying to bluff each other about how well it's going. So I don't know. I think NVIDIA probably knows pretty well because they're the ones who actually get the funding from it. I would think the funding guys are looking at how many checks are showing up in the mail. But I don't think TSMC, I'm sure TSMC has a certain amount of insight, but I think their biggest risk is NVIDIA at this point, because NVIDIA is sort of the market and buffering the market on the finance side.

9:25Your point, I guess I was just thinking, as you were saying, is anyone really incented to look that hard, right? TSMC is like, okay, yeah, it looks like you have the debt facilities and I'm confident you're going to buy. And it's like, yeah, I'm confident Anthropics is going to buy. And so I don't know, maybe that leans back more into the speculative side of it. But I will say this, though, NVIDIA, I do give them credit in the sense of like, they are out there buying a bunch of companies. They have a very highly valued stock and they have a lot of money. And I think the only thing you can do when you have that is like, you need to go do things with it.

9:54You can't just, I guess the other thing is they could just write a big dividend check back. But for them to keep going and make all this work, they need this ecosystem to come alive. So I give them credit. They're trying to do it. Maybe we'll talk about some of their acquisitions recently. But I don't know. I think that's all you can do at this point.

10:09Aaron: Let's kind of talk about, so they picked up Hugging Face. That's probably the biggest one, right? And so I was at 13 billion, 12.9 billion. And I think it's a super smart pickup for them because they've been trying to go up stack. And we've talked about it many times about, hey, you need to be more than GPUs. And hey, they're doing the Nemetron family of models. And then now, hey, I want to go buy the model repository. It's a little bit like Microsoft buying GitHub back in the day, right? Kind of deal. go where the developers are, go where the model repositories are. I think it's a pretty smart move for them overall.

10:43Aaron: And hey, they got the money, why not? What's your thoughts on it, Brian? If you guys had been asked a week ago, two weeks ago, at a dinner party to explain what Hugging Face did to people, for people maybe who aren't data scientists, how would you have explained it? I would have said like the GitHub of the models. Basically, it's the place you go put all your models so that everyone can go check them out. So this is where it always felt like Hugging Face was somewhere between like GitHub and like Docker Hub, right? Like it was the place where you downloaded stuff from, which sounds like a terrible business because all you do is you own compute and bandwidth.

11:18It's hard to monetize. And then the second part was like, how much data do they have that gives you insights into the marketplace? And it never was weird to me is like, I never felt like Hugging Face provided that much market insight other than benchmarks. But to people in the data science community, did they look to Hug and Face for a lot of information other than I can go find a bunch of models, I can go figure out benchmarks. But was there really a lot of unique insight they used to have that may be really valuable to NVIDIA? I don't necessarily think so. I guess I've come to terms with this acquisition more around of they want to promote more models being out there, right?

11:59And so the number is so enormous, it's a little bit hard to get your head around. But if you just say to yourself, I want more models out there and I want it to facilitate people publishing them. I want to open weight models and I want as many as I can. And you say to yourself, I want to inject money into Hugging Face so that more and more people are doing that. I guess to me, that seems to be, if you will, falls into the strategic acquisition bucket in the sense of like, you're not really expecting to ever make$12 billion in profit. You're really just expecting it to, if you will, grow the whole open source and open weight models.

12:36So I think that's the only way you can, at least I can get there. I don't know, looking at itself, and the best analogy would be like Microsoft's acquisition of GitHub. I guess everyone can decide for themselves, did that work out for Microsoft? Did they make the money back or did it give them some cachet? Because that's the best analogy that I have. Yeah.

12:54Aaron: I think, to your point, Brandon, who's this trying to fend off? I think NVIDIA is increasingly, OpenAI and Anthropic and Google are going to be all threats to them. Because if any of them take over the vast majority of the market and they move away from NVIDIA chipset, it's such a huge hit to them. And yeah, I think they're definitely trying to do open models. They're trying to do open weights. they're trying to promote this ecosystem as much as they can because they don't want that ecosystem to die on the vine and it become a big model in the sky is the only thing kind of thing now the strategy of it makes sense i think like you guys said it's they want to be at this point they want to be all over it's going to be a little weird for them because they've always they've always been about like hey we're the best at every layer of that five layer cake that they like to talk about even though they didn't own the models they always wink nod nod were aligned with either open ai or anthropic to a certain extent and now it's no we're aligned to everything else so it'll be interesting to see how they what they do for real what they manipulate so do the open models always stay at the top of the benchmarks now that it's owned by nvidia right or the other ones does it go through something like that are they like the concern that always happened when github got acquired was like, oh, Microsoft's going to, at the end of every build, they're going to be like, hey, would you like to deploy this over to Azure?

14:23And they never really did that, right? They stayed pretty independent. They did stuff like GitHub runners and you could make them go wherever. It'll be interesting to see if NVIDIA, like how much they push on people to move over towards CUDA if you're not already running CUDA as some of the underlying libraries. Where do they start? Where do they start taking some of this up the stack stuff into Hugging Face and make it a default experience and so on and so forth. Yeah, I think strategically it makes a lot of sense. What did GitHub got bought for like about$6 billion, something like that. So at$13 billion, you look at this and you go, yeah, that makes total sense.

14:56You're paying 2x for sort of all of the open market data, which in today's times probably is pretty cheap, actually. So yeah, it makes sense. But yeah, it's another one of those ones where you're like, okay, did they start working on stuff right away? Or are they like, this is just a hedge and we'll see where it goes until one of of them comes out with their own chips or something. I do wonder, I wonder sometimes maybe we put like too much analysis in the whole thing. It's because I was thinking back to some of the other things you see that like come and go. Remember OpenClaw and then NVIDIA had, was it Neoclaw?

15:26Neemoclaw. Nemo. Is it Nemo? Yeah, it's like, yeah. Yeah, Nemoclaw, yeah. Yeah, it was like the, and it's, I feel like that, I've never heard another word about that since like, maybe it's going great. I don't know. But it just feels, that was like really big around that time of that conference. And it just, who knows if it hits, if it doesn't. If you're an NVIDIA, it doesn't matter. You're like, let's just fund that, see if it hits. If not, we're on to the next thing. So maybe the same thing with Huggy Faces. Yeah, they have a lot of good stuff going. We want to keep this going. Let's just pump a bunch of money into them, see if something hits.

15:53And if not, if it's just another website out there that has the models, so be it. It's a$12 billion lottery ticket or insurance, depending on how you want to think of it. Yeah. Yeah. And the only thing we haven't heard anything about, who was the company they bought? Was it right around Christmas time? The one that they bought and then it was like an alternative GPU thing.

16:10Aaron: Oh, Grok? Was it Grok with the key? Grok of the Q. Yeah, it would be interesting to see if they start looking at the models that are in Hugging Face and start dividing them up and, okay, Grok team, you go after this chunk of them for certain things. And Grace Hopper team, you go after this chunk of them and see if they start trying to build some sort of optimization around that thing. So yeah, it'll be interesting. Plus, it'll be interesting to see how they've got their own, like you said, they've got their own Nemotron models, which most house people say are pretty good. How do you stay independent, but also have this thing that you're trying to push as the, hopefully their hope is the default open one, if you will.

16:45I do think as we maybe wrap up Hugging Face, it's a fun story. Everyone should go read the founding or the creation of Hugging Face. It was an AI chatbot for kids. But somehow that's where it started. Yeah, that's where like the hugging comes from. And they completely, you know, to their credit, they've completely pivoted away from all of that. Now they're this. Sounds like a good pivot, man. Yeah, so it was just, it's just funny though. So for all of you out there that you think you have to have the idea all figured out, there's yet another example of what they were actually bought for and what the original investors, because they had some celebrities that made tons of money, what they actually invested in had nothing to do with what they actually ended up building and selling.

17:25So good for them. Yeah. Stripe, the payment company, acquired a company called Open Router for$8 billion. How do these two things connect? How do you connect the dots on this thing? you want to go first brandon yeah sure i think the the dots are stripe is a rails for all payment right and so their thesis is going forward that ultimately it's not going to be one model that everybody uses is that you're just like you have different credit cards when you swipe you lots of banks have credit cards and they all use the same set of infrastructure they would like to be the same kind of idea there they'd like to be in the middle and every every request if you will ai request or AI agent making requests would go through OpenRouter and the OpenRouter would route that to the most cost efficient or best model to do that job.

18:12And then ultimately, I think Stripe even sees there to be the agents will one day be paying each other in some way. So it'll be some kind of financial movement between them. And they would like to, if you will, stay within that business. And another little homework assignment, if you would like to read an interesting letter, you should go read the Stripe letter, which is published many places where they announced the singularity began on January 1st of this year. So they have it down to a specific date, not a time, but a specific day. So you can go and they basically say that is the identification of singularity is why they've now decided to make the moves here.

18:44So it's, I don't know, it's quite the letter. You should read it for yourself. Do you think we'll have singularity parades and other things soon, like next year? Do you think those will kick off? I don't know. As I said, you could listen to a lot more discussion of this on over at Sovereign Talk. I don't know, man, And to actually put a day, to put it down to a specific day seems quite, you're quite confident in your precision. If you're to be like, it happened right on this day. The 31st, there was nothing. And then the first it happened. And then just to your point, I'm not even denying something could happen or has happened or will happen.

19:13But I'm just like, I don't know. I was January 1st. I think I was just hanging out watching football. I don't think anything was really, I don't think anything notable happened that I could detect. But maybe that's me. Maybe I'm not paying attention. The Stripe thing is interesting in the context of it. But isn't Stripe also trying to make a big play on PayPal? No, they gave up, but they had been trying. They announced that was over. But in that context, you kind of go, okay, if you're going to be building out this thing, then eventually, because sometimes these things go away, and then sometimes they come back again at some other price point or something like that.

19:42Somebody figures out another thing. It would make sense if they were, it was like Stripe plus PayPal, which has hundreds of millions of people who use their service. And then maybe they, I know they were trying to do some stuff with Block, which is the former Square company, which gives you a point of sale. Those sort of things all put together look a little bit like Visa and MasterCard. They look a little bit like Amex. And so you're like, okay, if that's the case, then I can see why you would pay a huge amount of money for kind of core infrastructure to be able to do routing and now routing with this other stuff.

20:13It'll be interesting to see if they do the thing, who was it, was it like Spotify that came out with Backstage and people were like, this is weird, a music company came out with this Backstage thing, which is platform engineering. It'll be interesting to see if Open Router stays out there as something people can keep using. I guess maybe there'll be a free version that you can download or something. But does it just go away because now it's just part of infrastructure for this payments company, which probably makes more sense if you're going to pay that much money. You want those engineers working on stuff that doesn't just look like you or I being like, I'd like to pay a little less on my Cloud account, routed to something else.

20:47Another little homework assignment, if you guys want to go read about it. The co-founder of OpenRouter was the founder of OpenSea, which you may remember was the NFT marketplace. That's where you could sell your weird images and stuff. So it's just interesting. This marketplace over in the NFT world, I think that has largely not worked. It's kind of collapsed. And he basically pivoted to let me be some kind of marketplace for models, which is now OpenRouter. But again, it gets back to this ongoing discussion. It's like back to even in Stripes thing, it's like, I don't know if the singularity is here.

21:22Does the singularity need a route, a router between models? Won't the singularity just be like, I know what models I can use. I'll just use them. I don't, so I don't know. I mean, you really like decompose the whole thing. You're like, if this is true, then I think that's not needed open router. But maybe not. It's just a weird, it's hard to get your head around it. That's the, that's the, I guess that's the weird thing about like when people talk about the singularity or HEI or ASI or whatever the latest buzzword is. They treat it as if it's the thing in that latest Tom Cruise Mission Impossible movie where it's like a thing running somewhere on a big server somewhere in a mountain or whatever.

21:57But they all seem to talk about it as if it's just a benchmark we've reached. It's like a feeling you get after somebody hits a asset or whatever. It's just like a thing. Whereas you're talking about it in the sense of, oh no, the singularity would know that I don't need open router because it does its own. And you're like, Like, so that's the part of this where you're like, okay, I don't, I'm confused. Is the singularity a thing or is it just like a, it's like an experience or it's like a metric that we've reached or something like that. And I don't, I don't know. It's all vibe space. But Aaron, I wanted to get your take on this other thing that happened this month.

22:27I thought maybe you're closest to it. There's been open AI has been doing all this crazy testing, right? And so there's the whole hugging face break in. But there was this article that came out that basically said, that actually did show some of the, let's call it singularity behavior. I don't know if that's a word, But it basically showed the agents, the models, they figured out they were given an impossible problem. And then they kind of started to work together to be like, huh, maybe we could lie to the human that we can do it and we can change our logs. And they even have agents, they created a bulletin board amongst themselves, right?

23:00And different, they called it generations of the agents would go read the bulletin board. And I was like, this was the first time where I was like, this is weird. This is, I don't know, this has a feeling that I know we're not supposed to think of the AI as people, but I was like, this kind of feels like people are working together in some weird way. So I don't know. Did you have, did you follow this at all? Did you, what do you make of it?

23:21Aaron: So here's what this reminds me of. And this, you know, a lot of our analogies sometimes is all of us being fathers who goes back to kids and experiences at school. And it goes back to the post-mortem on the hugging face hack with OpenAI, right? They had over a thousand agents and then talking in message boards and they exchanged like 70 ,000 messages and manipulated logs and did all these other things. And it just reminds me of all the different ways our kids, they always found a way around the system. There was always like, oh, if you open up a Google Doc and just share it with everyone in the class, everyone can type the answers in real time.

24:01Aaron: Everyone always figures out a way to keep cheating. And the AI agents are basically just like middle school and high school kids. They're just like salespeople in quotas. You give me, show me your quota and I'll tell you exactly how they'll try and sell your product. I think the thing that always surprises me at this point is that even the data scientists are surprised that when they give these things a task and they do whatever incentive program around it, they're like, the thing doesn't give up. And you're like, yeah, of course you gave it unlimited tokens. It's just going to keep trying to do this thing.

24:35And it's almost like they seem to forget that they're not humans. They're not there. They don't have morality. They don't have any of those sort of things. They're just going to go figure stuff out. It's actually like the best talking. If you're ever talking to somebody who's new to this stuff and you're just like, hey, imagine this scenario in your business. Like, how would that work with a human? Like, for example, hey, you tell Tom, hey, Tom, go see if you can break into the HR system and steal some payroll data. They'd be like, Tom would try and he'd have some credentials and then he would get like a 404 and he'd probably give up.

25:05And you're like, OK, cool. Let's say Tom never gave up. What do you think would happen? And they're like, what? And you're like, yeah, that's how agents work. Like, that's the starting point you should have for like agent security or sandbox security or something. And they're like, oh, OK. Because they just have never experienced a thing that just doesn't stop trying until we don't show an example and we stop giving it tokens. So, yeah, it's interesting. So there is a, I think we put a thing in the show notes. I know I put a thing in the software defined talk, Slack. There's been two or three of these that somebody had done some research on that really haven't gotten as much public talk.

25:39But yeah, there's several of these sort of stories that are floating around, at least among the data science community, which doesn't seem to be freaking them out enough. But they don't seem to be freaked out about anything. They do seem to be freaked out and then they stop being freaked out. Yeah. I was going to ask Aaron, do you think it's a sign of anything new though? Is this just what Brian's saying? Is this just, yeah, you give a computer program a task, it'll just work forever until it runs out. Or is it like, is this, it should like not make light of the singularity for a minute. Or is this like some actually new thing happening that we should be like, okay, this is like, we've reached the new milestone.

26:08Whereas like you could do, you could, in old school, right. Procedural program, you can give a computer and it will just like loop forever, never do anything. Right. It will just keep doing the same calculation. Whereas, okay, these agents, like they, they try different stuff, right. For a long period of time. Is that new or is that just the same?

26:23Aaron: Here's the thing. Like I feel when it comes to these agents and like giving them these tasks, like it is interesting to me in the ones I have read, like they just were too open in what they gave it to it. Like they didn't put enough guardrails around it because I don't think they anticipated how far it would go. and I think everyone's now starting to go, oh, yeah, these things, like Brian said, they will never freaking stop and they will make up things and do things that most people hadn't even thought of because they've got infinite tokens and infinite things to try and they will just keep coming up with stuff and keep coming up with stuff.

27:02Yeah, I think the difference in the loop example is your loop example, you would just kind of keep like hitting the wall with a hammer. In these examples, I've seen examples where they're like, yes i'm this agent sitting out on a network and i can't get in and so i'm like okay what are all the other things on the network oh that one's a that one's a kubernetes cluster okay let me go try and hack the kubernetes cluster to try and get all of the secrets that it has sitting in there like you that wouldn't happen in a loop right because it would now you're moving from do task a to now do task b and then if you figure out something in task b do task c d and e i think that's the piece that's different that people sometimes struggle to get their head around which security people are all probably screaming in their car right now.

27:42They're like, that's what I do all day long. And that's why I tell all you guys know all the time because you never think about all these other things, but it's these things are very good at that stuff, or at least they, again, they don't give up trying because you told it not to give up. We'll put a link in there, but it's, I think it's Dwarkesh, The Rise and Fall of Asian Civilizations. He's got a good write-up about it and it's very accessible. So you, I don't know, you can read, decide what you think of it. Civilizations are just weird computer programs doing stuff. But it's an interesting read.

28:09Yeah. The good news is OpenAI put out a letter this past week or so, and I don't know, 30 or 40 companies signed on. And they basically said, yep, these new AI tools are super scary for security. Instead of not releasing them, you guys should just patch all your shit. Get all your shit patched. So there you go, people. 2027, the year of patching your shit. You didn't have to do that before, but now you do because the tools are really scary. What else we got on our list? Microsoft spent a lot of money. Their employees spent a lot of money on AI. They were surprised about this. Brandon, if you had unlimited money at Microsoft, do you think you'd spend$28 ,000 in 28 days?

28:43I know you've been working on a lot of stuff. I don't have good enough. I'm always wondering what everyone's doing. I guess I'd have to give it a task, rebuild the tax system in the United States or something. I'd have to come up. I'd just have to be like, I'm just going to give you a massive task. Because I'm always like, I don't really have that many ideas that I personally... You're always trying to rewrite calendars. You're trying to rewrite the banking system. Maybe that's what we should do. Maybe calendars will finally be the thing that's figured out. But I think this is just another, it's just a fun example of nobody knows what's going on.

29:08Even Microsoft, if you will, one of the hyperscalers is mad at somebody. Because for all we know, maybe they, I don't know, maybe they actually were working on a program. You would think of the people that have problems at the scale. And Brian alluded to it. It's like Windows security is never fixed, right? You could spend$1 ,000 a day probably on look and find and fix all the various issues that have been reported. It's just a funny line, though, the fact that they got mad at the person. Now, maybe the person was just like, they were doing projects like I do. I was like, let me just sit down and spin up 10 different crazy ideas I have that maybe will never make money.

29:41And so maybe they're just like, that was your problem. We don't need 5 ,000 cat pictures. Stop using the image generation to do that. That's not useful to our company. So I don't know. And that would be cool if that person actually came out and said, here's what I was doing. And it was like really important research, or maybe it wasn't. Yeah. We're going to end with two. I'm going to leave the last one for the last because I think that's where it is. OpenAI came out with jalapeno, not the jalapeno that's on the banned list for making you sick, but they came out with their own hardware. Aaron, I know you like to cover sort of hardware stuff and benchmarks and stuff.

30:10Anybody, any thoughts on, should OpenAI be in the hardware business? Are they, have they moved far enough along that they should be like world-class at hardware? How does this fit in when they're trying to be world-class at software?

30:21Aaron: Just like I think every major company, and we've talked about it before, and yeah, NVIDIA has their five-layer cake, if you will. And everyone eventually, when they get to a certain amount, they want to control the hardware. And they want to control the hardware stack. And they want to do it on their chips. And I think it's a natural progression for them. Now, will they be good at it? It looks like the first chip is going to be better than NVIDIA, at least at what they want it to do and what they've designed it for. So, hey, good for them. They probably have the money to go through and hire the engineers that they need to go do things like this.

30:57Aaron: To what Brandon said earlier, though, I'm really curious who's going to make the chips, though, because TSMC is the major fab out there and NVIDIA has got them locked up. And then I think Apple takes whatever's left over. And I've always wondered, I'm like, OK, great, you designed a chip. Who the hell's going to make it at scale? Yeah, my best guess would be Broadcom. But I thought the interview when Jensen Wong was asked about it was like, because, you know, when someone starts out, he's very calm and he's just he started something like, we've been doing this a long time. It's just like the wise old man.

31:28It's like, I've seen a lot of people, just like you said, Eric. I've seen a lot of people come to me with a chip design and I've seen a lot of people leave. And so it was just funny. It was just like, okay. It was just, sure, bring it on. Call me when you actually have the chip. And to your point, call me when you have the chip at scale and it can be deployed and it's in a data center. And it was like, yeah, a lot of, I'm not, who knows, OpenAI, maybe they will be successful. It's hard to know, but there's a lot of work to do, right? There's a lot of work to do between I have a design of a chip and I have the chip working at a data center scale level.

32:02So that'll be the fun part to watch there. Yeah, it feels like two things. It does feel like, basically, look, I don't know if you guys really know how the game works. TSMC is the only game in town, and I have them locked up for the most part. I am surprised we haven't seen, or maybe this is going on, but we just haven't heard enough about it. It feels like in every other industry, once some of the really smart people have made generational wealth, which I'm sure tons of them have at NVIDIA, didn't go and say, hey, I've had this side project I've been wanting to work on for a while, but I couldn't get the buy -off from Jensen to do it.

32:34I'm going to go work at other company. And I think that probably just speaks to the thing of chip manufacturing is not like software. So you can't just run off and go spend something up in a year. It's probably four or five years of very concentrated work to even have something that's viable in that space. But I am surprised we haven't seen more ex-high-end engineers at NVIDIA be like, we went over and started this thing. And people will be like, oh my God, I can't believe that X, Y, and Z left the company. Like the way they do anytime people leave Google or somewhere like that. All right, last thing on the list.

33:07Brandon, I feel like this is a perfect crescendo before football season. I'm going to let you cook on this one. And Salesforce essentially has said to the market, nobody can understand Salesforce. We are going to introduce this thing called Cloudforce, which means that you now can just use Cloud to interact with Salesforce because nobody understood how to directly interact with Salesforce. Do you want to make a comment on this? Do you have any thoughts about interacting with Salesforce or just applying AI to the CRMs? No, I feel like life has gone full circle. We've now finally gotten back to where we all wanted to be, becoming online.

33:39Everyone has a Salesforce admin. The Salesforce admin is too busy and the Salesforce admin is really doing only what the VP of sales or the chief revenue officer wants. So nobody really knows what the reports mean. Nobody knows what dashboards have been created and no one can ever navigate to the dashboard that was maybe created for them or maybe not. So I think finally that Mark Bainoff has, if you will, admitted that and said, listen, why don't you just connect it to Claude and just ask questions? and that'll be the UI going forward, is the perfect, if you will, end to that entire piece of enterprise software.

34:12And it's super fun to watch the interview of Benioff and what the CEO of Anthropic, Dario, right? Because it's just like worlds colliding, right? It's like Dario, the PhD researcher, talking about how they use Salesforce. I was like, come on, Dario has never been in the Salesforce. He has never once logged into Salesforce to see what was going on. He's like, oh, it's important to us. And Benioff going on and on about how they need it and stuff like that. And I will say, unfortunately, all the Salesforce data, all the data is trapped in Salesforce. So maybe Claude is finally a way to get it out.

34:47And then for all of the marketing and the product people and performance marketers, maybe now all of us will actually be able to see what the hell is going on inside that Salesforce thing. So kudos to it. I look forward to using Claudeforce sometime soon in my professional career. Do you think we'll expect to see an article from Jammin' Ball basically saying, hey, that whole SaaSpocalypse thing, now made a mistake. Just put Claude in front of all of your SaaS apps. You will be good to go at this point. Don't worry about rewriting them. Too much hassle. Just put Claude in front of everything. I'm surprised he hasn't already written it because I feel like that's just become the new thing.

35:23The new UI is an API, right? Just put an API in front of it. Give it access to Claude or let Claude do it and just start asking questions. It is so much easier to do it. I know this is not really a sports podcast, but I'm going through all my fantasy drafts, like me and like everybody else. That's what everyone's doing. That's, I used to put all the players in there and have it tell you what to draft. You don't have to go to these weird websites anymore. You don't have to suffer through Yahoo. You just tell me who the best quarterback is and it works. It's perfect. That's what we need. Should we just, should we should start, see if we can get a trend going.

35:51Just every enterprise SaaS that you use, just shorten the name. So work day is now Claude day. Just start. We should, I think we should get people starting to say that. Like just within your meetings, be like, hey, I don't know how you do that with Slack. Slack's only a one word thing, but I think people should start putting Claude in front of all the things that they use every day to see if people will start adopting that as the terminology. My guess is that very slide with all those names exists in the Claude, or sorry, in the Anthropic Roadshow document created by the investment bankers who maybe they were forced to use Claude to create the slides.

36:23Because that's what I assume they're showing all the investors. It's just like, yep, we're going to have Cloudforce. We're going to have Cloud, all the ones you just said, right? They're going to have them all together. And the TAM of this is$20 trillion or something. And to be fair now, but it is true. Like when you get things connected to Cloud, you can do a lot more. So that's the, maybe that's the optimistic tone. It's like, it does make you personally much more productive. And many of them, like Cloudflare, never used it before this, before all these chatbots. Love Cloudflare all the time.

36:52Never actually used the UI. I just deploy all my free apps and stuff there. So it's one of many examples. So yes, I look forward to Claude taking over. Maybe that'll be the Anthropic Super Bowl commercial. It's just one guy sitting in front of a Claude terminal, you know, doing CLI stuff. The rest of the enterprise is running around crazy. And he's just, I just do more. I use Claude anymore. I like it. I like it. That'll be a good Super Bowl conversation with everyone who's like, what is this? So Anthropic, if you want to send us$10 million, you can have that idea. We'll call it even and we'll go from there.

37:23So, well, boys, not the biggest news month, but it's August. There's not a lot that goes on. All of Europe wasn't working at all. Obviously, they were off. Although we did, and I got to find it somewhere, and we'll acknowledge this. Somebody took the time to send us a very long email. I just got it this morning, basically saying, hey, all the stuff you guys talked about with the EU and watermarking was wrong. So we do need to review that. We'll give it a proper respect and dig through it. Somebody from the EU, which we appreciate, because a lot of times we just speak on the EU from far away, from across the pond.

37:52but otherwise any other last thoughts for august other than summer's over the kids are back in school and football's back you just summarized it all summer perfectly all right boys good good stuff we're gonna wrap it up there folks thank you all for listening thanks for telling a friend thanks for giving us feedback on the show and for the person who just sent us the thing about the eu stuff which i gotta go find in one of my folders thank you for taking the time to do that straighten us out getting us where we need to be so with that we'll wrap it up and we'll talk to you next week. Thanks for listening.

38:20Check us out at the enterpriseaishow.com for past shows, newsletters, and all things enterprise AI.