We Discuss
The Jensen Huang Scandal Is Bigger Than You Think
Tech moves fast. So does work. Digital Disruption NOW helps you keep up and make sense of what matters.
What happens when Big Tech depends too much on one person, and we depend too much on AI? This week we’re covering the Jensen Huang controversy, the backlash against AI data centers, and how AI coding tools could weaken critical thinking.
00;00;02;06 - 00;00;14;14
Jeremy Roberts
Jensen Huang has had an extramarital affair with an individual who works at a customer of theirs based out of Singapore, who the US government thinks is probably working as a spy for the Chinese.
00;00;14;18 - 00;00;19;22
Geoff Nielson
And not not just like any customer. Like this is a multi-billion dollar for.
00;00;19;25 - 00;00;24;09
Jeremy Roberts
Partner. Jensen Huang is arguably the most important executive in the world today.
00;00;24;16 - 00;00;31;28
Geoff Nielson
They're like the central bank of AI.
00;00;32;00 - 00;00;33;18
Geoff Nielson
What about your call of duty or battlefield? Oh.
00;00;33;19 - 00;00;48;11
Jeremy Roberts
A battlefield for sure, because I like being part of a team. I can't let the whole game just rest on my shoulders because, like, I'm not that good. Call of Duty or Super Soldier Battlefield. You're just part of the team. Sometimes you're in a tank and it just gets blown up and you go, I guess that's my career.
00;00;48;17 - 00;00;56;08
Jeremy Roberts
It's just ending. I think it's probably in many ways much more realistic in terms of, you know, combined arms warfare. Right?
00;00;56;10 - 00;01;07;20
Geoff Nielson
And I like it. I guess we should jump into it. So we've got a few stories to talk about today. The first one is Jensen Wong and what this guy has been up to. And the.
00;01;07;20 - 00;01;08;04
Jeremy Roberts
Spy who.
00;01;08;04 - 00;01;25;09
Geoff Nielson
Loved Me, what you're calling affectionately the spy who Loved me. We're going to talk about data centers. And, you know, if they were actually the biggest loser of the midterm elections. And then we're going to talk about a new anthropic report warning about cognitive offloading and the risks of that.
00;01;25;10 - 00;01;30;01
Jeremy Roberts
An engineer going public saying, I don't think as much as I used to, and maybe that's a problem.
00;01;30;04 - 00;01;35;18
Geoff Nielson
I'm shocked. So maybe with that, you do want to kick us off with the Jensen Wong story?
00;01;35;20 - 00;01;50;01
Jeremy Roberts
Yeah, this is a really compelling one. You know, we weren't sure. Just peek behind the curtain for our for our viewers. And our listeners weren't sure if this was appropriate for digital disruption now, but it really took the media ecosystem by storm. This is like by far the biggest story in tech. And I think for really good reason.
00;01;50;02 - 00;02;12;25
Jeremy Roberts
So Nvidia today is, I think, the second most valuable public company in the world, after Apple, at least in part because their stock slid on news that their CEO and founder, who's been in since 93, Jensen Huang, has had an extramarital affair with an individual who works at a customer of theirs based out of Singapore, who the US government thinks is probably working as a spy for the Chinese.
00;02;13;00 - 00;02;34;14
Geoff Nielson
And not not just like any customer. Like this is a multibillion dollar for prestige partner. Yeah, Mega speed. Is that what they're. Is that what they're called? I think so I get that right. Yep. In any case. So, yeah, she's the CEO of it. And if you look at what this customer does, they buy billions of dollars worth of chips and video.
00;02;34;15 - 00;02;56;13
Geoff Nielson
They put them in data centers, and they rent out that capacity largely to Chinese firms, which is allowed, right? They're allowed to do that if they're selling the chips to Chinese firms. That's not allowed. Right. That's that's bad. So it's already kind of a pretty like, juicy, kind of like right on the fault line of a lot of geopolitical tension.
00;02;56;14 - 00;02;57;08
Geoff Nielson
Right? Absolutely.
00;02;57;08 - 00;03;08;05
Jeremy Roberts
Well, Jensen Huang is arguably the most important executive in the world today in video, being the supplier for basically the entire AI building. I think I saw a number that said like 90 plus percent of AI.
00;03;08;09 - 00;03;13;08
Geoff Nielson
They're like the central bank of AI. I've heard them described, okay, I can see their funding, everybody.
00;03;13;09 - 00;03;34;02
Jeremy Roberts
So, you know, this is kind of a big deal. And it's extra because while Jensen Huang doesn't actually have like voting control over Nvidia, he is highly visible. He owns, I think about 3.5% of the company making him, you know, a multi multi multi-billionaire. So super rich powerful guy. And he sits at the nexus of a really weird organizational structure.
00;03;34;04 - 00;03;38;08
Jeremy Roberts
I think he has something like 60 direct reports. And he called.
00;03;38;10 - 00;03;39;03
Geoff Nielson
The.
00;03;39;06 - 00;03;41;19
Jeremy Roberts
You have company. No I don't I don't think I could handle six.
00;03;41;23 - 00;03;46;07
Geoff Nielson
I've done I've done 20 and I don't recommend it 62 like that's that's a lot of people.
00;03;46;07 - 00;04;00;05
Jeremy Roberts
You got to change up your, your management style. When he gets interviewed about this he always says, oh, the way that org charts traditionally work just doesn't work for me. I don't think he does one on ones. I don't think he sees any value there. I think he's very vertically focused instead of horizontally focused in the sense that, like, you can be a relatively junior person.
00;04;00;05 - 00;04;18;16
Jeremy Roberts
I was reading a story in the New York Times. You can be a relatively junior person in a meeting with Jensen Huang, because he happens to care about that particular project that you're working on. So it's a unique organizational structure, and it puts Nvidia in a tough spot. They've announced an investigation. They're going to hire an external law firm and he's going to lead them through the investigation.
00;04;18;16 - 00;04;38;25
Jeremy Roberts
But they're totally confident that nothing was leaked and that nothing bad happened. He sent an email to all Nvidia staff, basically saying, yeah, I did have this affair. And it was really bad for my for my family, which, you know, I understand that there's probably some personal implications for, for Mr. Huang there. But I think this puts in vedere in a really tough spot because they can't exactly push him out.
00;04;38;25 - 00;04;40;16
Jeremy Roberts
I mean, like, can you name another Nvidia employee?
00;04;40;18 - 00;05;05;06
Geoff Nielson
No, it's a super tough spot and I can't he has no clear successor. But like if this was just Nvidia and no other big company or big tech company was like that, to me that would be a much less interesting story. This like weird structure, this weird corporate governance we've got around Jensen Wong with no obvious replacement like this has actually become endemic in Big Tech, right?
00;05;05;07 - 00;05;27;05
Geoff Nielson
Like if you think about what's going on with like, the Sam Altman's of the world, the Mark Zuckerberg's the Dario Amadeus, like Elon Musk, right? Like there's so many of these larger than life personas, all male, by the way. Generally, yes. And it just speaks to the like the human risk that like, we've kind of deified them, right?
00;05;27;05 - 00;05;49;25
Geoff Nielson
We've treated them as like they're so systemically important to these companies, and we've concentrated power in them. And we've been very happy to do that. And we've believed that, you know, oh, this stuff is like too dangerous or important for the market to try to sort out. But that just comes with this huge downside, which is if they screw up and when they screw up, you can be left holding a very, very big bag.
00;05;49;25 - 00;06;14;16
Jeremy Roberts
Well, it's like a key person risk, like completely a key person risk. So like let's look at how big tech is structured. So obviously the best example of this is probably meta with Mark Zuckerberg founder led. And there's I believe a structure in their governance that gives him pretty extensive controls even beyond what Jensen Huang has. I expect that if Mark Zuckerberg got caught up in a similar scandal, the board probably wouldn't even have the ability to remove him.
00;06;14;18 - 00;06;35;11
Jeremy Roberts
Like, I'm not sure what would happen. The board could theoretically remove Jensen, right? But I don't know that, as you said, there'd be a logical successor there and he'd still own a significant portion of the company. It'd be bad. Elon Musk, excellent example. He has been causing problems for anybody who works in governance at any of his companies for basically as long as they've existed.
00;06;35;11 - 00;06;44;13
Jeremy Roberts
He always pushing the boundaries. He speaks on controversial topics. He just put $500 million into a political campaign just a couple of weeks ago, which I'm.
00;06;44;13 - 00;06;45;25
Geoff Nielson
Sure great for him. I understand.
00;06;45;28 - 00;07;08;29
Jeremy Roberts
Yeah, that didn't work out. His his podcast, I think after after he, after he lost that race, he said, now we need data centers in space, which is gonna that'll be our second story. We'll talk a little bit about the implications of that for the for for data centers, the midterms. But like these companies, as you've said, are structured heavily around big personalities and those big personalities.
00;07;09;01 - 00;07;31;26
Jeremy Roberts
I mean, I think it's a combination of two things, right? Number one, as you become this untouchable person with everybody around you telling you all these good things about yourself, your your capacity to self-regulate, I would imagine decreases. Right. And number two, you don't actually change all that much. As an individual, I would expect, I think that you are still subject to all of the things that make us human.
00;07;31;26 - 00;07;51;20
Jeremy Roberts
So just because Jensen Huang wears a leather jacket everywhere doesn't mean he has a shield against all of, you know, the faults of humanity. And then a third thing that's relevant is that will you become a target? Right. So let's talk about the cybersecurity and just the general security implications of individuals. I mean, people are the entree for most hackers or bad actors into companies, right?
00;07;51;22 - 00;07;53;22
Jeremy Roberts
I mean, that's how that works.
00;07;53;24 - 00;08;17;22
Geoff Nielson
No, it's it's like it's like, frankly, a disaster. And it's so challenging for boards because they know that investors are investing in the individual. Right. And so what I imagine you're going to see with Nvidia and if this happens, you know, elsewhere at a structured company like this is an extreme reluctance to do anything with Jensen other than some like finger waving and like hand-wringing.
00;08;17;22 - 00;08;18;25
Jeremy Roberts
Yeah, don't do that again.
00;08;18;28 - 00;08;27;12
Geoff Nielson
Right. And the biggest risk they have is if something really damning comes out and their hand is forced, right? And they feel like they have to do something.
00;08;27;13 - 00;08;29;15
Jeremy Roberts
You can't really think of anything more than this.
00;08;29;16 - 00;08;32;19
Geoff Nielson
No, no, no. But I mean, if they do this investigation and oh.
00;08;32;20 - 00;08;34;25
Jeremy Roberts
And they determine he gave some sort of a secret.
00;08;34;25 - 00;08;54;25
Geoff Nielson
Or he gave some secrets, that's it. You know, actually ended up putting like a large amount of chips in Chinese hands. Like there are there are facts about this story that could be really, really damning for him and for Nvidia that forced the board's hand. And I'm sure they're like individually and collectively praying that that doesn't happen because of how bad it would be for the company.
00;08;54;25 - 00;09;13;23
Geoff Nielson
And to your point, like this is a company that's market cap right now is equal to about the same as the bottom 213 companies in the fortune 500. Like, this is so structurally important for the US economy and the Western world. Like it's just it's so much concentrated risk and so few people.
00;09;14;00 - 00;09;40;07
Jeremy Roberts
Right. And so it's totally disastrous. And I will say, just for formality sake, Jensen Huang admits the affair. Yes. In vedere acknowledges the affair, says it's going to do an investigation. And the Chinese government denies that the woman in question is a spy. That said, they probably would deny it in vedere denied the affair up until they admitted the affair, so I don't know how much I trust any of their other denials, and I guess we'll have to wait and see what happens when the when the dust settles.
00;09;40;13 - 00;09;58;29
Jeremy Roberts
A couple of other takeaways that might be relevant here. Number one key person risk matters a lot. So if you're watching and you run a smaller company, while you may not be the target of an operation like this from Chinese intelligence, it's good to interrogate the motives of people who worm their way into your circle. Human intelligence is growing in importance.
00;09;59;04 - 00;10;21;01
Jeremy Roberts
And, you know, we've seen this with North Koreans getting jobs at companies. And the recent Binance hack, which was through job interview on a DevOps engineer, that they breached it that way they got access to a laptop. It's very important that people are aware that these types of things do happen. And if you're involved in a critical space for national security, well, your security posture should reflect that.
00;10;21;02 - 00;10;21;07
Jeremy Roberts
Well.
00;10;21;07 - 00;10;48;00
Geoff Nielson
So let's let's talk about that for a second. Because we were talking last week about the Binance story and about the, you know, how this hack was able to happen through one person and one laptop. And we talked, you know, for a reasonable amount of time about, you know, having a zero trust security environment. Right. This is actually having basically a security architecture where even one compromised person or endpoint is doing limited damage in the system, like can that withstand a Jensen size?
00;10;48;02 - 00;11;16;23
Jeremy Roberts
No, it can't fundamentally know, like, I mean, if you compromise the CEO and the CEO says, I think we should do it this way. I mean, like the the key backstop to this is the board of directors, right? Which is, as far as I understand, largely independent. So a wide variety of folks, but it's not super powerful in Nvidia's case, and they haven't moved towards firing him, which makes me think that while legally they might be empowered to do so, they're relatively toothless because of the way he set up the company with no obvious successor and the 60 direct reports or that that massive number.
00;11;16;23 - 00;11;35;16
Jeremy Roberts
So, like, I think the indication here is that these big companies that are pillars of the global economy probably should have stronger, independent governance. And maybe this case will be the thing that forces that into legislation somewhere. Maybe the external pressure will get so bad that the Board of Governors removes him directly, the board of directors removes him.
00;11;35;18 - 00;11;46;17
Jeremy Roberts
But I don't know. I mean, I think that, yeah, if the CEO says, hey, I'm going to do this bad thing, that's like the guy steering the ship, pointing it at an iceberg, like there's not much you can do. That's the guy with the wheel. So. Well, I.
00;11;46;21 - 00;12;09;10
Geoff Nielson
Say that somewhat reluctantly, but I think there's like a broader sober takeaway for like, executives everywhere, which is when you're executive, you don't have the benefit of having a Chinese wall that feels like, like a bad pun that it wasn't meant to be, but like a Chinese wall between your professional life and your personal life. And like the reality is that that just doesn't really exist anymore.
00;12;09;10 - 00;12;16;10
Geoff Nielson
And anything, you know, any indiscretion on the personal side can very easily be an enterprise level risk.
00;12;16;10 - 00;12;35;21
Jeremy Roberts
And so this happens all the time. Executives get in trouble for their personal relationships constantly. Like this happened at McDonald's not too long ago. CEO got pushed out for a personal relationship. I'm struggling to think of some other examples off the top of my head, but I guarantee you you can find it. Because guess what? When you have a lot of power, people try and worm their way into your circles.
00;12;35;25 - 00;12;54;10
Jeremy Roberts
You're extra vulnerable to that type of targeting, and maybe you're blind to it because you know you're finally getting what you want in life. So yeah, I think this is a wake up call for anybody in a position of power, interrogate the motives and honestly, just try not to behave this way because it never works out well in the end.
00;12;54;11 - 00;13;05;11
Geoff Nielson
Yeah. And it just creates such a systemic risk. And it's up to boards and to everybody to make sure that we're creating a system that holds people accountable. And yeah.
00;13;05;14 - 00;13;10;10
Jeremy Roberts
Yeah, I know you can't you can't rely on individuals. But you know what you can rely on Geoff?
00;13;10;12 - 00;13;11;16
Geoff Nielson
What's that? Jeremy.
00;13;11;20 - 00;13;13;17
Jeremy Roberts
The great research produced by Infotech.
00;13;13;17 - 00;13;14;29
Geoff Nielson
Do you want to add read time already.
00;13;15;00 - 00;13;16;15
Jeremy Roberts
Yeah. Time to read that ad Geoff.
00;13;16;15 - 00;13;17;03
Geoff Nielson
Let's do it.
00;13;17;04 - 00;13;19;15
Jeremy Roberts
Let people know.
00;13;19;17 - 00;13;40;09
Geoff Nielson
Well, Jared, as you know, I talk to a lot of CIOs. Yes, everybody knows that running a successful IT shop is extremely hard and AI has only made it worse. It has, you know, the drill immature processes, a fragmented data ecosystem and unlimited demand for it. It all makes for an impossible job. And yet nobody has invented a unified IT operating system until now.
00;13;40;12 - 00;14;01;10
Geoff Nielson
That's why Infotech created CIO analytics, a single platform to run the IT department. We call it API driven advisory. It's powered by your data, our analyst insight, and it works with your existing tools right out of the box. Better context, better insight, better IT performance only with API driven advisory, only with Infotech. So check it out at the link below.
00;14;01;12 - 00;14;01;27
Jeremy Roberts
Infotech.
00;14;01;27 - 00;14;23;11
Geoff Nielson
Com and I do have to quickly say, by the way, that as insane as this sounds, the story we were just talking about, which is like executive indiscretions, this is literally something that Infotech has helped with in the past, if you believe it or not. This is I'm just going to go into quick story mode. But the very first call I was on like 15 years ago at Infotech, I was shadowing this call.
00;14;23;11 - 00;14;43;14
Geoff Nielson
I was with somebody with a cyber security expert. And there is a senior security leader at this large organization and says we have this huge problem. You're never going to believe what it is. Oh, no. And the the security expert on the infotech side says, let me guess, your employees are watching porn at work. And the client gasps and said, how did you know?
00;14;43;21 - 00;14;54;06
Geoff Nielson
But there's more. It's way worse than that. And he said, let me guess, your executives are the worst offenders. And he said, how like, how do you know this? And he said, because I deal with this all the time.
00;14;54;08 - 00;14;55;04
Jeremy Roberts
It happens constantly.
00;14;55;04 - 00;15;12;28
Geoff Nielson
Right. And it comes down to educating and making sure people know the risk of what's going on. And so, you know, I know it sounds insane. And we've been talking about like the technology side, but the people side is even more important than that. So I digress. I just needed to get that anecdote on the table. I also just think it's like pretty funny.
00;15;13;00 - 00;15;14;04
Jeremy Roberts
It is pretty funny.
00;15;14;07 - 00;15;40;17
Geoff Nielson
Like depressingly funny, I suppose. But with that, let's talk about our second story. And that is data centers and the role that they played in the US midterm elections. So I'm not going to dive into all the election results. You can watch a thousand podcasts that will do that for you. But what I do want to talk about is the fact that a number, if you're looking at Texas going almost entirely blue, including a blue governor, you've got a blue governor in Ohio, we've got just Georgia.
00;15;40;22 - 00;16;05;20
Geoff Nielson
In Georgia, we've got just a wave of Democratic candidates, a lot of them sailing on this popular discontent and backlash with data centers. We've got more than 50% of people saying they don't want a data center in their community. And it's just become, you know, bankable. You can build a political career as an anti data center candidate. And we're seeing, you know, stocks drop around that.
00;16;05;22 - 00;16;30;06
Geoff Nielson
We've got, as you mentioned earlier, Elon Musk saying well I guess it's time for these straight to space right to space. But so I mean I guess I wanted to ask you, Jeremy, because, you know, a little bit more of the technical side of data centers than I do, like what happens to AI broadly to our usage of AI, to open AI, to anthropic, to, you know, space acts.
00;16;30;10 - 00;16;36;12
Geoff Nielson
If we can't build out data centers at the pace that we've been talking about for the last couple of years.
00;16;36;18 - 00;16;58;00
Jeremy Roberts
So a couple of things I'm going to change the premise of your question. Number one, it's not that we can't build data centers, that we can't build them in places where citizens have revolted. And I take media queries. And I got an interesting one a couple of weeks ago about data centers in Latin America, because I think there are folks down there who are like, well, if you don't want to build them in North America, maybe you can come to Brazil or to Chile.
00;16;58;01 - 00;17;19;21
Jeremy Roberts
You know, we've got extra power in, I think Uruguay, for example, very heavily hydroelectric powered. Right. So like there's lots of incentives. So I think the first thing that happens is the data center builders go jurisdiction shopping. And there's a whole bunch of technical reasons why this is difficult, like building power lines is extremely infrastructural intensive. It takes a long time.
00;17;19;24 - 00;17;27;10
Jeremy Roberts
There's if it fails at any point in the stream, you're unlikely to be able to power a data center at the scale, which is part of the reason that the US is so popular.
00;17;27;18 - 00;17;33;03
Geoff Nielson
But it is like I have to imagine, like data residency and like governance, privacy, security concerns become a risk as well.
00;17;33;04 - 00;17;34;01
Jeremy Roberts
Oh absolutely. Right.
00;17;34;02 - 00;17;35;10
Geoff Nielson
Like but they're not insurmountable.
00;17;35;11 - 00;18;06;03
Jeremy Roberts
No, no. And I expect that for large language model training, which is what a lot of these data centers would end up doing, I think that there's definitely a case to be made for going to novel jurisdictions. So number one, I think people start looking elsewhere. I think intra inter excuse me, jurisdictional competition is is very likely to ramp up when it becomes unobtainable in popular places like Georgia and Texas and Ohio that you mentioned, which have all been sort of fairly heavily data center concentrated.
00;18;06;03 - 00;18;32;14
Jeremy Roberts
Virginia used to be the leader. They've sort of rolled back on this. And then I think the other thing that it does is it inspires sort of technical innovation, like I wrote in my notes, that this might not actually a bad be a bad thing for data centers and sort of society at large in the long term, like when it's when you are backed into a corner, when you are forced to say, okay, I've got to confront with a like a megawatt limit on my data centers power draw or I can't build in a particular location, I have to build somewhere else.
00;18;32;14 - 00;18;51;01
Jeremy Roberts
Or like, you know, I have to go to space or I have to go underwater or I have to go to the Arctic or something like that. Sort of necessity breeds innovation, and perhaps we see a lot of really high quality innovations that come out of this. Maybe our data centers are able to cool them more efficiently with less water.
00;18;51;01 - 00;19;11;10
Jeremy Roberts
I know there's been some some big moves in that space. Maybe networking gets better so I can build more smaller data centers that are less objectionable to people. Maybe I figure out how to get launch prices down and actually build a stupid space data center, which is stupid. Maybe Microsoft revives Project Nemo and they start dropping them in the water for really efficient cooling, keeping them close to coastal communities.
00;19;11;10 - 00;19;17;15
Jeremy Roberts
You don't need a data center in LA. You can have one in the harbor, you know, and people complain about it less. Maybe the fish don't like it.
00;19;17;16 - 00;19;41;10
Geoff Nielson
There's I mean, all of that makes sense to me, and I expect we're going to see a lot of that. The piece that I'm still grappling with is like, I guess, facts versus feelings. Like there's so much backlash here and it seems like some of it and I'm not here to just be like a data center apologist. Some of it is based on like, hard facts, and they absolutely take a ton of, electricity to run.
00;19;41;11 - 00;20;00;19
Geoff Nielson
That's a big build out. I, you know, the water piece, from what I understand is actually, if you look at it in the context of things, they're not taking nearly as much water as like, you know, like public golf courses or like, you know, raising cattle or, you know, other big, you know, projects. Does that matter? Like, is does it matter what the facts are here?
00;20;00;19 - 00;20;16;27
Geoff Nielson
Is there any point in like, kind of public education or like if we're actually looking within a jurisdiction to try and get a a data center built? Is it is it financial? Is it like, hey, we will pay, we will financially incent the community to put a data center here.
00;20;17;00 - 00;20;41;08
Jeremy Roberts
So data centers are unpopular for a bunch of reasons, right? They are perceived as dirty. And there have definitely been reports there. Dirty. They use a lot of water relative to a non data center build out in a given place. They're big, they're ugly. They are loud. They can be very loud at community meetings around the country. People I saw a video of a guy who showed up and he actually played the sound that he hears, you know, every day from a data center.
00;20;41;08 - 00;21;01;24
Jeremy Roberts
It's a loud wine. It's the the powering of them is loud, the cooling of them is loud. And I think what's key is that this isn't like a factory in your neighborhood, like, oh, a factory is coming. We're going to employ everyone. This is like the inverse of a factory. This is a big building is coming. And not only is it not going to create jobs, this is where all the computers that will do your job are going to be right.
00;21;01;24 - 00;21;10;00
Geoff Nielson
And we're going to actually siphon like it's it's extractive. Right. Like it's, there's, there's a few billionaires and they're going to just like really make a ton of money.
00;21;10;01 - 00;21;26;00
Jeremy Roberts
Exactly, exactly. And so, you know, to your question, you know, is it is it an image issue. Is it an education issue. Is it a PR issue? I mean, sure, it's a PR issue, but it's like a pretty fundamental one. It's like, okay, this Ted Bundy fellow, he's got a PR issue. That's right. Ted Bundy, not a popular man.
00;21;26;01 - 00;21;35;17
Jeremy Roberts
Let's peel back the layers a little bit here. Oh, it turns out the reason he's not very popular is because he did some really bad things. And I would argue that data centers are not quite to that level. But like.
00;21;35;18 - 00;21;38;06
Geoff Nielson
You heard it here first, data centers are not quite.
00;21;38;08 - 00;22;01;18
Jeremy Roberts
Serial killers. Yeah. It's their unpopular because they're not inherently good and they take away jobs from people. At least that's how they're perceived. And now, again, I think that they can do a lot on the environmental front. I think that they can do a lot on the noise front. I think that, you know, once power grids come online and we got some nuclear and some hydroelectricity that can power this, you know, some of the noise complaints will decrease.
00;22;01;21 - 00;22;18;18
Jeremy Roberts
To your point about can we pay people off. They've tried that. Small parishes in Louisiana, meta made big fuss about this. There were some teachers getting bonuses in the five figures, you know, $50,000 because tax receipts went up in a small parish and the school district was able to afford to pay them. And that's, you know, because they built a giant campus there.
00;22;18;19 - 00;22;49;11
Jeremy Roberts
So that is a thing. And I'm sure that on the outside that can happen. But like, these are incredibly unpopular, I think, because they speak to something fundamental. And I have some numbers here, 56% of voters. This is according to a AP poll with a consortium of news organizations. They said that data centers were a crucial or major factor in their vote, which makes it like a top five issue, which is pretty wild since most people probably don't know what a data center was until a few years ago, and 76% of voters said they opposed new data centers around them, and 76% of people don't really agree.
00;22;49;15 - 00;23;09;21
Geoff Nielson
But that's it, right? Like, what's so fascinating to me is that this is so bipartisan in an era where there's so little bipartisan agreement, it's like we all agree, we don't like data centers. And so, like, I think this is like a roadblock that's here to stay. Like, I don't think we're going to see a data center construction boom in the US anytime soon.
00;23;09;21 - 00;23;10;23
Geoff Nielson
I read something, actually.
00;23;10;27 - 00;23;11;21
Jeremy Roberts
We're in the middle of one.
00;23;11;21 - 00;23;29;27
Geoff Nielson
But while sorry, I mean, it has to like it has to hit that class to level up. It's so unpopular. And I think anywhere, you know, I've read something this morning about, oh, Canada can help. And it's like, okay, unless Canadian politicians want to get reelected. Like it just seems like it's so deeply unpopular. So like, where does it go from here?
00;23;29;27 - 00;23;48;28
Geoff Nielson
Because one of my thoughts is that it just becomes costs absorbed by big tech and probably passed back on to consumers by saying, like, look, it's costing us more to build these. We have to, you know, put them in Uruguay or outer space or wherever, or we can't build as much or there's more R&D and we have to get them more efficient.
00;23;48;29 - 00;23;54;05
Geoff Nielson
Like, I don't know, like all of these roads to me just end with costs rising.
00;23;54;06 - 00;24;11;07
Jeremy Roberts
Yeah. And we're seeing that on the balance sheets of the tech companies, hundreds and hundreds of billions of dollars every year per big tech company, mostly going towards data centers, the buildings and the stuff that goes in them. Sometimes they buy the chips before they even have the building, which could be a problem for Nvidia if people stop stockpiling chips that they don't have buildings for.
00;24;11;10 - 00;24;39;20
Jeremy Roberts
So yeah, where does it go? I mean, I'm not great at predictions, but I will say that I expect that again, backed into a corner. They will be forced to innovate. They'll get creative. Whether it's one jurisdiction saying, you know what, our voters can tolerate it. You know, put them here. Whether it's an international expansion, whether I don't think it'll be space, but it could be something like, something like, underwater or whether they go, like, weirdly decentralized, like everyone's got one GPU in their backyard, I don't know.
00;24;39;21 - 00;25;02;18
Jeremy Roberts
And then the networking improves to facilitate something like that. I think, again, those engineers, when they've got $200 billion per big tech company at their disposal, are incredibly creative. And I really do think that they'll figure out a way to get around this. So I think that the building is going to continue apace. I just don't think it's going to be 500 acre campuses in the middle of farmland next to a subdivision.
00;25;02;19 - 00;25;04;05
Jeremy Roberts
I just don't think that's palatable anymore.
00;25;04;06 - 00;25;29;21
Geoff Nielson
I think, personally, my cynical favorite part of this story is that it's just like, I hope, forced the hand of a lot of these companies to, like, reckon with how largely unpopular what they're doing is. Right? And like most people in most places are not, you know, in the in the wealthy world are not sold on this. And like, they're not buying what they're selling.
00;25;29;21 - 00;25;37;23
Geoff Nielson
Right. And they need a better story. They need to bring better ideas and better approaches to market if they're going to, you know, continue this growth.
00;25;37;24 - 00;26;04;02
Jeremy Roberts
Well. Before we jump on to our next story, which I think is equally compelling, I'll just say there's a famous saying in politics that all politics is local. And I think that the geopolitics of data centers are very clear. You want to win the AI race, you got to build as much as possible. But that's bumping up against the old adage that all politics is local, and it's almost impossible for me to put them anywhere, even though I need them, maybe in the aggregate, which is why there's been some tension between candidates and the white House, because the white House is still very pro data center build out.
00;26;04;02 - 00;26;37;04
Jeremy Roberts
So maybe that's an interesting place to take it. Now, our next story, Geoff is from anthropic. So an engineer basically came out, I believe he's a fellow named Adi Osman, and warned that overreliance on AI can cause your skills to atrophy. Basically talking about how in work as a software developer software engineer, you spend less time these days actually sort of solving problems, getting to the root of things, understanding the code that your bot generates.
00;26;37;06 - 00;26;52;25
Jeremy Roberts
And that's cognitive offload. And I wanted to talk about that. So Geoff I mean what's your take on this. You're a leader. You're an executive. You hire people. You've had 20 direct reports at one point. Are you worried about the future of of hiring people and building your employees?
00;26;52;29 - 00;27;11;26
Geoff Nielson
I mean, first of all, I'll say like from my vantage point and I don't know about yours, Jer, but like this, like report or this, you know, release that this anthropic engineer is saying, like, to me, where I'm sitting, this is the most nakedly obvious, like statement I've ever heard that like people who are using AI more are thinking less on the most part.
00;27;11;28 - 00;27;36;24
Geoff Nielson
Like it is blindingly obvious and it is so bad for business. It's so bad for people in general, right, that we've just used this as a cognitive shortcut and like surprise, It's being advertised to cognitive shortcut. And B, it's just like you've given people a tool that actually at face value is a cognitive shortcut. Yeah. It's what it's for you.
00;27;36;26 - 00;28;16;17
Geoff Nielson
Yeah. It's ostensibly what it's for, except that it's worse. Like the output is worse, that you're letting the muscle atrophy like, you know, the muscle there being your brain, you're not actually thinking productively about, is this good? Am I engaging with it? And so, I mean, I worry I think there's a lot of smart people out there who can figure out how to use this effectively, and I worry that there's a lot of people out there who are either dumb or they're overwhelmed, or they're in some sort of circumstance where the pace of change and the pace of their demands and their life have gotten so high that they feel like they have no choice
00;28;16;17 - 00;28;36;11
Geoff Nielson
but to put out, you know, AI driven, I'll call it slop. And I think everybody suffers for it. Right? There was, you know, an article I was reading this past week about remote work. And the conclusion they had about remote work is remote work is actually better than in-person work. If what you care about most is quantity of work.
00;28;36;14 - 00;28;43;20
Geoff Nielson
But in-person work is better if you care about quality of work. And I just thought to myself, like, and this was not in the article, but it's just like.
00;28;43;21 - 00;28;46;09
Jeremy Roberts
You're really going to upset some YouTube commenters, by the way. But keep going.
00;28;46;11 - 00;28;50;07
Geoff Nielson
Like quantity of work, like as a standalone metric is so stupid.
00;28;50;07 - 00;28;50;28
Jeremy Roberts
Oh, it's absurd.
00;28;51;01 - 00;29;10;11
Geoff Nielson
Like like you just oh, look at how much I did like that is so stupid. And I worry because I've seen firsthand and secondhand what some of this so much work is, and it's thoughtless and it's not going to help your organization. It's not going to help you professionally. Maybe it will over the next like six weeks. But this is this is a huge issue.
00;29;10;11 - 00;29;25;16
Geoff Nielson
And I see the chickens coming home to roost like sooner or later. Like how much damage are we going to do to ourselves and to our organizations before we figure out that this is a terrible AI use case? I don't know, I hope it's less rather than more.
00;29;25;17 - 00;29;35;12
Jeremy Roberts
Well, if AI as an assistant for coding is a terrible AI use case, I've got a several trillion dollar bubble that you can go go ahead and well.
00;29;35;14 - 00;29;42;23
Geoff Nielson
Hold on. Like, first of all, I think coding versus general, oh, it's going to replace knowledge workers or it's going to ten x.
00;29;42;26 - 00;29;49;14
Jeremy Roberts
This is about coding like this. This sort of our entree into this conversation. But what Adi Osmani is talking about is software engineering right.
00;29;49;15 - 00;30;10;13
Geoff Nielson
Right. And so coding let's specifically hone in on coding. And as you know, my team right now is not coders. I've worked extensively with coders in the past. But with coding specifically, I think there's a lot of good that can be done with AI, that it can write code quickly, that you can tell it what you want to do.
00;30;10;13 - 00;30;34;29
Geoff Nielson
And by the way, this part is extremely important. The better you tell it what you're looking for, the better the code is going to be. If you have like a one line prompt, it's going to be shitty. And if you don't catch that on the other side, right, like it's you can have an AI sandwich, but you have to be smart off at the front of it asking what you want, and you have to be smart off the back of it, actually validating that it did what it said it would do.
00;30;35;05 - 00;30;45;17
Geoff Nielson
Right. If you can do those things well, you can have an absolute force multiplier here. If you do both of those things lazily, it's going to be a disaster.
00;30;45;19 - 00;31;06;00
Jeremy Roberts
Well, I see it as sort of a tale of two coders. Okay, I've been working. I practice this in the mirror this morning. So you've got, on the one hand, people like Linus Torvalds, who have been in the game for 30, 35 years plus and are widely renowned as deeply understanding their fields of expertise. Right. If I want somebody to work on the Linux kernel, he's going to be top of my list.
00;31;06;01 - 00;31;42;18
Jeremy Roberts
And then you have an entry level software engineer who maybe graduated from high school during the pandemic and is in the first couple of years of their professional career. And when we talk about using AI to augment developers and engineers, we actually are often talking about both of those people at the same time, and we should not. I have sero zero concern with Linus Torvalds using this, because I know for a fact that he understands what he gets out the other side, what I am horrified for and terrified for, and name the adjective that's how I feel is when we have a young person who is just getting started, who should be using that early
00;31;42;18 - 00;32;00;22
Jeremy Roberts
career to what was it Miss Frizzle said, get messy, make mistakes, whatever, to do all of that, and they're just pumping out this slop that they can evaluate the efficacy of. And those junior people who are pumping out slop will soon become senior people if they don't get pushed out of the business. And that's a huge issue they fundamentally don't understand.
00;32;00;22 - 00;32;19;10
Jeremy Roberts
So in much the same way that nobody knows how to read a map anymore, because we all use Google Maps, or in the same, same vein that, I don't know, there's like two phone numbers that I know and one of them is 911. We're going to forget how to do a lot of this stuff. And this is terrifying to me because we talk about doing higher order work as something that AI enables.
00;32;19;10 - 00;32;41;20
Jeremy Roberts
But if you lose those skills, how will you be an architect if you can't even do basic entry level design work or entry level engineering work? And so to me, the reason I picked this story for inclusion here is because this is a crisis, and not just in software engineering. It's a crisis everywhere. How are you going to be a good journalist if the Financial Times or whatever newspaper uses AI?
00;32;41;26 - 00;33;01;08
Jeremy Roberts
I'm not going to accuse them specifically. I don't know that they do this, but pick your newspaper. They use AI to generate stories and have a senior editor look at it. What happened to the role of the entry level researcher? Same goes for consultants. You used to move little boxes around on PowerPoint decks. Remember that? Now, I heard someone say I never move box around again.
00;33;01;08 - 00;33;10;14
Jeremy Roberts
I have a I do that what happens that entry level person accountants, lawyers even in the medical side maybe the less hands on medical side. I'm very worried about that. Geoff. Are you.
00;33;10;15 - 00;33;34;11
Geoff Nielson
I'm. Well, I'm very worried about it. And there's one piece that I want to hit on that we talked about earlier that especially concerns me, which is AI actually creating growing inequality in the workforce. And what I mean by that is like, if you actually listen to what we're saying, we're basically saying, like, you know, young people, entry level people don't use AI, you're too stupid to use it properly.
00;33;34;11 - 00;33;49;03
Geoff Nielson
But the people who, like, already know what they're doing, the experts, they're allowed to use it. So they they're going to use it. And you know what? They're going to be like ten x better than they already were. And they're already ten x better than you. So now there's like a 100 to 1 gap between them and you.
00;33;49;05 - 00;34;20;20
Geoff Nielson
And you have to just like inch your way forward the old fashioned way. And they're just going to, you know, exponentially improve that. Like that doesn't sit well. Right. That seems like destabilizing is the word that's coming to mind. And I think we have to talk a little bit more about that. And like, what do you do to make sure that you don't like, again, like to bring a bunch of conversations together, create key person risk because you just have a few people who actually know what they're doing, and everybody else is just left behind.
00;34;20;21 - 00;34;37;29
Jeremy Roberts
I keep person is Claude know in this case, Geoff? I mean, I would say and is my hot take of the day that like, not every problem actually has a solution. There's an asteroid barreling towards Earth, people. That seems unfair. Yeah, it is kind of unfair, but there's not really all that much that we can do about it.
00;34;37;29 - 00;34;56;14
Jeremy Roberts
So in this case, this software nuclear bomb has just dropped on the software development industry, and we're just arguing about how radiated we're all going to be. I mean, like it's yeah, I agree with you. Everything that you said that really does suck for entry level software engineers. And I think many of them, if they're listening right now, probably felt that pain.
00;34;56;14 - 00;35;11;03
Jeremy Roberts
They feel that AI has completely eaten up a lot of the work that they would do. Some of them maybe love it, some of them don't. Some senior engineers maybe love it, some don't. I just, I don't think there is a just world where there is a correct answer. I think there's versions of less bad and that's what we have to settle on.
00;35;11;03 - 00;35;24;09
Geoff Nielson
So I'm not quite I'm not quite ready to just like let you off the hook and say, yeah, sure is bad for young people. The end. All right. You know, cut to credits. Yeah. Like to me I want to get to like so so what do you do? So what if you do what do you do if you're entry level, you're learning.
00;35;24;11 - 00;35;55;08
Geoff Nielson
And like AI atrophy is a concern. And I think the answer is you still need to use AI. Like like I think the like well, I'm never going to use AI is also not the answer. You probably still need to use these tools. You need to be more critical about is this right? Whether that's interrogating AI further about the answers, whether that's talking to a senior or a person who knows what they're talking about, making sure that you understand what you're asking for and you understand what you're looking for.
00;35;55;15 - 00;36;12;25
Geoff Nielson
To me, these are all the ingredients. Like it just comes down to critical thinking and deliberately applying, you know, those neural muscles versus like, how is this just a productivity tool for me that that to me is a conversation that I think is really worth having.
00;36;13;00 - 00;36;28;16
Jeremy Roberts
So if I could offer one piece of advice then to somebody managing entry level software engineers, I would just say make sure they understand what they're submitting to. You force that in your reviews, don't let stuff fly by and really screen for people who are throwing slop grenades. As to Luca, absolutely.
00;36;28;16 - 00;36;48;19
Geoff Nielson
And I think like this is true in coding and it's true in general and knowledge work, if people are submitting AI generated content that they have not read, that is like a cardinal sin in my books like and as a leader, you should not be standing for that. People should know that that's not okay. It's not okay for you as a leader.
00;36;48;19 - 00;37;04;05
Geoff Nielson
It's not okay for the business, it's not okay for them and their development path. So like, I know it's like a stupidly simple like takeaway, but read the damn output and understand it before you deploy it into production or share it in an organization.
00;37;04;05 - 00;37;22;26
Jeremy Roberts
Maybe as a final point, I had a conversation with someone the other day and they said that one of their colleagues sent them AI slop and they corrected them. Didn't stop sending up, going back and forth, receiving the slop, putting it into their own AI, sending it back. And then they would take the slop and at certain point it just became two AI's having conversations with each other.
00;37;22;28 - 00;37;36;27
Jeremy Roberts
And the amount of productivity, I think went negative. Yeah, especially considering they were burning tokens to do this. A complete waste of everyone's time. You do not want to end up in that situation. Wow, what a what a pleasant point to end the show on. Geoff, do you have some optimism before I read this out?
00;37;36;27 - 00;37;59;18
Geoff Nielson
Yeah, I mean, the optimism is that like, I don't believe that this is a nuclear bomb. I believe that actually this is a tool that can be really positive for productivity if it's deployed effectively. And it's up to all of us to make sure it's deployed effectively in terms of how we're using it, in terms of how the people in our organization are using it, and we can use it to get ahead.
00;37;59;19 - 00;38;05;04
Geoff Nielson
But just having it here is not going to make us smarter. We have to be thoughtful.
00;38;05;09 - 00;38;07;18
Jeremy Roberts
We have to be thoughtful. I think that's a good point to end it.
00;38;07;20 - 00;38;08;09
Geoff Nielson
There you go.
00;38;08;14 - 00;38;12;06
Jeremy Roberts
You are watching Digital Disruption now, Geoff do you want to take us away?
00;38;12;06 - 00;38;15;29
Geoff Nielson
I was going to say thanks, everybody for watching. But Jer, why don't you read us out?
00;38;16;01 - 00;38;24;07
Jeremy Roberts
Well, I was gonna I was gonna read us out from memory. You're watching Digital Disruption now, which is a production of Infotech.
00;38;24;09 - 00;38;26;03
Geoff Nielson
Yes.
00;38;26;06 - 00;38;43;00
Jeremy Roberts
You're watching Digital Disruption Now, which is a production of Infotech, a leading research and advisory firm. We talk about stuff like AI all the time, so make sure you check us out at Infotech. Com. Geoff Nielsen is our host. My name is Jeremy Roberts I'm a co-host. The show is produced by Maria Knight. Special thanks to Bryan Moggach for operating the camera.
00;38;43;00 - 00;38;51;25
Jeremy Roberts
James Groome for doing the same thing. Our technical director Tim Champion, and our editor Tim Lai. We will see you next week with more insights on AI and the markets.
00;38;52;01 - 00;38;52;17
Geoff Nielson
Looking forward.
00;38;52;17 - 00;38;54;00
Jeremy Roberts
To it. Yeah, it's gonna be a great time.
The Next Industrial Revolution Is Already Here
Digital Disruption is where leaders and experts share their insights on using technology to build the organizations of the future. As intelligent technologies reshape our lives and our livelihoods, we speak with the thinkers and the doers who will help us predict and harness this disruption.
Our Guest Chris Lattner Discusses
The AI Coding Revolution Has a Dirty Secret
Chris Lattner, creator of Swift and LLVM and co-founder of Modular, joins this week’s episode to talk about coding, vibe coding, and why more code doesn’t always mean better software.
We Discuss
The Jensen Huang Scandal Is Bigger Than You Think
What happens when Big Tech depends too much on one person, and we depend too much on AI? This week we’re covering the Jensen Huang controversy, the backlash against AI data centers, and how AI coding tools could weaken critical thinking.
Our Guest David Weinberger Discusses
AI Has Exposed the Biggest Lie About Humanity
Author, philosopher, and technologist David Weinberger joins Geoff Nielson to explore how artificial intelligence challenges our assumptions about knowledge, human judgment, and business leadership.
Our Guest Jeremy Roberts Discusses
AI Data Centers in Space: Stupid or Genius
Are AI data centers in space a practical solution for the future or just an expensive fantasy?