America at 250, the AI Intelligence Grid and Export Controls

Hey everyone. Happy Thursday. Well, it's July 2nd, but we're celebrating July 4th.

So happy Fourth of July, everyone. Welcome to the Techstrong Gang. I don't know how we got there.

First, it was like, this is the way of the world. First, we take one day before off, now we're celebrating it two days before, but whatever. We're not going to be here tomorrow to do it on Friday, so we're making today, July 4th on the Gang.

Let me introduce you to our gang because we've got stuff to talk about today, and I'm sure they're looking forward to talking about it. I've got my friend down Austin way, R2 Robert Reeves. Robert, it's always great to see you, man.

Thank you. Happy to be here. Happy to have you.

Also joining us, Barbara Rose, not Ruse, I apologize, I introduced her wrong on our last Techstrong Gang she was on. Barbara, it's great seeing you. And then we have two of our regulars, the one and only John Swartz, and the only- Hey, everybody.

Good to see everybody. Yep, and the only one still in an undisclosed location, Mike Vizard. I'm wearing Yankees red, white, and blue today, so there you go.

Well, that's good. If you were in a place that had real lighting, we might be able to see that, but- ... we're going to have to take your word for it, I guess.

Do you believe? Yes. Do you believe in miracles?

Alan's already studio shaming. Awesome. Yeah, here we go.

Look, not everyone could have Techstrong studios right in the office. But no, all kidding aside, it is July 4th, and there's nothing more American than baseball. And if you're talking baseball, there's only one team that God loves, and that's the New York Yankees.

So Mike, with that intro, I'm going to let you kick things off. Right. Well, let's just dive in because you have a post over on Digital CxO that basically talks about, well, what does the meaning of freedom in the age of technology, and what do we need to do?

And you equate it in some ways to, there's a difference between, quote-unquote, free beer and a free platform, as it were. And as you read into your article, it seems like what the gist of it is saying is that it's important for us to be able to open the hood of these platforms, and if we can't open the hood, we can't see how they work. That's pretty much not the definition of freedom.

Now, Robert, I know you've been close to this open source conversation for a long time, but did this resonate with you? And have we lost sight of this as kind of a fundamental right? Absolutely.

The saying was from going back to Eric Raymond, Richard Stallman, back in the day of open source, free as in beer, free as in freedom. The idea that, hey, here you go, you can use this, harking back to the old shareware BBS days. This is a free app, help yourself, pass it along, but you can't have the code.

So freedom in that sense is like, yeah, you can use it, but you can't change it. And so there, that's where Free Software Foundation came from, the idea that it is the code running on your machine, and you should have the rights to change, update, and share those changes. Now, of course, that's for the individual developer.

What about businesses? Should they have the freedom to be able to inspect the code, make changes, and share with others? And that's where we've seen the success of platforms like Kubernetes, where companies, Fortune 500, Global 2000 companies, pick open platforms because that gives them the freedom to not only make sure that those platforms are going to run right, but they can pick and choose their vendor.

And that gives them the power, and that balances it out. It also creates a huge ecosystem for large providers, AWS, GCP, Azure, the rest. Everybody's making money off of selling Kubernetes implementation services or services.

So there is a way to make money with this and protect freedom and the right for developers to experiment and try new things. So here's my little take on this thing is I get all what you're saying, and I conceptually agree, but then I go talk to people, and they all seem to come back to the same issue, which is that's a lot of work. It's a lot of responsibility.

It's a lot of effort on our side, and they're just looking for the easy button, but they don't necessarily understand that that easy button comes with restrictions. Yeah. And then they wonder how come, as Alan pointed out in his article, my data is suddenly locked up somewhere and I can't get it and I'm totally dependent and I can't move it, and then I have all these business issues that come with that.

So- Doesn't sound very free ... yeah. So Alan, do we need a free data movement?

I mean, maybe we kind of speak to what the point of this thing is. Well, I don't know. Because I don't think the problem is whether the data is free or not, or even whether the software is free.

The problem resides between the ears of the users, right? Anyone who thinks something is free as in free beer is going to learn a lesson about life. Nothing in life is free, my friend.

You may not pay for a piece of software or to store some data somewhere, but in spite of Dr. Stallman's Society of Atheistic Saints- No one's giving you nothing for nothing. There's always a price to be paid down there.

And then the question, and this is something I input in the article, but there's an inherent connection between the free as in beer and the free as in freedom. The more free the beer is, the less freedom you get. The more you pay for the beer, the more freedom you have about what you do with it.

Right? And so- Alan, you're describing the entire home brewing market. There you go.

Who knew? I thought I was talking software. But- You were mentioning though, we're in this era of metered AI economics, so to speak.

Well, that's where I was going with this- Yeah. Sorry ... this article, right?

Which is- Yeah, no, I was going to say- ... how is this going to apply when the intelligence grid is up and running? That's my new name for what we're building.

It's not data centers and GPUs anymore. Well, Alan, it's going to drive open source AI. As people start getting sticker shock and realizing that they're restricting users because it's too expensive, they're going to start evaluating whether, okay, do we need these big, expensive frontier models, or can we use maybe a free open source model that's good enough?

It's good enough beats perfection any day. So I- Well, I- Yeah ... I wanted to chime in here.

When I saw the free like a beer versus free like freedom, the other thing I thought about was free like a puppy. You get a free puppy, it's not free. And- Yeah, god, no ...

I think that's the same thing that we're talking about here. There was a semi-analysis study that was released last month that I read about that they maxed out usage across every Anthropic and OpenAI subscription tier and compared the actual compute cost to the subscription cost, and what they found was that a $200 a month Claude Max plan can cost Anthropic up to $8,000 a month in compute, and a ChatGPT Pro account can cost OpenAI as much as $14,000 for one month versus what you consume. And they're getting you hooked.

They're giving you a free puppy, and they have the control to change their pricing model later once they've got you addicted. So maybe there's a mix of a heroin and a puppy metaphor here somehow. Yeah.

But we've seen this play before. That's the first time I've ever heard heroin and puppy in the same sentence. Well, they are dopamine providers, so...

All right. So let me ask Barbara this quickly. Barbara, do business leaders get this conversation, or is this just something that kind of, they just write the checks, but they don't think twice about it?

I think the CFO gets it, but I don't know if other business leaders are getting it yet. And it's kind of interesting because I think the other thing that they don't necessarily understand is choosing an AI vendor is a lot less like a traditional procurement process, and I think a lot more akin to hiring a member of your executive team. You need to be scrutinizing this vendor the way you would scrutinize a leadership hire.

What are their philosophies? What are their belief systems? How do they see their future in your organization?

It's not just buying a piece of software and running it for a while, and then you have the freedom to walk away from it. You're changing how your organization functions around this tool, and that's a very strategic decision and investment. Alan, the point that really jumped out at me was when we talked about this, I alluded to the metered AI economics.

So basically, when experimentation is practically free, the idea, I guess, is organizations naturally encourage a culture of try it and see, see what happens. But when every experiment shows up on this real-time financial dashboard and money's involved, finance teams step in with this pre-approval process, and then it kind of just, the spigot's closed. You know what, John?

Listening to you, I'm taken back to my youth, going up to Forest Park up in Queens, right? " Huh. Right?

And then you'd come back next week and have to pay for it. Do you ever go to Forest Park, Mike, when you were... No, you were in the Bronx, you didn't come down to Queens.

But that's what drug dealers do, my friend. And- Go to Queens, yeah, I know. Well- ...

I know, the Bronx got a bad rap. Alan, they prefer the term amateur pharmacologist. Okay.

But it was always the first one's free. Mm-hmm. And is that the AI industry today, right?

Yes. First one's free. Yeah.

Yeah. But this has always been the model, right? Tom Siebel was doing this, like, three decades ago.

Nah. Mm-hmm. Is that not right?

And then- Deep cut ... Deep cut. And you hear this term forward engineer, right?

Well, that used to just be I hired- It's not a old technique ... a sales engineer, and I embedded him in your environment. I charged you for every dollar of software, I charged you $15 in services, and then I called it a great business model, and we're doing the same thing over again here.

It's just a different play with AI at scale. Mm-hmm. Yes, it is.

For a big, fat paycheck. Well, and Barbara, that was just way insightful about you have to treat this as an executive hire. This is a really big decision.

And also, it's a big decision getting a puppy. And it reminded me of the line from Jamie Zawinski from- Old school Netscape days, and he said that Linux is free if you don't value your time. Open source, yes, it is free, but there has to be an investment on the organization's part.

So, the iron triangle, right? Good, cheap, easy. If you want the easy button, great, but it ain't going to be cheap, and you need to make that decision.

And I welcome this. I like to see this, but this is also a time right now where business leaders and technology leaders can take advantage of those inefficient pricing strategy that Barbara's talking about, how usage of Max and the $20 a month, $100 a month, it costs far more to provide those resources to the provider. This is just like Uber and Lyft back in the day, where it was far cheaper and easier than the cabs.

Now, look at it. So business leaders need to make that decision and understand that, yes, we are going to pay more later, and we need to go in open eyes, knowing that this is going to be difficult. Or we can just be smart about it and make sure that whatever workflow we're creating when using AI is not locked into a particular LLM, and that's kind of the architecture of this thing, and too many people are out there- Yes ...

kind of buying whole hog into one or the other without realizing that at a certain point, there's an interface there that basically is set up to lock you in, and you won't be able to swap out one LLM for another because you've now written a million extensions in the workflows that are all tied to one LLM. Well, the problem is they're hooked on the convenience and the potential cost savings, and they don't want to think about the human side investment that needs to come to be ready to be able to transfer and move between services. That needs to come from the workflow and from the investment in people understanding how to develop these and be able to port them across different systems, but no one wants to make that investment in people right now, and they're going to pay the piper later.

Just like anybody that makes the choice to lease a car versus buy or to sign up for a SaaS service versus buying a software license, not that you can buy a software license anymore, but once upon a time, it was a choice. We're across everything in our lives, deciding between the trade-off of convenience versus control. So John, will somebody create the rights of AI man, and we'll have a tea party and all kinds of stuff?

Is that coming down the pike? What do you think? Oh, inevitably.

Yeah. When you listen to this, it's like history repeating itself, right? And it harkens back decades or even centuries, and what Alan said about, it doesn't just apply to AI, but AI's following the path of human nature and human consumption and human addiction in a certain sense.

We're heading that way again. You know what, I'm- Accelerated at a bigger stage, though. So I will tell you, John, I know your book's about to come out.

Yeah, late October. I actually started working on a book. Oh, good.

Congrats. Well, I'm ear deep in it. That's exactly the point, that we've seen these waves before.

We've seen these waves before. We know how they end or there's rules involved that are going to govern them. But, Mike, your comment about rights of AI man, I think it's no.

It's the rights of men to AI, right? When AI becomes indispensable, like electricity is indispensable, like the internet is indispensable, you have to ask yourself, as a society, should we have rights to it, right? Now in the US, we don't have rights to electricity and the internet, but there are places in the world where that is part of the citizen's right.

And even though it may not be a right in the US, we've basically regulated it, right? We brought electricity out to rural towns that no way could afford the cost of bringing that infrastructure out there, but we did it. We brought highways out to places to connect them.

We connected them to the grids, the electric grid, the transportation grid, the computer grid. We need to connect them to the intelligence grid. It's going to be a right.

All right. Well, since you keep mentioning intelligence grid, maybe we should shift to the B block. What do you say?

Okay, let's go to the B block. All right. ai talking about the need to build an intelligence grid, and this intelligence grid's going to look something like a utility, I'm assuming, and it's going to drive access to AI, and maybe we'll all have some equal access to it, and what we do with it is going to be great.

Alan, from your perspective, is this maybe the next moonshot program that we should be launching as a country? We already did. Have you been sleeping?

Have you been sleeping up there in the cabin? We're going to spend damn near $1 trillion this year on this grid thing, and that doesn't maybe even include some of the second-tier guys like Oracle and CoreWeave and the Neoclouds. The big four or five hyperscalers, Max Sevens, I think are already on the hook for about $725, $750 million this year.

Add in the other guys, you're near a trillion. But, again, let me just give a quick plug. I actually put up a new white paper on Techstrong AI today called "The Dynamo to the Token," and it compares the rollout of the electric grid to AI.

And I think it's worth reading. Go check it out. I don't even know if you have to register for it.

You can just download it. So, wait, you're saying we're doing this already. What I see is a dozen or so independent grids that are not connected in any particular fashion, and I'm supposed to pick one of these grids over the other?

This is not a national utility in my mind. This is just a boatload of venture capital being thrown at the wall and building out a bunch of isolated data centers that may or may not be part of something someday. Well, these data centers are connected to the internet.

Let me put it to you this way. Let's, again, look at this dynamo to the token thing. The electricity utilities, the electric utilities, traded uncapped profits for guaranteed placement, guaranteed real estate.

So in other words, Con Edison serves New York City. I don't know who serves you in your particular areas. FPL down here in Florida.

They became public utilities. But when we talk about, let's say, something like the internet grid, whether today I access it via AT&T, or tomorrow via Comcast, or tomorrow, the day after that, another provider, Verizon, or whoever your internet provider is, we're all accessing the same grid, right? And I think that's where the AI grid's going to go.

Regardless of who operates that data center, that intelligence is ubiquitous, and you may not even care what data center you're getting it from unless you're on the sovereignty kick. Well, unless you're using us-east1. That's the only time you care about- Right, then you care.

Because that always goes down. That all sounds very optimistic and hopeful, but John? Yeah.

6 billion. And Sam Altman, he has not only been talking to Lutnick and Treasury Secretary Bessent and Trump, he's been talking to Bernie Sanders about this public wealth fund that they want. Yeah.

And it wouldn't just be OpenAI. There'd be Anthropic, Meta, Google, similar percentages. So is that, let me ask you- Those projects ...

let me ask you straight out, is that a grid, or is that just the latest form of corporate welfare that we're kicking in here? Oh, you know what? You and I think a lot alike, my friend.

I think it's basically corporate welfare, and it's also- Wait a second. How is it corporate welfare? Because we're using US tax dollars to subsidize a company that- Oh, he wants us to pay for that stake?

Yeah, Alan. Yeah. Yeah.

Oh, no. Screw it. No, no, we're not going to pay for that stake.

I thought they were donating. What nice guys they are. Oh, it's all out of the goodness of their heart- What?

right before they go public, yeah. Yeah, sure So this gets to the heart of this book I'm writing. Look- Oh, good ...

when things become indispensable, they've got to be regulated. Right. And if that means capping profits and everything else, so be it.

Well, sure, but remember, there's the other side of this, that is, the free market, we are told and we would hope, and my undergraduate degree would hope for, is that there is going to be an eventual adjustment of supply and demand here. That we all know that the price of technology tends to go to zero. And Alan, what you're laying out here is that, okay, we're allowing you, government, to regulate us if you give us a monopoly.

In return for the monopoly, we're going to cap our profits at a certain percentage, and we're going to be regulated. Okay, great. So there is going to come a time where that will make economic sense, or those providers will fold.

It's just the way it goes. They may, in fact. But Robert, what you're describing sounded a lot like trickle-down economics to me.

No, it's not. Don't you put those words in my mouth. Where did you get the idea that anybody's going to cap their profits just because the United States has a stake in something?

That's just not going to happen. They're going to maximize- I didn't say take a stake We're aligning this with utilities. " That's what these regulatory utilities do.

The same thing with the, it happened in the phone business, it happened in the railroads. Mm-hmm. Mm-hmm.

It happened in the railroads. Hey, Alan, so in this, I'll allude to your book in progress, but so there's no previous competing wave that's ever come this close in terms of impacting, I would think, physical infrastructure or supply chains. So because of the constraints, and I think you mentioned that AI data centers are going to need an additional 92 gigawatts of power by 2027.

So does that mean the build-out's going to force governments to rewrite permitting laws, utilities are going to have to overhaul decades of demand forecasting, and then these regional economies are going to have to be restructured around data center loads? Is that- Right ... kind of where it's going?

So here's the whole thing. If this AI is really going to be what some people claim it is- Right? And we as a country are getting behind it, and we are installing our next great grid to build the next generation of greatness on.

Yeah, we got to do that. But there are models, there's historical examples for this, right? As a matter of fact, if you look at the money put into, let's say, building the internet out, building the highway system out, building the phone networks out, building the railroads out, building the electric grid out, as a percentage of GDP at the time they were built, it's roughly equivalent to what we're seeing now in AI.

What? So what- The numbers are bigger, but we have a higher GDP ... What a great example.

Robber barons who built the railroads, we're going to use that for AI. But so let me- Yes ... So I'm doing a little research here.

The men who built the railroads went bankrupt several times. Yes. Because you know why?

They thought they were building railroads. They didn't realize they were building transportation. It was the people who rode their stuff on top of the railroads that became lasting wealth.

People like Carnegie and Rockefeller and Sears and Roebuck and Montgomery Ward. Right? And the whole city of Chicago was built.

But the guys who paid and built the actual railroads, they went bankrupt, other than Vanderbilt, because he got his money out before regulation and everything else set in. Barbara, is this making any sense to you? Is this kind of just a replay of a drama from the 19th century?

Well, I think it's a replay of the drama from the 19th century and from the history of the technology industry, too. This is kind of classic. The infrastructure builders capture the value at the beginning, and then the value starts to migrate up the stack to the application and services level.

You know? NVIDIA is- 100% ... making huge money right now, but think about Intel and the build-out of the PC ecosystem, and then the value moved up to Microsoft and then up to the software that sat on top.

Same thing with phones. The money isn't in the chips inside the phones as much as it is all the apps. You want to do my co-op here?

Sure, Alan. I'll let you know my hourly rate after this. Okay.

Go ahead. It's a lot, Alan. A lot.

It's cheap. It's got to be cheaper than this AI. Here's the pressing question on the behalf of the average American: Does this mean my taxes are going down or up?

Up. I think the average American is going to pay for this one way or the other. And the question is, is it in the form of taxes, where we actually, if the government is involved and has a say in regulating, we get a vote in what happens with this build-out and this technology.

Or right now we're being taxed on it, it's just in the form of our utility bills. So we're subsidizing these companies through the electric bill because they're not paying their fair share of what they're consuming. And so it's still hitting our pocketbooks, but with no vote in how they operate and what they do with it.

So we're paying one way or the other. Do you want to have a vote? I don't know.

I think I can hear the AI Minutemen getting together and heading to various cities and towns to defend- Oh, you're hearing them. They're campaigning. They're out in force right now running for election and running against AI and all we're talking about.

Barbara just mentioned- Well, we've discussed that, too, right? The AI tax- I have a lot ... on both sides of this ...

Yeah, it's happening. It's accelerating. There's a megaphone.

We certainly, we've seen the results of that in Utah, where an election was lost. The candidate that lost believed that it was the candidate's support of AI data centers. Well, people- It's happening now It is happening ...

that Stratos out there or whatever it is. There's a social democrat knocked off a 15-year congressperson in Denver, right? Well, yeah, but that has not as much to do with AI data centers.

No, but you can imagine that they're going to line up on this thing. Well, that's the mood. That's the influence, right?

And this is part of that story, right? Progress stops for no socialist. I never thought I would be wistful of the blockchain days.

Yeah. Going back to a previous technology and the tsunami about that. No, but wait.

Let's not kid ourselves. You really- I beg to differ. How is it that a government taking over and funding a utility is not some form of socialism?

Well, it is. It is socialism. It is.

It is. And so why don't you go ahead and check out the utility for your cabin. I guarantee you it's run by a co-op, which is locally owned, and a bunch of people got together and decided to run a utility many, many years ago.

It is socialism. Just like the fire department. But a little socialism's a good thing, I guess, then.

I guess. Just not around Mike. He's in New York.

He's in New York. But seriously, look, if you don't think the American taxpayer has paid for these previous grids one way or the other, you're kidding yourself. You're- Yes, we have ...

first day of Econ 101. We all learned there is no such thing as a free lunch. Yep Back to frees and freedom, but there is free beer.

Let's end this whole little segue right now on that. Mike, what's our third segment today? Our third segment is, what exactly is America going to be like in the age of AI?

" We now have opened that up again, somewhat, a little bit. But John, you've been following this space. It's hard to understand exactly what our plans are as a country here in terms of how we're going to manage AI, and can we manage AI?

And at the end of the day, is this kind of just the Chinese are going to win this conversation anyway? Yeah. They may benefit, and they may probably benefiting already.

It's hard. We joke about, we kind of half-jokingly talk about this and how things were going to change. This has been going on for this stringent two-week export restriction on Mithos-5 and Fable-5.

It's over as of Wednesday. Following, again, this is Alan and I and Mike laugh about this, an intensive national security review. We've heard that before.

So in a sense, my take, and I'll just be blunt about it. So I think we're going from nobody watching to the government pulling the most powerful models offline in under a month. 6.

Oh, yes. So what's taking shape, I think, is this pre-release testing, where the government gets early access and effectively clears the strongest systems before they go wide. And it's interesting because in a sense, when this happens, we're going to be having these improvised rules and companies trying to figure out how to adhere to what the government wants.

Maybe this slows down their process of the models and how they reach the public. And I think the edge kind of goes to whoever can prove their guardrails hold up under a government tester, not whoever ships first. And I think we're seeing, and I'm not familiar with these models, but I talked to a security expert who mentioned Japan has, was it Sakana AI with Fugu?

2, which are kind of, in a sense, these new models that are kind of moving faster than the US can move. And this whole idea of the government oversights, which also perhaps will evolve or devolve, as you want to put it, into these stakes in companies. This, to me, is just kind of worrying, but again, things could change.

They do every four or five days- ... with this storyline. Robert, are you and your open source friends undermining the national interest of the United States of America?

Yes. Just like we did with PGP back in the day. Oh, wow.

Yes. And you're welcome. There was a time where, remember when the US was restricting strong encryption, and here comes PGP and everybody freaks out.

People got arrested, blah, blah, blah. Did not move the needle, and in fact, caused problems because for a while, the internet had very lax security. We had low, very easy keys and those sorts of things.

Luckily, we figured it out and made that go away because what's the idea of how can we restrict access to software in a time where anybody could show up at an open Wi-Fi in the US and download it? What about VPNs? So the same thing, restricting the access to these things, it's going to get out.

What is really interesting about this is that the companies, Anthropic and OpenAI, are willingly doing this. There was no lawsuit. This was not hostile between the two entities, the companies- Well, no, they did- ...

and the federal government That's a good comparison, Robert, because I'm thinking about the Pentagon, that whole issue. This was resolved fairly quickly. That other issue- Yes.

And notice that that happened and then it went away, and so it seemed that what was the result of that? Because DoD is using Anthropic. Yeah.

They're using it in cases. Oh, they settled that. They settled that- Yeah ...

very quietly. So some of this stuff is performative, but the reason why they're accepting this is because they know if they push this, there is a chance that it gets tied up in court. They're just sitting there waiting for the courts to figure it out.

So they're going along with the current administration so that they can continue to operate. If anybody believes that this is for the greater good, it is not. All right?

The federal government is trying, like, "We don't want our foreign entities to get access to it. " And then Anthropic, OpenAI, the other companies just don't want onerous regulations. So they're giving a little bit to avoid any laws being written.

Exactly. Let me ask Barbara this question. When it comes to technology and business in general these days, are borders increasingly becoming irrelevant?

Of course they are. Yeah. I don't know how you could-- The borders are completely porous at this point.

I don't know how you can argue anything else, and I think that there is this move toward I can't remember what it's called, the Sovereign AI? Mm-hmm. I think it'll be interesting to see if that picks up steam.

But right now, there are no borders, effectively. Well, I think we saw that with the Lutnick letter originally that shut down Fable and Mythos. It wasn't just that you couldn't export it, they said foreign-born nationals here in the US couldn't use it.

So what am I supposed to do? Check your passport ID green card before I let you download a piece of software from New York or Miami or Austin? It was ridiculous.

I have some thoughts on this whole thing. Number one, if you are going to be the ones who are calling the shots over when something can be released or not, when it's going to be available or not, you can't also have your hand in the cookie jar. And when you have the Trump and Lutnick families making billions of dollars this past year on stock and crypto, in these very technologies that they're putting their fingers on the scale of, well, that just smells bad.

It's corruption, and it's wrong. And we've got to do something about that in this country. And it's not just Lutnick and Trump, frankly.

We've got to stop the insider trading by members of our government based upon actions that the government's taking in these technology areas. That's number one. Number two, you know what?

Dario Amodei said this over a month ago. He called for a pause. He called for a pause because these models are getting more powerful.

There is more potential for bad things. But he said he can't pause because he has a duty to his investors, to his shareholders, to be as competitive as possible. If he pauses, OpenAI gets ahead of him.

And if the US pauses both OpenAI and Anthropic, guess what? Strategically, we are giving the Chinese and Japanese and anyone else an upper hand. So there are no borders when it comes to this.

If you're going to regulate AI because it's that dangerous, it's got to be a worldwide thing. There is no single entity or company or industry that could stop this train. It's a runaway train.

And- You're right, Alan. It's like back in the day, we can restrict uranium and plutonium- Right ... to stop nuclear weapons.

Apparently, restricting chips isn't slowing this down. No. No.

All it does is it forces the other people to increase their research to make their own chips. That was the second day of Economics 101. Thank you.

We need to make it not about country versus country. We need to make it about good actors versus bad actors, and we essentially need a United Nations of AI. Somebody who's looking out for the collective good and where everybody is- Yeah ...

making shared commitments to guardrails and oversight that we all commit to. I love that idea. I agree with you 100%.

We've talked about that concept, Barbara, yeah. That's absolutely- But we can't make this United Nation work. What makes you think we're going to make the next United Nations work?

So this one's going to be different, Alan. This is going to be different this time. It's going to be different this time.

It's different. So how far are we willing to go here? Alan, are you essentially calling for nationalization of AI infrastructure companies?

I think you're going to have to, yeah. It's indispensable. It's a strategic asset, and if it's really going to determine the next century, yeah.

Mike, Alan's wrong. No, it's not nationalization, it's regulation. Oh, can I throw a cynical hat about this?

Good. So Anthropic wants to go public. So does OpenAI.

And I'm thinking about the juxtaposition between Anthropic standing up to the government and the Pentagon, and then this agreement, this accord. So in exchange for lifting the export ban, Anthropic has now deeply integrated its roadmap with Washington. So they're granting federal agencies pre-release access to future systems.

They're sharing- The roadmap ... data, and they're engaging in joint safety research. And come on.

But let me hear Robert. He vehemently disagrees with me. I think we're going to hear what happened on the third day of Economics 101.

No. Mike asked you if you're talking about nationalization, and you said yes. I disagree with you.

Either I don't think you meant that, I think you meant regulation. I said it tongue in cheek. Right.

Yeah. We don't want ownership of this, we just want regulation. Oh, no.

In return for capping profits, we're going to regulate you, but we're going to guarantee monopoly. Well, you don't think we could do it more efficiently, Robert? Always.

I'm a DevOps guy, of course. I'm all about efficiency. Uh-huh.

I don't know. But you know what? We're almost out of time here.

I wanted to just take a quick wrap-up. What's cutting through today's discussions on all three segments, and it cut through most every day's discussions here lately on The Gang, is that AI is pushing us, pulling us to places we're not comfortable quite yet with how it should work. Right.

But the thing is, it's not waiting for us to get comfortable. So it's going to be a little uncomfortable. Strap in.

But that being said, let's end things that, hey, America's 250 years old this weekend. If you went back and asked the Founding Fathers, Ben Franklin and the rest, Tom Jefferson, George Washington, Monroe, Madison. If they thought we would be here at 250, still with the same Constitution for the most part, other than the original 10 Bill of Rights, we've probably only done- A few minor changes.

Yeah. Maybe 18, 15, 18 amendments. Would they have thought so?

I don't know. It's still a grand experiment, and we've got a lot to celebrate and a lot that we need to do. It's far from perfect.

But I wish you all a happy, happy 4th of July, and let's come back next week, because we've got work to do. Mike, John, it's always a pleasure having you on here all year round. Barbara and Art, too, thank you for coming on when you do.

It's Gang members like you that give us our zest and make things interesting, so thank you. Thank you for watching, most of all. Hey, you can watch us on Techstrong TV, Techstrong TV YouTube channel, or the Techstrong OTT app on any kind of screen that you like to watch your videos on.

But until Monday, have a great holiday, everyone. We're out.