Google’s $40B Anthropic Pledge, Claude Code’s Quality Conundrum & AI Coding’s Pricing Reckoning

Happy Monday and welcome to Tectron Gang, where we cover all things AI, security, DevOps, IT, cloud, you name it. And it's a crazy world to cover. Just this morning, I just mentioned to the gang in the green room that OpenAI and Microsoft revamped their partnership in what is a truce, essentially.

And we're going to probably talk about that a lot more later this week. But for now, we're going to focus on what's happening now, and there's a lot going on. So let me introduce you to our sterling panel.

Let's start with Garima Bajpai, who I have not seen in a while. Hi, Garima. You're out in Canada, is that correct still?

Yes. Where in Canada? I'm in the capital town, Ottawa.

Ottawa. Yeah. Nice.

Well, it's good to see you. I'm happy to see you on the gang again. We also have Chris Blask, who I believe also...

No, where are you, Chris, today? Hamilton, Ontario, the Greater Toronto area. Yes, also in Canada.

Nice. Okay. And last, but certainly not least, is Stephen Foskett, who I believe you're in Cleveland usually, but you travel a lot, so I'm not sure where you are today.

I'm in a gray, nameless void. No, actually, I am in Cleveland. We can't quite see Canada from here, but if I drive a little north on 90, I can see Canada.

Oh, nice. Very nice. Hey, before we start-- Yes, go ahead, Stephen.

Yeah, I think we were just going to say the same thing. Actually, I'm going to be seeing Garima soon at AI Field Day, which folks will be able to watch right here on the Tectron TV channels, wherever it is that you watch this. That's going to be May 13th through 15th, and we'll be in sunny San Jose for that one.

So I look forward to seeing you there. I was born and raised in San Jose. Wow.

Haven't been there since, though. But hey, I'm going to mention, we mentioned earlier this kind of wacky news cycle. It's almost daily.

I mentioned OpenAI and Microsoft, and late Friday, there was another development. Google and Anthropic dropped some significant news of their own. Google pledged to invest up to $40 billion in Anthropic.

Now, it's for all intents and purposes, it's a $10 billion cash infusion, and the remaining $30 billion is contingent upon Anthropic hitting certain specific performance milestones. The reason why this is significant is because a few days earlier, Amazon promised to plow as much as $25 billion in Anthropic. So there's this great land rush, and starting with Chris, I wanted to ask you this because I'm almost of this thought that these deals, specifically the Google Anthropic deal, do these signal the end of the model war and the beginning of an infrastructure war?

Yes. It's funny, a conversation I got off just before this one. The models are fine.

Models aren't everything. We are all similar human models, but depending on the training and the context and the harness and everything else. So I was going to open up with exactly that.

Yeah, this is it. This is going infrastructure. You have to have this supply chain in place.

Google is Google. Stephen, we were talking in the green room about this, right? This is what happens.

What was the previous condition? Early-stage startup. This is where you start talking infrastructure.

The agreements you had in place before will get modified, perhaps. But this is the big game. Yeah.

In that vein, I was going to shift this to perhaps Stephen. So if you're a developer or a business owner, does this news make you more or less likely to build your future on Claude, knowing it's backed by the two biggest cloud providers on Earth? Well, I think that there's a lot of, let's say, shifting sands here.

Early on in the AI rush, essentially these startups, OpenAI and Anthropic and the rest, really desperately needed partners like Microsoft and Google and Meta and so on. But now I think that they've grown, or at least on paper have grown, to such an extent that they're starting to reevaluate those relationships. Frankly, I wouldn't really be all that affected if I was making a business decision about which cloud provider to use or which AI provider to use based on these sorts of partnerships, because it feels like gorillas dancing more than it does gorillas really partnering.

I think that they're just all jockeying for position. They're trying to decide how revenue share flows. They're trying to decide where and when they can run and which models they can use.

And frankly, the end result, I think, is going to be pretty much the same from the perspective of the users. Right. Right.

I also wanted to add something here because what Stephen mentioned, it's spot on from a strategic perspective. When we talk about, let's say, a practitioner's view on this, and I have to bring this to this forum, because what we see from a practitioner's standpoint is two major opportunities here. One is that AI is breaking all the cloud assumptions.

So what it means is that how cloud providers can pivot to the next steps or next generation, cloud agnostic, model agnostic, AI native stack is in the question. Now, it is not who is the fastest, but who is more agnostic, right? So how the AI native stack would fold in a couple of years, there can be a Google AI native factory, which is more for regulated sectors, which are more trustworthy in one way, right?

And then you have people like Azure who are more enterprise savvy, moresignificant for the desktop users, for example. So there is a niche segmentation of these capabilities which people will tap in. And then when you see companies like Anthropic, they will also create this kind of AI native hub, which is an opportunity for them because they will become infrastructure agnostic.

So there's a lot happening in this space. I'm keeping an eye on this because it has several pivoting opportunities as we go along. Now, there is one other silver lining which I have started to see.

I'm not a very big fan of platform engineering from the beginning, but now I can see this, that platform engineering and there is a strong opportunity with this AI ops capability in making how the traffic routing will happen, how model abstraction could bring in more value, observability, legal ops, data ops, all that in a mix. I think this is a great opportunity for platform engineering people to see how they can be part of this AI native, neutral, and agnostic stack in the longer run. So I, from a practitioner standpoint, great opportunities in making, is just that we have to think about agility and bring more experimentation on the ground.

Yeah, that's absolutely true, and I definitely see, like you're saying, Garima, that there's different emerging leaders for different people in different spaces. And so Anthropic is 100% going for the enterprise market here. That's what they want.

And then another thing I would love to, another part of the onion that I would peel back is this question of what really are these investments anyway? So Google's giving Anthropic $10 billion for an equity stake with a promise of $30 billion more. But as it says right here, the expectation is that that'll all flow right back into Google for use of hardware and data center, and we've seen that again and again, where these companies are essentially, if you zero out the dollars on both sides of the equation, what we're getting is AI companies basically giving away part of their equity in exchange for usage credits so that they can run the models.

And that's a real different message than Google is investing real dollars in Anthropic. At least it sounds like a different message. Yeah.

It's also hedging its own model, right? So think about this. It's putting a neutral stake on all this model opportunity and putting money back into Anthropic with Gemini on the other end, and also opening up doors for other sectors to say that you are not vendor locked in, right?

Right. But Steven, so you mentioned, it's interesting. In the vein of what you were talking about, this arrangement, critics refer to these as circular investments, basically.

So Google gives Anthropic money, and Anthropic immediately hands it back to buy Google Cloud services. And in the bigger picture, is this a legitimate business model, or is this an accounting trick to inflate cloud revenue? I'm just being a little bit cynical here, but do you have any thoughts on that?

It sounds like an accounting trick to me. I don't know. Just when I see these things, I wonder.

It's almost like play money at times. And maybe I'll go to this part. Here's something I'll throw out.

Maybe Chris or anyone can answer this. So Amazon and Google both are pouring tens of billions of dollars into Anthropic. So that begs the question, who actually owns the relationship, and does Anthropic risk becoming a puppet of two different masters?

Well, let me take this one. So I spent most of the last decade looking at supply chain security, software bills material and so forth. And that has developed to the point that now every Monday morning, there's a coffee talk with a dozen or so interesting cats, and we're trying to figure out where this is going.

And this morning, we're talking about AI BOM. The AI bill of materials. Mm-hmm.

And a couple of years ago, when I first attended one of those meetings, I was thinking like a software bill of material, I have this code, this individual file, and I have a contents for that, and I can see that. An AI BOM a couple of years ago, we're thinking, okay, to earlier comments in this show, is it the model and the harness and the so forth? It's like, wait, hold my beer.

Pour my coffee, in my case, because everything you just said, John, and everything we're talking about here. So where exactly is any of this? How would I account for the provenance and the context of the output, this thing that I'm using for critical infrastructure or high-risk compliance, legal information systems, when I can't even answer these questions?

Who actually even owns the infrastructure that I'm using? Where is it from? What's happening to it?

What are the intents, motivations? Is it a circular relationship where I really shouldn't trust that infrastructure because it's not a real investment? If we can't enunciate that, how do I make business decisions based on output from these systems?

Yeah. It's interesting. With this deal, and I'm going to say anywhere from $10 to $40 billion.

Anthropic's valuation now is $350 billion. That's the equivalent of the combined market cap of Disney and Pfizer, and it also percolates this idea with this kind of valuation, is that sustainable for a company that's still scaling, or are we in the middle of this AI bubble? I'm going to bring it up again, even though we've brought it up the last two years, but it just still begs that comparison or that thought.

Well, AOL Time Warner, right? AOL obviously was a persistent brand down to this day, providing value in the free diskette space. I don't know.

Yeah, but let's accept this, that the models are not the bottleneck. The bottleneck is compute. I'm surprised you picked Disney and Pfizer.

We keep hearing that AI is going to revolutionize drug discovery and make movie-making obsolete. So maybe it's completely correct that it's- Yeah, I only used those two because it added up to $350 billion. It keeps not happening, but we keep hearing that that's what AI is for, right?

Yeah. Sure. Yeah.

No, absolutely. It'll be interesting to see what happens. Today, I just mentioned earlier, OpenAI and Microsoft kind of revamped their relationship as kind of a truce, and there's also a report resurfacing that OpenAI is going to be working on some sort of phone.

I don't know if that's kind of the next iteration of the rumor of them with a device. I don't know. So it's- Can we make a segment out of it?

Yeah, we could. Probably tomorrow or Wednesday. Yeah, until something else bumps it out of the news cycle.

In any event, we're going to move on to topic B, which is trouble with AI code, and we're going to start with Garima. So GitHub has suspended some new sign-ups for several of its Copilot subscription tiers, a decision that follows a surge in demand driven by agentic coding workflows, which consume far more compute resources than earlier models of AI assistance. So we're referencing this, and I kind of wanted to ask Garima, and maybe it's a bad comparison, but I will throw it out there nonetheless.

And is this kind of like a Netflix moment? Where Netflix used to be cheap and shared by everyone until the costs of the content and the infrastructure forced them to crack down into the new subscription model where you couldn't share the same account. And I'm wondering, is GitHub just the first of many AI tools like Cursor, Harvey, or Midjourney to hit this wall?

I think it's not a unfair comparison, but I'll come to that point. Hmm. The first thing which I mentioned in the first segment as well, that it is evident that the bottleneck is compute, right?

So that's the pivoting point of everything, what you see around all these changes. Now, this article especially, which we are referring to, refers to cloud agent sessions, and these sessions are costing more than user monthly subscriptions because the sessions prolong, right? And the context is too kind of stretched out.

So that's the problem here, and that's for breaking that, the economics are not looking good and that's the reason why we see these kind of shifts. Now, this is again, you go deeper into this and see why this is happening and how competitive assessment of, let's say, we actually did a competitive assessment with a few folks and what we did was we actually looked at developer productivity and specifically in this case, one developer who is in low-cost location and comparing it to a coding assistant where we are leveraging a coding assistant or a coding buddy and focusing on cost and delivering outcome. So we compared these two things and we found that in several cases, the direct cost was broadly comparable from a human developer standpoint and consistently generating higher value was the key theme.

So when you employ a human developer in a low-cost country, it still is delivering higher outcome than you have a coding subscription, which is low cost, but it cannot deliver that value. So the economics are screwed up, the outcomes are screwed up, and now what we are seeing is that with this session prolongings and the economics being not matched, I think it sounds interesting how this will unfold in the future. Because there can be eat-all-tokens-one subscription model, which you can think about, but I don't think so this would be a viable option for the next few months and maybe few years as well, because it does not make any economic sense.

The second scenario will be that enterprises will pivot to a developer-pay-per-burn model, which is true usage-based, which can be an option where you are a conservative company who's looking at the economics, and economics should match to your what expectation you have from these coding buddies and these AI partners and this whole kind of bubble, which is in making. But again, from engineering point of view, if you want to measure developers per burn rate or true usage-based model, then you will have to have strong telemetry in the system, strong cost allocation mechanisms, and team has to be comfortable with per token economics, right? And the third model can be an hybrid option where you have tiered subscription models, where you have a base pricing for something which is very elementary, which is auto commits, for example.

So they don't allow long sessions, for example. Right? So in memory context and all that.

So it will be auto commit cases, and then you have pro models, which are tier two and tier three, which will be more session-oriented and context-driven, and they need to be kind of also be very sensitive in terms of pricing. So there are a lot of things in making here. Right?

And the second point is that economics are not matching up. People are creating hype of these coding buddies will replace developers, and there will be a lot of substantial de-skilling in the developer ecosystem. So we'll have to be watching out this very carefully.

Yeah. Stephen. Oh, no.

Go ahead. I think, if you're not part of this industry, and Garima's exactly right. And if you couldn't follow all that along, the simple takeaway is that this is a little indicator where we are in the industry.

I've been through this before with various security products, firewalls and SIMs and whatnot. And at the beginning, you say: "How do we monetize this? What do we charge for?

" And if it's successful and it becomes endemic and everybody's doing it, that's sort of where we are right now, then that may or may not have been a good idea. I think the last segment and this segment, we're coming back to the same thing. This is an- Yes ...

emergent thing. The actual, what we're paying for and what it costs, the people behind it, these companies are a couple years old, the Googles and the big ones, again, they're just folks going along, like the rest of us, watching shows like this, trying to figure things out. And yeah, the entire economic stack may make no sense whatsoever because now we're doing this.

Because the obvious thing on this one is we didn't think that AI would code that fast. It's like, "Oh, yeah. " So maybe the first round of pricing just didn't make any sense to begin with.

Yeah. Please, go ahead. So there's another aspect, yeah, in here too, Chris, and to your point there, and Garima, 100%.

It reminds me so much of the FinOps conversations we had, I don't know, last year, year before on Techstrong Gang. There's another aspect here too, in that you're letting your vendor essentially twiddle the pricing behind the scenes while you're using it. And that is absolute madness.

Terrifying. Can you imagine if Ford could make your car get half the miles per gallon because they felt like it, or if they could decide that the new software update that your car is using uses three times as much gas. People would be in the street with pitchforks.

And yet that's pretty much literally what's happening here. 7, token usage. 6.

But what I'm hearing as well is that it's actually much larger than that in practice. In fact, they're seeing as much as 3X token usage for this model that doesn't actually generate much better output than the previous one. And the reason is because of some of the knobs that are being turned behind the scenes.

So for example, to the point of Adrian Bridgwater's article here that we're going to be linking to, there, what we found was Anthropic had decided to throttle down the level of thinking or reasoning from high to medium in order to give a faster response to users. I know that because I've been using these models, and I'm one of those people who likes to drop down the box and watch the reasoning happening, and it's pretty cool to see what it does. It's actually fun to watch it reason.

But when it takes 10 seconds, 20 seconds, 30 seconds before it even gives you any kind of response, if you didn't have that drop-down, and if you weren't watching it, you might be like, "What the heck is this thing doing? " And at the same time, they can adjust things like that in the background. To your point, Garima, if companies are going to usage-based models, what does that do if the vendor can then say, "Oh, really?

You're going to pay per token. Watch this. Now you're going to use three times as much tokens.

Have fun with that"? Well, they can literally do that right now with these models, and they are. And those of us who are using them and trying them and experimenting with them and what we're finding is that we run out of tokens real quick.

And since these things are so far underwater, I would not be surprised to see these vendors... And it's not like some evil conspiracy or something. They've got to turn those knobs up because they've got to make money, or why are they doing all this?

I guess that's a bigger question. So we're seeing this movement or move towards token-based billing or, I guess, metered usage. So maybe, Garima, will developers accept a world where every save or refactor has a literal dollar cost attached to it in real-time?

I think we can say that this is, from a developer standpoint, it doesn't matter, to be honest. It's your CFO strategy. So how much you are willing to pay for things which can be done in a cheaper way, and with a better outcome.

And mind you that a lot of these companies who are advocating for developers to be out of the door because now we have these coding buddies who can do the job will start to rehire in 2027. This is my wishful thinking that this hype would be going down a little bit because now the code is in production. Now we are seeing the implications of it.

So a few days back, we talked about this license laundering, how open source licenses are being forged, and we are taking code from there. The third aspect is that the code which is being written is not understood by developers so much because now the coding buddies are writing the code, so they're not commenting in the code. The ownership of the code is lacking.

And the last part is de-skilling, right? I am very concerned about this because if, let's say, we continue this movement without any thoughtful strategy on how we build the pipeline of future developers, I think this de-skilling will be a big problem for us down the line. So I feel that there is no one answer to it.

Maybe these auto commits for simple jobs would be remaining token-based or something like that. But in the longer run, we see a hybrid tiered model where we can have both options, and maybe companies become more sensitive and mature to see how they adopt. Maybe they also question back on how the models are generating the code, like what monitoring measurement or what SLAs they can put onto the model itself.

Because now, how many rollbacks you do for a code which is generated is not measured or monitored anywhere, or how many vulnerabilities you are introducing. So those are also the kind of matrices I probably... I'm very hopeful companies or communities like Dora would come up with some kind of intelligence around this, right?

So there is a lot more to be done, John, on this. Sure. Yeah.

It's interesting, Chris, I like the idea that you tied this into the first segment because in a sense, if GitHub is struggling with the strain, I wonder what this means for this, quote unquote, $40 billion investment Google just made in Anthropic. Is that money just going to be vaporized by the sheer cost of running these models? Just- Is yes too short an answer for our purpose here?

It is. I want you to overflow me with your brilliance and beyond the "Yes, yeah," if you have any thoughts. Well, yeah.

And again, yeah, but to be clear, I don't particularly know. That's down in the details, those accountants, lawyers, and again, I am fairly infamous these days for talking about local AI, sovereign systems, building it all from the ground up, right? And it's not just because centralization has all these risks and so forth, it's just that the answers end up being opaque.

How do we know we're building systems? And as a security person, and we're doing things this way these days, right? And putting them out there in the world, and again, it's not the same as saying that massive global enterprise systems and global systems and so forth will work out that way, but it seems to, right?

Individual computers, the amount of information you would need, any one of the four of us who are companies or whatnot, is tiny compared with all the information on the Earth. And when we try to say the cloud or the Googles or whatnot are going to have the infrastructure that we all work in, and you try to work back to, "I am a retailer, manufacturer, whatnot, out in the world, a NGO, a public sector entity, and I have these commitments and responsibilities that, again, aren't every regulation on Earth, but they're mine, and I need to demonstrate diligence and responsibility, and I don't know where any of my data is or what's happening. " Look, short of comet strikes and so forth, evolution doesn't get a lot punctuated.

Big infrastructure and so forth is necessary. It's not going anywhere. These companies will be generally fine.

But the fact that every week, we get to come on, I get to do this every week and come back and say, looking around, yep, yep, still doesn't make a lot of sense, not in the long term. That's why we have things to talk about here, right? Yeah, absolutely.

Right. And it just keeps shifting. It's like shifting sand every other day.

Once you think you have a grasp on what's going on, then something else hits you on the side of the head. So it's fascinating, it's dizzying, and it's kind of overwhelming, but that's what makes this show and what makes this particular period in our industry's history so compelling to me, is just so unpredictable and moving so quickly. And that's what makes these types of shows interesting, I think.

Finally, last but not final, we're going to talk about Qlik Connect Field Day. And now I know you were there, Steven. Can you please provide us with the field report?

And the floor is now yours. All right. Excellent.

So the cool thing, as Chris was just saying, the cool thing about all of this is that we get to really dive in and consider and learn from each other, and that's really what I look for when I go to events. That's what I look for when I come on shows like this, when I meet with people like you. And I feel like there's so much to be learned here.

And it was really illuminating to go to the Qlik event. Now, we didn't do Tech Field Day presentations there. Your eyes aren't fooling you.

There's no hidden secret presentations. We did what we call a Tech Field Day Experience, which is where essentially I bring a group of folks like, well, us, to an event, and we attend the event. And we learn, and we attend the keynote, and we have briefings, and we have meetings, and we talk to the attendees.

And then we do essentially what we do, whether that's writing on LinkedIn, which I did quite a lot ofOr recording roundtable discussions, which we did and published, or podcasts, which we did and published, or interviews or whatever it is that we do. And so I think what you're going to find there is that you watching those things are kind of having a similar experience to me at the event, which is essentially my goal is to learn from other people. And so I got to sit down with really sharp people from Qlik, like Mary Kern, who knows more about the data industry than I certainly ever will.

And it was really an illumination when she started to explain how, both on camera and off, how data and analytics has informed the perception of data people about AI. They see it very differently than AI people or infrastructure nerds because, of course, in the data space, they have been focused on data quality forever. They have been worried about not just sort of in general terms, but in very objective terms, is this data trustworthy?

Where did it come from? Is it described properly? And not an on/off switch either.

I love that too. They have a whole data quality score, and they're looking at it sort of on a percentage basis. Like is this 80%?

Is this 85, 90, 95%? And the way that companies like Qlik in the analytics space are leveraging AI is they're incorporating that concept of a data quality score when making decisions. So for example, they were saying their agentic integration with ServiceNow, they would only have it act autonomously if the data quality was maybe over 95%.

In other words, if we're pretty confident that it has the right information to work with. And I was just like light coming down from the ceiling hearing this because the problem with so many of these things is that the input data is garbage. Garbage in, garbage out.

And then, another thing that we talked about, I talked to Sydney Drill, who is focused on FinOps for Qlik and helping customers navigate FinOps around AI applications. " And it was really illuminating there too because, again, these people come from a background of having worked in the data industry and having considered things that we're just now considering. And so it's so incredible to basically step away from the same echo chamber and into, I guess, a different echo chamber and hear the echoes that they're making about this stuff.

So, Chris and Garima, I know that you all have sometimes worked with folks in the data space. Does what I'm saying make sense to you? What other areas should we be learning from when it comes to data quality, and thus AI quality?

Yeah, I think Chris has more to say on this, but I will just say a few lines- Yeah, he's like, "Mm" ... and then give it to Chris because I think technology has never solved any problem alone, right? So if being AI or being data or being a developer, I think we will not be able to do full justice to this whole ecosystem because we are in the S curve of technology.

The technology itself is evolving and there are multiple S curves, right? So the data engineering part is kind of maturing, then the AI part is maturing. The developer ecosystem is also kind of taking a leap of faith, right?

So I think where we are missing the point here is that we probably are not focusing on the most important part of this, which is people, right? Why do we have technology? We have technology for the people, right?

And I do believe that a lot is happening, but one thing is that there is a lot of irrelevant expectations on leaders, on senior managers, on people themselves, practitioners themselves. And we have to pay close attention to this because fatigue is real. Every time we come to this show with 10 things happening in the background, and whenever we end the show, everything we talked about becomes like a legacy, right?

So I think we need to kind of be very careful on how we manage this from a people point of view, from a process point of view, from an ecosystem point of view, and that is what I think you are alluding to, Stephen, that we already have these movements like the data foundation movement. We have the FinOps movement. We need to look closely on how to connect the dots in the ecosystem.

Yeah. Context is everything. So on LinkedIn today, I posted an article because I just do that a lot.

I've been trying to break down all of civilization into seven major version numbers. 1 right now. Right?

From the time we started living in groups more than 50 people and trying to get information moving, something other than first-person. And it's interesting, we think we're all... Everybody on the screen here is old enough to remember things when they were a little less, right?

" No. Right? Wikipedia looks like this huge data set.

It's an amazing thing that pretty well describes the Earth from my perspective because I'm of the cohort that built it, but you rotate the perspective a little bit and it's empty, right? Most information that actually matters to anybody is not on the internet. It's not in Google.

It's not in the cloud. It is hyper local. It's because right now I am whoever I am.

I'm a company. I'm a person. I'm going through the world doing things.

There's a lot of data that's happening right now that just isn't out there. And so as I look at this, like in the last couple segments, compute and infrastructure and so forth, it's out there, and a lot of this AI stuff isis hardware heavy, and we do a lot of work in developing countries with small computers in the first place. But you find that all the data, all the information you need to do what you're doing is already right there.

And if you need to access some external resource, if you can define what that is, why you're doing it, in your context, and use it, then you can live with that inside your own space, your own canon, and build up from there. And Steven, this event, I have to look at it more. We have a team doing a Hackathon Agency 2026 this week, and just looking at the same things as, from our perspective, it's all about provenance and context, and if you have that, pick the scenario, you can work your way through it.

And if you don't, you can rent the answer from someone who thinks they have enough information about you to tell you who you are in the first place, and maybe that works, and if it's low risk, but it's not, right? This is an evolutionary crux, where all the things us talking heads talk about on shows like this is defining spots in the space that in retrospect, I mean, already in retrospect. John, when was it, Alan and the folks, this show has been running a couple of years now.

This conversation on AI, looking back at it, there was no, wrong, wrong. Oh, wrong, but in interesting ways, and now we're talking about something based on that. So what are we wrong about today?

I can't wait to find out. Well, that's a great way to end today's show. I love the historic perspective you bring, Chris, and how you think about these things in the much higher, 37,000 feet perspective.

Yeah, it's interesting. What we thought was ridiculous six months ago is actually happening now or far beyond or behind it. So, again, I think Alan has always said this, and I think we all agree that there's never been a more vital time going on in AI and in terms of the tech industry as there is now.

It's just overwhelming, but the possibilities, I think, overwhelmingly are great. And that's why we thank you for watching the show. I want to thank Chris, Steven, Garima, and all of you watching.

We got more programming on "Techstrong TV" later today. And so until next time, we'll see you tomorrow. Same time, same website, same channel.

And we'll see you later. Bye.