Skip to content

Innovator vs Operator

Episode 599 · November 20th, 2026 · 55 mins 25 secs

About this Episode

This week, we discuss Nvidia without Jensen, Microsoft Ignite's agents, and the week everything went down. Plus, a blameful post-mortem of the FAA router outage.

Watch the YouTube Live Recording of Episode 599

Runner-up Titles

  • SDT After hours
  • Maybe he's caught in some scandal
  • Awkward family reunion
  • Circular financing
  • Doesn't wear leather jackets
  • How AI-pilled is Microsoft?
  • Go where the money is
  • The ship's on fire, and then they're like, do we want the captain back?
  • I need twice as much money and it's everybody else's fault.
  • Don't say I need one extra engineer. Go big.
  • We should never create an MBR deck again.
  • Microsoft doesn't want you to use apps.
  • You don't even patch your servers, so why do you need this advanced attack?
  • You don't talk about the competition for your sponsors.
  • The strength is your weakness.
  • He's living the executive life

Rundown

Relevant to your Interests

Sponsors

  • Decodo gives AI agents access to clean, live data from any public website. Try it free at decodo.com/sdt, and use code SDT10 for 10% off any paid plan.

Conferences

SDT News & Community

Recommendations

Edited and produced with PodEditor, the native Mac podcast editor.

 

Transcript

Brandon

Well, Cote is out doing something important or not doing something important. We never really know. So I brought in Brian Grace Lee to chat along with Matt and I. So everybody knows Brian. He's over there at the Enterprise AI show. So welcome back, Brian, to the show. And I want to start with an important question I think both of you will have good answers for. You know, we do nothing here if we don't cover travel. So hypothetically, I want to throw this out to you. You're in Oxford, Mississippi. You have a presentation that's going to happen in Norman, Oklahoma, the following day. You have a flight scheduled to get there. And there are approximately 100,000 people that are waiting for your presentation. Your flight gets canceled. It's an approximately an eight-hour drive, something along those lines. The people that run the presentation said, hey, listen, we understand you're in a bad place. We could move back your slot, but it's really important you're here on Saturday. It's really important. We had 100,000 people. And if you need to come in a little late or you need a little extra time, we'll even get you a hotel. Get in late, you can sleep in.

Read Full Transcript Hide Full Transcript

Brandon

My question is, do you, and I'll start with you, Matt, do you actually... Get in the car, make the eight-hour drive, and make it happen on Saturday? Or do you tell them, no, I got to be in my best. Got to have my beauty sleep. Have to be ready to perform. And I need you to push me back to Sunday. What do you do there?

Matt

Well, I definitely need more beauty sleep. But also, the wrinkle to your story is there's like 150 of me.

Brandon

Yeah, well, there's more of you.

Matt

It's not just me hopping on a bus.

Brandon

But I'm going to say this. The people, and for people who don't know, what I'm referring to is Ole Miss, the football team, needed to go to Oklahoma. And I'll tell in a few minutes what actually happened. But they have the capability to get 150 people moved. You would have the capability to get yourself a rental car. Ole Miss has the resources to get 150. It's not a big deal to them. They could do it. They have buses. So my question still is, are you making the trip? Are you like, no, no, no. I got to be my best. These people, they're paying to see me at my best. I gotta have them go Sunday.

Matt

Um, I, I think, I think, uh, I think you keep it on, on to Saturday. Um, you, you, you put in the effort and, uh, you know, I, was it like a noon game or was it?

Brandon

It was an early game, but they, it was clear, it was made clear that they would put, I think they would have started the game as late as probably 10 PM central time.

Matt

So yeah, I would have gone with that.

Brandon

So would you would have driven, you would have done it?

Matt

I would have driven and asked for the delay.

Brandon

Okay. Brian, where do you come down this? Are you driving or are you waiting? You're pushing back?

Brian

Yeah. I mean, I'm probably driving. Like, I think about this. I mean, I do more stuff for work where it's like, we need you to fly 12 hours, do a 30-minute presentation, and then you can come home. Like, I'm used to doing this. I think the one thing I would ask for is some kind of a whore about points, you know, whether it's airline points or whatever. Like, I'm looking at the thing. I mean, it's straight across I-40. There's got to be some sort of Buc-ee's or something along. Like I need something that is some, some sort of points that I can use for something else, but yeah, I'm fine with that. I mean, The difference between my best and my four-hour sleep is not that different. So I'm fine with driving it. I drove eight hours this weekend to go see my kids, so it's not a big deal.

Brandon

You make it happen, right?

Brian

Yeah.

Brandon

Now, I would answer for Cote. I think under no circumstances would Cote drive. He would prefer, he wants his points, he wants his airline mileage. If you had told Cote, he had to drive, he'd be like, I can't do it. He's like, no way can I do it.

Brian

Does he even drive anymore?

Brandon

I would even go, if it was as far as anything outside of five hours, I think he would reject and be like, there's no way I can do it. I absolutely need to. Now he's not here to defend himself, Matt. So I'm going to let you, if you wish to, you could defend Cote and say he'll drive. But I am almost a hundred percent sure he would not drive under any circumstance.

Matt

No, he's got that lounge access. He's living the executive life. We're hit the bus, drive ourselves. that that's that's me.

Brandon

That's right so that's that's what separates us from the art one american in the netherlands is we would drive as long as he's not here yeah well i mean listen if he returns to the show i don't know where he is maybe maybe he's caught up in some scandal we don't know we don't know if he's going to come back well we'll see i mean i'll be back he's not caught up in anything he's not caught up in a scare oh he'll if he listens to this is really just ensuring he listens to the show and that he can come back and he can defend himself but i think if he's honest with himself.

He knows he wouldn't drive. So just for everyone that has no idea what we're talking about, the story is Ole Miss football team had to play the Oklahoma football team. They chose not to drive in the buses. It was a big, if you follow college football, it was funny because Oklahoma people were pissed, were very pissed. And in fact, Buc-ee's, the gas station we always talk about, the coach from Oklahoma said, even said to them, was like, it would be fun. Why didn't they all just load on buses? And they could have stopped at Buc-ee's It would have been a great time, Now that coach learned an important lesson that we would never make the mistake of software to find talk. You don't talk about the competition for your sponsor. So it turns out Oklahoma is sponsored by the Love's Truck Stop. So the following day at his press conference, he brings like a giant Love's container, fills it with water, takes a giant sip of water with the logo in there and just said that's his favorite truck stop. So there you go. In the end, maybe everyone lost. That's what the lesson there is. So, all right, well, we'll see. Tune in next week to see if Kote would have driven. I don't believe he would have.

Decodo Ad

This episode is sponsored by Dakota. AI agents are only as good as the data they can reach, but most of the web wasn't built for agents. Pages load with JavaScript, block automated requests, and come back as messy HTML that burns through tokens. Dakota handles that. It gives AI agents clean, live data from any public website without the complexity. Teams get three things. Agent-ready output is clean marked on our JSON, so models spend tokens on content, not markup. One API key for searching the web, mapping a site, crawling it, and fetching single pages. And it plugs straight into coding agents through Dakota's MCP server. And reliability without the upkeep because blocks, retries, and JavaScript rendering are handled automatically. Try it free with up to 10,000 requests a month at dakoto.com slash sdt. That's d-e-c-o-d-o dot com slash s-d-t. Code s-d-t-10 gets 10% off any paid plan. And thanks to Dakota for sponsoring our show.

Brandon

Now, in other news, we're a podcast and we cover technology. And in fact, Brian and I have been talking a little bit about this on the Enterprise AI Show. But Matt, you had the week off. You kind of missed out on the craziness of the Jensen resignation slash affair and things like this. So as we stand today, he's officially resigned. He's officially out. Turns out the company still makes incredible amounts of money. That was all released yesterday. But do you have any initial thoughts on on the jensen uh.

Matt

Resignation where where.

Brandon

Did the matt ray uh opinion come down on all this.

Matt

Uh i mean i think i think he made the the right choice to resign i think he's making the wrong choice to try to come back um i mean they had a code of conduct uh i think to me, The highlight for me, in a darkly sarcastic venue, is his internal post to the employees of NVIDIA announcing that he was leaving. Two people gave it the explosive emoji for celebration. I'm sure they picked the wrong one, but everything else was hearts and sad pandas.

Brandon

Was this on Slack?

Matt

Yeah, something like that. yeah okay and uh but but i think, there's a lot happening here i mean obviously you know the the whole you know, was she a spy was she you know intelligence you know was he compromised that way he was pushing nvidia towards things that uh, broke the law full stop you know um and then allegedly allegedly broke the law come on let's get the facts right there yeah we're we're all about allegedly here um but you know allegedly breaking the law they're starting to become rumblings of well was there a little bit of mark manipulation also happening were they you know this circular financing specter looms over everything um you know were they propping up uh their sales um, and then of course he also reportedly allegedly had affairs with two of his, you know, 560 underlings. So, you know, that... Those are like, you know, two, three strikes, you know, I don't care what sport you're playing, that's disqualifying. And...

For him to come back is probably a distraction because they're going to spend a lot of time getting grilled on this. And then you've got this blue wave of Democrats coming in who are looking for a fight. And there's rumblings that the Trump administration was covering this up. They knew about it. And they allowed NVIDIA to continue to make 20% of their sales channel through to China. So I don't see him coming back because, you know, it's like the abandoned ship, ship's on fire, and then they're like, oh, do we want the captain back? They're like, no, we're busy firefighting here.

Brandon

That is on fire with like tons of money raining down on it. That's the other part you have to remember.

Matt

Maybe you're putting out the fire. Yeah, their stock is about where I sold it.

Brandon

All right, Brian, I know we've talked about this a little bit already. What do you think? Like, is he going to, because I think early in the week, it was like, he seemed to be kind of mounting a comeback.

Brian

Yeah.

Brandon

And we talked a little bit about this on the Enterprise AI show. But like, we thought he would have a resurgence eventually. I don't think we thought it would be by like Monday. What do you think? Is it over? Is he not coming back?

Brian

I mean, it feels like it, like the body language, the tone from the new chairman of the board and the new CEO, like, I mean, they, that was like the first thing they said on the earnings call was like, you know, this, this guy made personal choices that, you know, don't align with NVIDIA. He's not here and he's not coming back. Like they made it, they made it pretty clear. Like we've, you know, we've, we've done our adult thing. We made the decision. I mean, the, the CFO went on, well, she's the new CEO. She went on Kramer. And i mean she just had that look on her face like i have put up with some shit for the last week and i am not i mean like i was just looking it up she's been there for 13 years so she's been there a long ass time, uh how much do you think she's worth in nvidia stock versus jensen who's apparently worth like 180 billion.

Brandon

I don't know a few billion i would.

Brian

Think no 740 million dollars what so she's working she's working for a guy that's worth 100x her and she's like Like, I have to put up with...

Matt

And she's been there that long.

Brian

Yeah, she's been there for a long time. She's like, I have to put up with your cheating scandals. I now have to deal with HR stuff. I'm going to have to deal with the federal government. I have to, you know, I have to write this, I have to write the keynote for this week that you were supposed to, like... She was just like, no, I'm not having any of these sort of things. So I don't know.

Matt

So we know how she's voting.

Brian

Yeah, we know how she's voting. I mean, I think, the one thing that'll be interesting to see is, and again this is this is the difference between and I know you guys have talked about this like the difference between like a software company and a hardware company like if this was a software company, Jensen would rally like you know the hundred best engineers and they would go build a clone of whatever the heck they did but you can't do that with hardware like hardware is such a long-term thing like, there's kind of nowhere for these guys to go I mean I guess they could go to AMD or something but like, that's the only thing that feels like is really up in the air is like how many smart people end up going and leaving the company or something like that.

Matt

If he goes to AMD, that makes for an awkward family reunion.

Brian

Yeah, exactly.

Brandon

Well, there was some reports out that one of the things that the new CEO announced is that like the bonuses and stock grants are like it's being moved up. I think it's like sort of like, hey, let's make sure everybody, I'm going to shoot everybody here with the money gun just in case anybody felt like they were feeling weird during Thanksgiving. Everybody's getting more money. Now, I guess the real question is like, as I said before on the show and others, is that I don't think it really impacts their earnings and their ability to make money in the near term at all. And obviously, the earnings yesterday, I think, show that their forecast is like yet another 100% growth. So it's just like, you know, they sold all this. All this has already been sold.

Brian

Right.

Brandon

But what I wanted to do, and maybe Brian, I'll start with you, is one thing that I think will change is the, let's call it the aggressive financing and investing. Someone called that circular financing. I know Matt likes to call it that. Because it seemed like that was mostly Jensen's thing. Like he sort of just was the one writing checks to like the neoclouds and to like all these things. So what do you think going forward? What do you think the posture of, are they going to unwind these things? Will they keep doing it? What do you think is going to happen?

Brian

I feel like that's the real big question because they are sort of the bank of AI at this point. But the trick to it is, I mean, he was definitely playing this game where it was like, I need to finance things. Well, I need to make successful things that aren't the three or four biggest hyperclouds or hyperscalers because I know those guys also have the ability to build their own chips, right? So he could only make them so successful. He had to go out and sort of essentially be his own VC for the rest of the ecosystem. I feel like they have to do that to a certain extent. And, you know, we're going to find out real quickly whether or not those guys, you know, have viable businesses or not to do it. But I think that's a big what if. And then the other big what if is, you know, it's not clear how much they were selling through this middleman out in Singapore into China and whether that gets cut off or it gets, you know, rerouted somewhere else or something like that. So, I mean, those are the two big sort of pillars of like the unknowns of their financing. because you're right, the money spigot is still coming in like crazy.

Brandon

Matt, neoclouts, they've been a recipient of a lot of this money. I feel like, at least on this show, you're the one that has talked about circular financing the most. So what do you think? Does it end? If you're a neoclout, is it one of these things where you have to go to your bank and be like, whoa, before you raise that interest rate, let's talk. Don't just hit me with an arbitrary bad credit rating. What's going to happen?

Matt

I think this is kind of going to be their sink or swim moment. I mean, we've already seen several of them have failed IPOs or they've been relegated to junk bond status. Their ability to borrow money is being greatly curtailed.

And so they're going to need to show that there's a viable business here and, you know, the finance part is the the part of ai that's always been squirreling um i mean, obviously you know sometimes the the output can be but people are getting their heads around that and, you know there's definitely utilization there it's just the finances don't make sense and if you're one of these edge players who you don't have advertising or you don't have you know a cloud compute you know other business to you know finance things for you, you're going to start running into some headwinds and, uh, i think i think there's going to be a reckoning because i think nvidia is probably going to be tied up and under a lot more scrutiny and you know people are going to be, tearing into this stuff and you know the stock market's already starting to reflect this a lot of a lot of things are starting to like slow down and um you know the the anthropic anthropic ipo is supposedly this week that'll be interesting yeah.

Brandon

Now what do you think about brian what do you think about the whole uh You mentioned it, the Southeast Asia thing. As best as I can tell, it sounds like it was working like this. It was sort of like a lot of chips were sold there. They seemed to be deployed, not quite in China, but very close to China. And then the Chinese data, I don't know, let's call that model, people creating models were then using them. And this seems, from what I can tell, was like a very gray area, That it wasn't necessarily legal, but it wasn't really what the U.S. Wanted. So, do you think that whole business just ends? What was the thing called? Super mega?

Brian

Oh, mega speed.

Brandon

If you're working at mega speed, they're just like, no, no. The jig is up, guys. You're not like, we're shutting this down.

Brian

Well, I mean, again, we'll probably never get all the details. I mean, it feels like they probably, you know, they did the thing where they were as close to the line as you could possibly be because they were selling, you know, they were selling to a company that wasn't technically, you know, physically in China. Some of them were at least going into data centers that weren't in China. You know, there was probably a certain amount that they just went, well, you know, we don't know what happened to them. They got on the wrong boat and maybe they ended up in China. You know and they you know their excuse is probably well you know we sold to the company and we you know we can't control what happens after that sort of thing which you know again you're getting into these gray areas I mean I think, I think the thing with that is, It feels like if that were to go away, and then people say that's upward like 20% of what they did, it feels like they could make up for that pretty quickly with the hyperscalers who have sort of said for the last couple of years.

We can't put enough money into AI. We're supply constrained. So in theory, those GPUs could go into hyperscaler data centers. Um you know the the trick to that is you know that accelerates jensen's you know, assuming those guys get them they put them in they make money with them that accelerates the revenue that then goes into those guys building alternative chips you know for doing it so i mean i think the legacy of jensen's ultimately going to be like, the guy knew he was playing this game that that every path he went down had you know knives pointing at him and he figured out how to navigate them pretty well but at some point you know he was going to get stabbed with something uh bad analogy sorry but, but you know now now that but he was very good at making these like really high high you know high yield and high risk bets, you know the i think the thing we hear from everybody is like who's going to make those big bets now because most of the people around him probably didn't have that level of you know, complete trust in himself or you know think he could get around things or whatever it might be.

Brandon

Right i think this is the classic the strength is your weakness you You know, he ultimately made one of the biggest bets in tech history. He invested in all this stuff when nobody really saw it and he won out. Now, of course, like I always like to say, you know, there's always some hindsight bias. There are many people in Silicon Valley that made bets that didn't work out. So we're only remembering the one person. Matt, since you're the, you know, the expert investor, I wanted to offer you this hypothetical. Okay. Putting aside the Jensen for a second, because there's a lot of talk of like, you know, that Jensen is like the key man. And without him, NVIDIA won't continue to grow. I think we've already talked about it. He can't come back for obvious reasons. We'll see if, I don't know, some weird thing happens. Anything can happen. But let's use like kind of a different company. If I were to offer you the following, I said to you, I'll let you have the Apple returns from Steve Jobs' tragic death, right? You can have that as door number one. Or door number two is Steve Jobs makes a monoculus recovery and he leads the company for these past 10 years when.

I just forgot the name of the CEO, the previous CEO of Apple, you know, when he's been leading it, right? And I said, so door number one, door number two, right? You know, it's like, you want Tim Cook's returns or do you want the hypothetical returns that Steve Jobs could have done? And I'll just lay it out. Maybe Steve Jobs would have foreseen AI and he would have been like, Guys, I just discovered Surrey sucks, but he would have figured that out 10 years ago. And maybe he's the one that unlocked whatever, the AI revolution and some, like in Apple's worth $20 trillion today, something enormous. Or... Or maybe, of course, maybe he, you know, he's like, he's all in on the vision, with the vision pro and it didn't work. So which would you go with? Like if I said door number A, Tim Cook returns, door B, the hypothetical Steve Jobs founder led, greatest technologist of all time, leading the company for 10 more years. Where would you put your money?

Matt

I like how you're giving me, you know, the power of life over death.

Brandon

I already took that. I did that part. I just want you to tell me where the money is.

Matt

I brought it back to life i mean it's it's the question of you know an investor versus innovator um i personally i think i would have preferred steve jobs to be pushing harder um, i mean apple is a well you know a well-oiled machine and you know they put out consistent products with incremental updates and the few things that they've done that kind of but just so you know.

Brandon

The company went from like several.

Matt

Hundred billion dollars.

Brandon

To like almost i mean that's pretty.

Matt

Impressive i said well oiled i mean like bathed in oil i mean okay there's no friction here they they they showed everybody like how to you know how to make sure everything was hitting on all cylinders and and just delivered consistently, but the converse is you know on the on the um privacy side of things you know there have been detractors who are like steve jobs never would allow them to do this much like, you know, advertisement and selling, you know, access to their customers. And, you know, he would make bolder attempts at new products. Apple is, almost like the definition of a fast follower um they've gotten to the point where you know you look at like their their product updates it's like you've seen this other places but they didn't do it well, you know and apple is always seen as like well they're going to take it and refine it and make it palatable to the rest of us you know everyone is still holding out for siri to be the breakthrough but.

Brandon

I'm no i'm not holding out.

Matt

Most of us are not holding out but that's the the goal here is people are like, you know this i'm not going to use copilot at home but you know once apple sorts this stuff out siri is going to be just great well let's just.

Brandon

I mean we're going to bring brian here so the biggest bets that i would say that they mean vision pro i think that was i think we can all agree not good right uh and then more recently iphone duo which i hey let's i do i got a hands-on review maybe we'll do that later i actually i actually used one for a second.

Matt

So those.

Brandon

Are probably the biggest new thing so i guess what you're really betting on here is that steve jobs would have like, made those better or come up with something completely new? Because that's the only way the valuation... So, Brian, what would you take? You're taking the Tim Cook known returns? You're taking the Steve Jobs unknown returns?

Brian

Yeah. I mean, I think this probably boils down to pretty simply, like if you're talking to somebody on the business side of things, they would take the operator's returns because they were... If you take the technologist perspective and you go, hey, where would you want to be in five years or something you know you'd be like i i need somebody who's an innovator and especially like right now so i mean.

On the jensen side of things i mean this this is where it's going to get really interesting because you know two or three years ago people were like hey gpus are gonna they're gonna you know they're gonna wear out in two or three years you're gonna you know depreciate in two or three years you're gonna need new ones, i mean they're still talking about people running these gpus like, six seven years in still really efficient you know maybe not as efficient but like still making money off them you know it took them, you know they announced this what's the what's the latest one is it vera rubin or whatever the later i mean they announced that one back in like the beginning of the year they just now said they're just now shipping it, so i mean that's almost a year between between that and that they still made all their money, i mean in theory if they have the next you know two or three iterations of the of the product already taped out and sort of planned which i kind of would assume they they probably do even though they say they put something like they're, as long as they operate really i mean their whole business basically is going to is going to be dependent on two things one is, um you know does demand stay high whether that's through anthropic or open models or whatever and, you know do they still keep getting the the bulk of the tsmc you know you know time in the in the manufacturing line you know there's no like no war in china or something and if those are the cases um, then yeah, like a Tim Cook-type person, you know, maybe it's somebody who's already in NVIDIA, like they probably can run that company and make a lot, a lot of money for the next five or six years.

You know, I mean, like there's plenty of precedents to sort of show that that's possible.

Matt

And Chris has been their, you know, CFO for, you know, 10, would you say 13 years? I mean, that's Cook's background. You know, maybe she's the Tim Cook of NVIDIA.

Brandon

Yeah, well, I think that's definitely her pitch, right? And so, and- You better hope. It would be interesting. If I said, I'll give you this, like she's going to be named the permanent CEO or I'll give you that or the field. Which are you taking? Who do you think? Start with you, Matt.

Matt

I think it's going to be her for the next two years. Just because.

Brandon

Brian, what do you think? Someone else going to take it over or is it her?

Brian

I think it's her. Yeah, I think for at least a couple of years.

Brandon

I think so too. I think it's going to be boring for podcasting because we could have done brackets and everything and we could have had a whole bunch of people and they're just going to pick the obvious person.

Matt

And she doesn't wear leather jackets.

Brandon

Yeah, she looks very calm, normal. She's not going to say anything too crazy. I mean, you know what? She's killing the podcast angle for this, for us, for the next few years.

Brian

I mean, the real bet is, you know, how long is it going to take before, you know, Jensen has his next coming out party. He goes, hey, I went, you know, I went off on a beach and I thought about life and I figured out the next thing that's going to change the world.

Brandon

Well, that's why, like.

Brian

What is that thing going to be and when's it going to be?

Brandon

Yeah, and I think that's kind of the interesting, fun thing that we're going to be able to watch here. Because we kind of just laid it out. NVIDIA will become a little bit more conservative. If Jensen does provide this unique insight coupled with risk-taking, he's still out there. He can go do it. So if he feels like NVIDIA is unwilling to now do the thing that he would do, there's no one more well-positioned to do it. If he said, tomorrow, I'm starting a company, I mean, how, I mean... It's almost limitless the amount of money he could raise, right? That some of that people would fund and want to be part of it. So it doesn't mean they'll succeed. So that's why I think it's going to be interesting. It's like, well, we're about to see this really play out once and for all. Is it, was it the, was it the person or was it the ultimately the time of the company? So, so we shall see.

Now in other news, not nearly as salacious or exciting. There was Microsoft Ignite. That was 2026. Did anyone watch the keynote other than me? Matt, I know you didn't watch the keynote. You were asleep.

Matt

I had to read the articles because I wasn't aware there was a keynote.

Brandon

Well, as always, you know, they have the, it's like, I feel like I see these keynotes mostly through the fonts and the colors. It's like, you know, it's like, oh, the Microsoft colors up there and on the co-pilot sprinkled along everywhere. So I would say I did watch the keynote. I thought it was just okay, like normal. The things that I kind of came away with was like, one, they're putting, co-pilot everywhere. Two, they are basically taking... Are either one of you familiar with the IQ, all the IQs inside of Microsoft?

Brian

No.

Brandon

It's essentially kind of like all of their connectors to all of their various systems. So they basically took all of the various like work IQ, fabric IQ, foundry IQ, web IQ. They took all of these and they put them in like one layer. So that's kind of the layer of connections between agents and things like that. I can tell. Are you moved by that, Matt? Is that exciting to you in any way?

Matt

I'm just thinking about like Conway's law and the org chart of, you know, how they're reorganizing the engineering to be like, okay, we're all one product now. You know, I'm not across it.

Brandon

That's kind of my take. So, you know, I've been trying to do a lot of thinking about like how to be use AI more, right? And I just actually recently did an interview with one of your old friends, Adam Jacob. Matt, it's over on the Enterprise AI show. so everyone should listen to it. And he's got a real strong take about how you become AI-pilled and how to use it. And so as I was watching this, I was thinking about that. I was like, how AI-pilled is Microsoft? So one of the demonstrations they did, and I actually took some screenshots, you can see them at the bottom of the show notes, right?

Is they're going to show us how to use their new co-pilot workspace, okay? And the thing that they decided to do, the project they created is they're going to create a monthly business review deck, okay? So this is what they're creating. They're creating a deck. So let me stop right there. I was like, and I want you guys to push back. Tell me if I'm wrong, but I was thinking like channeling a little bit of like Anthropic, how they do things. I was thinking about Adam Jacob and it's like, no, you don't create like a monthly review deck, right? You create basically like an agent that would just collect all this information and would just be able to highlight this information to you all the time. And if you wanted to, you could have an agent monitoring those KPIs, servicing problems that you should either address or even making suggestions. Sort of like going from like, almost like, I'll call it like DevOps business. It's not like, you know, the operations team gets together and like reviews all the outages on like, you know, the first of every month, right? It's like everything's just happening all the time and things are getting addressed.

And I was thinking to myself, is the whole idea that the demo that they.

Chose to show us, basically a team of people putting together a deck with a little bit of agents, is that completely missing the AI-pilled use of agents? I don't know, Brian, do you have any thoughts on that?

Brian

Yeah, well, I mean, it's, you know, it's the innovator's dilemma, right? Like, it becomes really hard to go, here's our core product that we want every one of you to use that we get a seat license for that doesn't involve you, right? So everything has to sort of, you know, I mean, you come up with the demo and you're like, okay, what do people typically do? Like, how are we making it better? Yeah it's it's i heard it i heard an interesting analogy because apparently one of the things they did was they said hey we're we're pulling all of the office stuff into copilot right like it's they're now just like demonstrate connectors or features of copilot um, and it was interesting because well it's interesting for two reasons one anthropic you know did the thing where they put out their version of of the office suite this week or last week uh you know still to be determined if if people want to use that thing. But I forget who it was. I listened to somebody. And they sort of said, it's interesting to see what's happened with agents because now you have, Google doesn't want you to do searches. Facebook doesn't want you to talk to people anymore. Microsoft doesn't want you to use apps.

Like everything has just become some capability of the agent. And so it's like in Microsoft's case, you have this problem where like if you brought in like new kids, into your workplace or something, Like you could teach them that super AI centric workflow, but if you've got somebody who's like 35 or, you know, 40 or whatever, and they've been using Microsoft office for 25 years, like, I don't know how they get on, how they get AI pilled. Like I don't, like, I would love, I would love a case study of like, okay, this is the only way you can do it. Like, how long do you feel completely useless at your job until that thing? And maybe it's a, maybe it's a week, maybe it's six months. I.

Think that's the most interesting thing of the dilemma they're sort of dealing with is like, how do I do this without people being in the middle of it? Because people are in the middle of their software right now. That's their whole business model.

Brandon

If I went in there and I channeled some Adam Jacob here, I was like, I'm going to build the machine that builds the machine. And the way I'm going to do this is like, listen, I'm going to go in there and be like, listen, we should never create an NBR deck again. What we should create is an agent that can answer all of our questions. And if we have a question that it cannot answer, we should spend our time making sure that the agent can answer the question, make sure that it has all the IQ connectors, it can get all the data. And that way, if we want to meet as a team, We just walk in there together and then we turn on the agent. And if we have a question, we're like, hey, put up last month's sales. Or we say, hey, like, show me customer retention. And I said to Lizzie, like, we are outcome focused and we are not spending any time doing meta work. We're not creating the fonts on the PowerPoints to make them look good and cutting and pasting. Because if you look at, literally, if you look at their demo, they're cutting and pasting charts, right, into a PowerPoint, which, Frank, I know Brian has done this. I know you've done this. I've done this.

And so if I went in, I was thinking to myself, like trying to pitch this, and I think if I pitched this to like a Fortune 500 company, I feel like they would immediately kick me out. They'd be like, you do not understand how we do business. But I think that is the AI-pilled way of doing things. What do you think, Matt? What do you see in the, quote, real world out there?

Matt

I reject the idea that having a business review artifact is not a good thing. To me, what you want to have is you've got your AI watching all these numbers, watching all those metrics, you know, running actions for you, but you've got your AI. At some point, you need to have a get together with fellow humans and talk about the progress that's been made. And so that business review, I don't want us to sit in a room and ask AI questions. I want to say, here's what we're going to see. Here's the presentation. I don't want to spend any time editing it. I agree with that completely. It's like, dude, AI, spit out the deck for me. And these are the charts we need to see. These are the things we're going to talk about. We have that monthly presentation or whatever. And then they're like, you know, hey, how come it doesn't have this next month? It'll be in the deck. Right. And so you keep that as, you know, we are, you know, AI is going to handle the making the presentations. Goodbye, Office. Right. Yeah. So much Microsoft. But, you know, because it's like, I'm not going to do it in PowerPoint. I'm going to do it in HTML. And it's just going to be a click through deck.

I don't need Office anymore in this scenario. Right. So what I'm going to be maintaining internally is, here are my notes for AI. Here's my feedback. Here's things that we have going on. I'm going to have my regular conversations, interactions with it. But once a month, whatever our business for you cadence is, we're going to sit down and talk about what we're seeing. And that's not an office thing anymore.

Brandon

So I agree with you. I think the idea, the meeting and talking about their priorities, I'm going to say no one disagrees with that. Like that is the important part of it, right? And I think, but mostly how these things play out, having been a part of far too many, right? It's like you create the deck, right? And then you get in there and then someone's eventually going to like raise a question and be like, you know, what's going on in Amiya? I wish we had a drill down of this thing. So like my thought is like the whole deck is live. It's not a deck, it's a live. Like all the stuff you want to see is live. And then if someone does say two things I want to like take out of these reviews, one, if someone does say, like, I want to like drill down to this thing, it's like you could then ask, you could literally use the prompt, be like, show me this information. And the other thing that I think people I would like to do is when someone likes to say, it's like, you know, I don't think that metric or that data is right, where you could then have the agency, like, tell me exactly how you got the information and tell me how it's defined.

And if someone doesn't like it, right there in real time, they can be like, yeah, filter out all of the personal email addresses because I want to see something along those lines. Because that's where I think these meetings, that's where they lose their focus. And that's where I think going this AI native first, it's like, no, we're just in there reviewing data that we've seen before, ability to ask questions, and we're not getting sidetracked on what we're actually going. I don't know, Brian, is that too pie in the sky? Is that, you do a lot of these business reviews. Are your executives going to accept that?

Brian

Well, I mean, I think the thing that you're kind of pulling out and that I got, out of Adam's, you know, machines to build the machine thing is, like, there's no reason for there to be, like, point in time anything, right? There should no longer be the thing where it's like, well, the meeting's on Tuesday and, you know, by Monday night somebody will send around the thing. Like, you know, there should be no more of this, like I'm surprised when I walk into the meeting because I didn't get a chance to see this stuff ahead of time. That should all be kind of real time. You know, you ought to have, you know, stuff like, you know, like, let's say you create content and you're like, well, this has got a bunch of like internal data in it, but then we also want to share it to partners. So somebody's got to go through and scrub it. And it's like, none of that stuff needs to happen by a human anymore. That should just be a thing by the, by the agent or the LLM.

When you're done with the meeting if it's got to go to your partners it just automatically scrubs out the stuff or it does whatever and it posts it out publicly like, I think ultimately what Matt's getting at is like, you can envision a world, however you do your processes, whatever, but like Office isn't part of this thing anymore. You can very easily be like.

I've let the tool dictate how we can operate our business, and I got to stop doing that. And I think that's at the heart of the problem that Microsoft has is you can easily envision a world where you're like, yeah, I might interact with Microsoft in some way, but it's not around these static tools and, things that individuals can only deal with and all that. There's very easy to get away from that as the way you do business.

Brandon

Man you look you have some thoughts there.

Matt

Well i've i've you know lived this experience where it's like, uh yeah we don't use excel anymore we just have like you know a database somewhere that you know we run some salesforce queries today salesforce right but in the future it's like you know oh now we just keep those metrics you know they're, in a database somewhere and you know So maybe the future is, just MS SQL?

Brandon

Well, they did have, that was one of the, no, the future, I guess, is the IQs. It's all the new IQ stuff because they did announce a new Salesforce IQ, which I assume is Salesforce data, which is, I have to say, if you're Salesforce, you do get mentioned in all the presentations. Everyone wants to pull your data because of course you have the sales data. Okay, the only other things out of it, of course, there's a huge long presentation. So I know everyone did a lot of hard work at Microsoft, But I say the only two other things I wrote down that I thought were interesting. One,

they're building their own models and they have all the models inside of their foundry. My question is, Brian, is there any reason for Microsoft to be building models at this point? Don't we have enough models? Is anyone really using these things?

Brian

I think, I don't know. I mean, I think what we're ultimately going to see is I think we're going to see the hyperscalers who have a ton of compute. They've been the ones who figured out how to build data centers. They're just going to go, yeah, I've got, you know, I've got an anthropic OEM. I've got an open AI OEM. Those are fine. I've got, you know, four or five of the leading open models. And I think they will keep building one model because I think there's a certain amount of like, you know, if something goes wrong, I don't want to be completely stuck because, I mean, you know, you look at what these guys are doing. They're basically making the model the center of their world. Hopefully, they'll figure out a way to make a harness that's sort of like a universal harness that can go across stuff. But I think you still got to have some amount of like, I've got some smart people that, you know, if push comes to shove, if we have to do something unique with models, like we still know how to do that. Does it have to be Microsoft specific? I don't know. Maybe they should just be doing it on top of some open model and just don't tell people what it is.

I mean, I think the key to it is like You keep enough of them around so that you've got that skill set so you don't lose it, but you put all your energy into harness engineering so you can hide all that stuff.

Brandon

Yeah, well, the models they create, I guess they have Code2 and Code2 Flash and they're all part of GitHub Copilot. And I guess the only reason, I mean, I do think if you're forced in a Microsoft environment and they make you use them, I guess you'd use them. Otherwise, you're using cloud code or you're using Codex, right?

Matt

You shouldn't care. Yeah, you shouldn't care. And Microsoft doesn't want to give money to open AI and traffic. That's, you know, full-on stop. Why they're not using open models? Meh, whatever.

Brandon

Exactly. Well, and then Matt, I don't know if you had any thoughts on this. They had some security stuff. They've got the red, blue, and green agents. So basically red teaming, blue teaming, all that kind of stuff. All part of the Microsoft Perception product. That was another name. Just too many names. That's one thing. I just, I don't know. I hope the agents can figure that out. All these companies with too many names. But do you have any thoughts of it? I just kind of read this. It's like, sounds good. I mean, I'm not saying it's bad or good. It's just sort of like, okay, check the box. anything we should know about it?

Matt

Why is it coming from Microsoft and not GitHub? That's kind of my only branding question, but yeah, no, it seems reasonable. That way, that's more security dollars they get to hold on to before somebody else comes for them.

Brian

Hey, do you think there's any chance, you know, like we talked about NVIDIA and they're going to go through whatever changes they're going to go through. Like if you're Microsoft, do you make a play for the Nemotron team? Right, like NVIDIA was building this sort of, you know, big, pretty good open model, you know, to, you know, at least have some reference for people using there. You know, do you make a play for that team if you're Microsoft, because you're the only one who doesn't really have your own model. And then you've maybe got the best of both worlds. You've got maybe the best open model, but then you've also got this thing you can customize. Or do you think NVIDIA doubles down and says, no, no, we need to keep those skills in house, even though it's really expensive?

Matt

I i think it's a good time to be on the nematron team yeah because to me it's like you might as well be a mercenary at this point yeah yeah you're gonna make a lot of money go where the money is yeah.

Brandon

I think nvidia don't you think nvidia though whoa whoa whoa before.

Matt

You go where the money is.

Brandon

Let me uh let me walk let me go over here and get another briefcase full.

Matt

Of stock.

Brandon

Options and money for you.

Matt

Yeah yeah yeah.

Brandon

All right. Well, listen, now that we've covered the really important things, the final thing I wanted to cover this week was the end of the world. OK, so I don't know what everybody is. Everyone is everyone into full prepper mode. So this past week, the following things happen. The United States FAA shut down the entire, airspace over what is being depicted as a router upgrade somewhere in Arkansas. So that's what it took to take down all of the AirFlight there. At some point this week, I think, in fact, Brian, we were maybe talking, I tried to log into my Chase account. It wouldn't let me log in. I was like, oh, I guess, you know, I thought it was me. I actually called the number because it was like, and they were like, the next available person is like two hours of it. I was like, oh, what's going on here? So it looks like somehow Chase was down completely. I don't know exactly why. As well as somehow it affected several other banking applications. Microsoft, we were just saying how no one's really going to use Office. Well, actually, this week, we all got a lesson. And I should have noted that they did, so it did come out.

The first thing he did was apologize and did the old like, and this was a good lesson for us about how important it is to stay diligent, blah, blah, blah. That went down. And then I think that was pretty much it. So basically, we've had just nothing but...

Brian

Oh, no, no, no, no, no. The NSA got hacked. Their secret net went down. Apple went down.

Brandon

Apple went down. Yep, I tried to redeem a gift card this week and Apple couldn't do that. And then you're right. the most secure communication systems that we have in the United States, the NSA, that also went down. So I don't even know if I have any idea, like, what's going on here? The best thing I've seen that someone said, and I don't even think it's true, but it's like, what else can you say? Someone said the amount of patches that are just coming out, and therefore everyone having to touch their systems more, is actually making everything brittle. Do you buy any of that, Matt? Is that why all these technology problems are happening in the same week?

Matt

I do buy that. Having seen what a tire fire it is at so many enterprises, having to do a constant... Drill of, oh, we got to patch everything, get everything out. It's definitely introducing brittleness on top of the persistent threats that are happening. I mean, I think we talked about this previously. It's like, I fully expect this to be the MO going forward is we're going to just see outages happening again and again and again, because things are brittle and we've introduced a lot of friction.

Brian

Yeah what do you think brian.

Brandon

Do you think uh.

Brian

Beyond just.

Brandon

Patches what what do you think is happening out there.

Brian

I mean i i think that's some of it i i the other one i'm curious about and since matt's on you know he's he's the the security guru what's this other tool this this new tool with an a that the fbi was this like pen testing tool um, that people are talking about i forget what it's called it's like and and x aren't our necks or something like that?

Matt

Yeah, I didn't see that one.

Brian

Okay.

Matt

But there's pretty much general phobia of, you know, ever since Mithos, the whole industry has gone haywire, where people are like, you know, oh, I need to do pen testing across my stuff. I was like, you don't even patch your servers. So why do you need this advanced attack? There's a lot of basic hygiene that's just not happening. And, you know, various security vendors have put out like, oh, here's all the steps that you need to do. And like the first one is stay on top of your patchy.

Brian

Yeah. Well, and just to put a little bit like that, the mythos stuff in context, like I deal with it a little bit. Like it's not unusual for us to get these mythos scans from companies, like, you know, decent sized companies. And they'll send us 80,000. They'll be like, hey, we ran it against our servers. Here's 80,000 vulnerabilities we found. You guys filter them, triage them, figure out which ones are great. And by the way, give me patches for all of them by Monday. And you're like, what would you do?

Matt

This out. You already had the patches.

Brian

Right. And what would you do with them if you got them all?

Matt

Exactly.

Brian

So, yeah. Most customers will tell you it takes like, you know, 40 to 60 days to get patches into production. So, yeah.

Brandon

Well, the most important thing is here, we're nothing if we don't like to watch, a blameless post-mortem. And by that, I mean a blameful post-mortem. And what we got to witness this week was absolutely fantastic. Putting aside the politics, let's just put that aside for a moment. It was great to watch, Is it director of the FAA? Secretary of Transportation.

Matt

Yeah.

Brandon

Sitting on stage with the Secretary of Homeland Security.

Matt

Uh-huh.

Brandon

Like side by side, right? And of course, they go on and first the Secretary of, whatever, the Transportation comes up and basically goes up and says, it's not his fault. It was the previous regime. And he almost demands outright $50 billion to fix this problem, right? And that made me think, I got to learn how to upgrade routers. I was like, what? Like, how much? Like, what's going on there? I mean, it's like, it can't be that hard, right? But I mean, I guess maybe it is. Switching over. Sorry. So he is just full bore. It's not his fault. He wasn't given the money, although he was given money, but he didn't get enough money. And, you know, and then so he steps to the right-hand side. And then they bring up Director of Homeland Security and says, this had nothing to do with any type of terrorist attack or security. thing, it was 100% the fault of the Secretary of Transportation. And you see him side by side. He's in a zip-up, a quarter zip, and the other guy's in a suit. And you can literally see his face. He kind of looks down. You can also see a little vein in his head start to pulsate. And I was like, this, my friends, this is reality. This is what a good post-mortem looks like.

And then, of course, because they're just taking questions, and they're kind of like, I don't know, strategically basically putting blame on each person as it goes along. And then finally, just to bring it back to where we started, it turned out that, I guess on the following day, the Saturday, that the director of Homeland Security, he had tickets to the LSU-Texas game. Who among us would not have wanted to be at the game? I would have been there.

Matt

Did he take the bus?

Brandon

He did not take the bus. He took his private plane, not his private plane, a government private plane to go sit in the box and watch the LSU Tigers lose to the Texas Longhorns on the last play. So I just thought that was like, you know, we can all learn from like a good example of how to blame everyone. And of course, this week, he's at least one I've seen so far, like he's being asked millions of times about why he thought it was okay to fly and not, you know, be dealing with the emergency. So I don't know. No question there. Any thoughts, Brian, on hand to handle the blameless post-mortem in the federal government? Should we go in there with a little agile talk and a little bit of like hugs? What should we do?

Brian

No, I mean, I think that the ultimate lesson to take from this is that the guy who's the transportation secretary has had nothing but failures on his hands. I mean, he had, you know, planes crashing in Washington, D.C. He's had this giant outage. No, the thing to take away from this is no matter how bad it is, no matter how bad the entire country hates you, you need to have the confidence to go in there and go, I know I asked for $30 billion to fix this, but I really need $50 billion. And don't crack a smile. Just be like, this is the way to fix it. I need twice as much money, and it's everybody else's fault. You should do that with every single problem you have. Don't say, like, I need one extra engineer. Go big. Have that kind of confidence.

Matt

We need to acquire a company.

Brian

That's right.

Brandon

I think that's what everyone needs to learn. They need to update their decks or just be ready to say, listen, it's not like I'm asking for $50 billion to fix this problem. I just need a couple extra tokens.

Brian

This is barely a one-router level problem. It's just not $50 billion.

Brandon

Come on. Look how efficient I am.

Matt

Yeah.

Brandon

Well, I really enjoyed it. I'm glad everyone's back flying. It's going to be holiday season here in the United States soon. So we all need to get through this. We need to like a couple things if I could just request. Like, one, Jensen, just have Thanksgiving quietly. Don't do anything. Don't do anything crazy. We don't want to be podcasting while we have to have turkey. And two, let's just keep the planes going. Let's just keep everyone flying.

Brian

And the banks open.

Brandon

Yeah, the banks. Yeah, most of the time. Thursday, you can close the banks if you want. But for most of the week, can we get our money? Can we get where we want to go? All right. Well, listen, a couple conferences this week that we'll promote. Really just one, I guess, Build Stuff. It's December 2nd through the 4th. It's in a place, I can't say it, but it's in Lithuania somewhere. I know Kote's speaking. And then all the other conferences, they're not till the new year. So you know what? I'd say take the rest of the year off. No more conferences. Also, I forgot to mention, if you want a sticker, all you have to do is email me at stickers, or email your postal address, rather, at stickers at softwaredefinedtalk.com. I'd be happy to send you a sticker anywhere in the world. And we'll move on here. Matt, I'm sure you have a fantastic recommendation for us this week. What is it?

Matt

Um, my recommendation is, uh, a comedy show. I guess it's on Hulu in America. Um, but, uh, it's called adults. Um, as, as my wife pointed out, uh, these are the dumbest adults you've ever seen. So if, if you like, um, if you like, I don't know, sunny in Philadelphia to a lesser extent, Seinfeld, you know, people who are kind of not good people, but, uh, it's, it's had a fair number of laugh out loud moments so uh, it's pretty crude but i've enjoyed it.

Brandon

Okay adults what was it on again what.

Matt

Did you say netflix probably oh hello all right yeah all right brian.

Brandon

I don't know i don't know did you do your homework.

Brian

Do you yeah no i was looking forward to this i actually coincidentally i'm going to stick with the comedy theme uh there is a, there's a really good show on uh on netflix um it's about sctv which was like the the canadian version of uh of SNL back in the day. It's really good because it's one of those shows where, I mean, it's all the history of it, but you look at who the original cast was on that, and it was like John Candy, Eugene Levy, Catherine O'Hare, Martin Short. It was like all these people, and it's just one of those ones where you're like, none of these people were famous at the time. They all just happened to be in the right place in Toronto at the right time. And if you go back and watch it, it's better than the original Saturday Night Lives that, you know, even were out at that time. So it's good if you're into sort of historical comedy stuff. It's on Netflix right now.

Brandon

All right. Fantastic. All right. I'll give three this weekend. The one you just reminded me of, because it's Thanksgiving, I think a couple of Thanksgiving recommendations. One is, there's a Saturday Night Live skit that's going around called Sister. I don't know if anyone's watched it. Everyone should watch it in preparation for Thanksgiving because it has gone so far in my extended circle of people that normally don't even follow the stuff where people are sending it to me and there's a lot of commentary on it. So just watch it. I think it's triggered. I think somehow SNL has made something funny, but they have triggered a family discussion that I think will be happening at many Thanksgiving tables coming up. The other thing I'll say you should check out is Mobland Season 2. I'm really enjoying that on Paramount. It's good. If you're not into comedy, watch people kill each other and mob. It's really fun. I like that. And then final little tip is, I had done this on one airline, but not all the airlines. Make sure to go into your airline and sign up for the TSA pre-tech touchless. You have to opt in. I find you have to opt in on every airline.

So if you're not like Cote and you're not airline loyal like me and you just fly the airline that's cheapest, like you may forget. You're like, how do I not have this? I was like, oh, I had to go in there and do that. So get the touchless. That way you can get in and out of your TSA line without any issues. So with that, I want to thank Brian for being on the show, replacing Cote, Matt for making it back. And with that, we will talk to everyone next time.

Matt

Bye. And I've got another scheduled outage next week. So yay.

Download Transcript