Techstrong Gang – September 13, 2024

Hello, everybody. NVIDIA's, CEO says that some of the conversations with customers about GPU scarcity is getting a little tense. Then we have Serge Brand is saying, Hey, developers need to make better and deeper use of Gen ai.

And finally, well, there seems to be a disconnect between what's going on in big tech and how Washington perceives that. And maybe both sides just don't understand each other. I'm your host, Mike Ard, and you're watching Techstrong.

Hey, everybody. We're back and we're gonna get started with what's going on with Nvidia because, well, it's one of those head scratching moments for me as I kind of look into this whole conversation about what's going on with GPU scarcity. When you look at the 10 k, that Nvidia file last May, they send two customers account for just under a third of all their revenue.

And there's some analyst reports outta Wall Street saying that maybe a handful of customers actually accounted for 40%, which kind of suggests that there's a lot of companies out there who are interested in GPUs, and maybe the only way they can get 'em is through a cloud service provider. And, and that has some challenges in terms of, uh, where my data is. And it also has some issues in terms of cost because, well, it's an expensive way to get to A GPU.

John, let's start with you. What is going on with Nvidia here? There seems to be some folks who are saying the company is overvalued on Wall Street, and how can we have an AI revolution if we can't get GPUs?

Yeah, that's gonna, that's a bit of a problem, right? This is ongoing. GPU shortage problem has been going on for years, which has went to engineering terms like optimizations, smaller model size, et cetera.

We've got investors betting hundreds of millions of dollars on startups and software helps companies make due with the GPUs they've got. Meanwhile, we've got nvidia, which is wildly fluctuating in terms of its stock price. Uh, it's also under the, the guise now, the government was just looking at it and its, its power or its influence.

I know Futurum Intelligence did a report on the, the power of, and the, not just the expansion, but just the, the reach of Nvidia, more than 90% of the market. So it, it's, it's, it's at the focus, and you're right, Mike. Uh, how do we have an AI revolution when we don't have enough parts?

And we have a shortage from one of the most dominant companies in the market, which is catering to, I think you said a handful of companies account for half of its business. So we definitely have a log jam here that needs to be broken, and I just don't know where it's gonna go. And, and then the one thing, I think I mentioned this before, but Nvidia under being under investigation is kind of weird.

They're being, uh, in a sense, punished for their success because they got into the market early. And, and that also plays a part in the AI revolution, right? You're going after a company before the market fully develops, just because they were first to market and they, they took advantage of the, the sleepiness of Intel and, and others.

So we're at a kind of very, very fraught time right now, My question though is aren't there other companies, why are we just so focused on Nvidia? Yes, they're the main one, but if there's a log jam there, I mean, there's other companies, um, like a MD and Google and yeah, That's, can I Add several other companies offering high powered chips? Yeah.

That's something I feel to say is A A MD, right? They've said that they embrace open source to tackle GP shortages, so you That's right. And then there are other options.

So, Well, let's get into those other options 'cause they're not quite as available as we'd like them to be. So the issue isn't so much that these, uh, Nvidia can't make GPUs because they're contracting out to, to a company in Taiwan, and the company in Taiwan can't produce enough GPUs. And guess what?

The same people who were trying to buy GPUs from Nvidia are also the same people trying to buy GPUs from that company for everybody else. So it seems like the constraint is this manufacturing line in Taiwan. Um, the other issue though is not all those alternative choices work equally well, if you go talk to some folks about, uh, the training of the AI models using some of the alternative quote unquote accelerators that are out there, they will tell you that depending on the model, the accuracy rate drops dramatically when you're not using GPUs.

And so they sit there and they say, well, you know, we might use some of this stuff for some of these other models, but there's a trade off there in terms of usage, in terms of accuracy and well, gen AI already has a image problem with accuracy. So they're not exactly jumping up and down about using something that is less expensive, but less accurate. And you can't get stuff from a MD and Lisa, I'd love to get your opinion here.

Are we kind of on the cusp of maybe a marketing nightmare because we are creating this expectation of demand and we are not able to fulfill much of it, and somebody's gonna wake up one of these days and say, all these valuations of companies that investors are asking about related to AI are gonna have a, uh, less than positive outcome. Well, what investors are wanting to see, Mike is the ROI. Where is that?

In fact, Jensen Wong was recently acknowledging that tension with customers. You guys, both you and, uh, John mentioned, I think I read four companies constitute nearly half of NVIDIA's revenue. So he's acknowledging there's tension there with those companies.

We know NVIDIA's biggest customers include some of the Mag seven, uh, Microsoft, Google, Amazon, meta Tesla. We also are seeing, are we at a buy versus make kind of conundrum where we know that, um, Amazon Meta Tesla are investing in their own silicon, um, not necessarily to compete with Nvidia, but they are. And so from a, from a PR perspective, I think what Jensen Wong is doing is the right thing.

I'm sure what's going on behind closed doors with those top four customers is a lot of, uh, discussions, concessions. We know we're behind here. The, the demand is so high.

Um, so I, I think I, I don't know that it's a marketing problem. I think it's probably at Jensen Wong's level, and I, I presume that's probably what's going on in order to asage concerns, um, because we saw the, their announcement there earnings just a couple of weeks ago and showing how high the demand is still, and we know how much its top customers have been investing. So I think it's gotta come from the top.

Paul, I don't know if you wanna weigh in here a little bit, but, um, our developers sitting on their hands now waiting for, you know, something to work with. Yeah, Mike, I, you know, I, I, I like the way the conversation's going, but I'm gonna throw a different perspective out here. I, I think that it makes me a little bit nervous as, as, as an investor, but also just looking at the market overall.

When I look at what's going on with the, uh, you know, the hyperinflation of, of, uh, Nvidia right now, um, you know, one of the things that comes to mind, especially from the business logic side and the developer side, is education on how to execute and use these, these technology stacks that they're putting in place. One of the things that, um, that I often hear, and you know, when you get like roll up your sleeves and get into the nuts and bolts of things, is there's a demand for organizations to go out there and get these GPUs. But the re reality of it is, is many of the applications are not using the GPU.

They don't even need them. They, they can use CPUs and they can use and, and, and, you know, and conversely, they, there's a, there's a, a need for TPU and different, different areas across the, you know, the board. So the education of understanding how developers and business logic is using these, uh, this tech stack really has to come into play here.

I I, I want to believe that this will normalize, uh, soon and in, in the, uh, the, the, you know, technologists will get an, a full understanding of where their business logic is going to intersect with the demand for these, uh, the, this, the, the tech stack that goes along with it. Uh, I know I've been, uh, you know, having these conversations about the CPU utilization that, uh, GPU utilization and TP utilization, and frankly, um, I, I don't know if there's a, a, a, you know, a an immediate need to go directly to A GPU, especially if you're just getting started. I think there's trade offs there that people don't know how to sort through.

Because a lot of the benchmarks that are being surfaced for tracking the performance of AI models are borderline worthless. And it's not that the work that went into that didn't have merit, but each of these models is so different that it's really impossible. Your mileage varies so widely from one platform to the next.

Absolutely. And over-provisioned as well. And that's the thing that, um, needs to be taken into consideration, harmonized.

Yeah. And John, I wanna come back to you about something that Lisa said. 'cause she made an awesome point about, hey, if I'm AWS or Microsoft, they already are out designing their own chips for everything else, so why not A GPU?

And, uh, and I know Tesla, which is as GPUs in almost every car they make has the same idea in its mind. They're basically going, do we really need somebody in between us and Right, the Fed, Right? I mean, I could conceivably see that happening, you know, that it, it's, um, just kind of going back at the bigger picture, and it's something that we've all touched on, is that we consistently see the same storyline or narrative with ai.

Billions of dollars are being thrown at this. In fact, I think o Open AI is on the verge of getting yet even more massive funding that will push its market value up to $150 billion. We saw safe superin intelligence get a billion dollars.

So there's all this money pouring into here. We're getting all this hype. We're, we're focusing on just a handful of companies that at, at this point, allegedly hold all the hold, all the power, and yet we have shortages.

We have over, over baked expectations. We have customers expecting immediate results, especially from the top down. We don't, we have lack of training.

It's just, it, it's kind of, to me, it's like a massive traffic jam. Like there's that, there's 10 lanes trying to get off the same exit. And I, it just, inevitably it's gonna create, as you said, um, maybe other players emerging to gain some sort of independence, get outside of the reach of the NVIDIAs of, of the world right now.

I think it's really fascinating. We're really early, as we've always said, but things gonna have to start happening. We, we can't keep talking about this.

ai for us. Um, do we need to have an adult conversation with the business leaders about what's real about AI and kind of set some expectations? Yeah, I think, I mean, as we've talked many times, I think there's a little bit too much faith and expectations being put in AI right now.

There are many great use cases, but I think we need to take a little bit more time determining where we really need to incorporate ai, ai. And maybe that would slow down this need for all these chips. Mm-Hmm.

Lisa, uh, I think John mentioned that there are investigations underway, and I am not a, uh, SEC executive, but do you think maybe we should have some caution about what we're saying in our marketing literature versus what the reality might be before somebody stands up and say, um, you know, was that a misleading statement or not? Absolutely. I think marketing needs to be transparent, and it can absolutely aid here.

Paul was talking about the education piece. I think that's critical. I think one of the things too, that marketing can facilitate is those ROI stories.

They have to be out there with how much John mentioned, the billions being spent on AI hardware and software. There have to be ROI stories out there. So what Nvidia marketing can do is start working with those customers, work with the Amazons and the metas, the Googles, to understand where are they seeing that return on investment so far, so that the proof in the pudding is there.

And to your point, Mike, it's authentic. It's not misleading it's truth. And that will could go a long way on the trust factor as we navigate this bottleneck that's clearly there.

Hey, Mike, can I jump in for a second? That's interesting what, what Lisa said, because I can't, I can't divulge where, where I heard this from, but it's very, very good source with an Nvidia. They are actually, uh, transforming their corporate blog from the, the, basically, here's our success stories, et cetera.

They are evolving that blog probably for next year or later this year, where they're gonna talk about the implications of their technology, not just, uh, within their ecosystem, but on, on the climate, but they're also going to be going deeper into these success stories and why you need to use the product. So they're gonna kind of, it'll be kind of a more mature, uh, approach rather than the classic blog. They're gonna be doing these, these kind of in-depth stories and looking at their product from various vertical markets and why it's successful and why they are, um, not breaking any rules, so to speak, and why they are, uh, the classic must use example for ai.

That's great to hear. I would say also that Nvidia is only one part of the problem. And I'll go to Paul from my next part of this.

Um, the way pricing works for these cloud services when I wanna invoke these GPUs, is essentially it's a token model, and I pay per token and I pay for the input and the output. So every interaction is two tokens. Well, that sounds reasonable, but then when you start adding up the hundreds of thousands and millions of interactions, the pricing for gen AI gets prohibitive quickly.

And that's one of the reasons why you see so many people trying to figure out, how do I run this stuff on premise? Because if I'm gonna put something in production environment, can't afford to pay AWS and Microsoft for every input and output, it's just not sustainable. Um, so do we need to maybe go back to the cloud service providers and say, Hey, if you want this business and you wanna run outta the the cloud, you gotta come up with a model that's not based on these tokens.

Yeah, I mean, it goes back to what I was kind of alluding to earlier in the conversation, is understanding how you're utilizing the tech stack to achieve your goals. If you can do, um, a lot of your learning models, if you can build your implementations or your in instantiations on-prem and not have to use the, the cloud models today, you can obviously, you'll save a, a quite a bit of money to kinda get around that, that, that structure. Um, I believe that the, there's, you know, I, I'm, I work in, in my space covering app dev.

I, I work with a lot of emerging companies, a lot of, uh, uh, you know, series A and kind of emerging companies. 70% of my business is emerging companies. So, but with that said, um, there's a lot of emerging companies that are coming up that are looking at ways to kind of build that token bridge, so to speak, and the way to, um, uh, do the processing offline and, and then produce it in the cloud.

So therefore you don't have to have that incurring cost. This is very similar to, uh, if you're familiar with API gateways and API, uh, transactions and such, every, every interaction on an API costs you money. This is very similar, uh, in that model structure.

So if you can build a bridge that you can do kind of in a cashed mode or in a way that's offline, you're going, you're gonna reduce the amount you incur on the cloud. Now, with all that said, in order to make it more affordable for organizations to get to have a lower cost of entry, there has to be alternative models, right? Uh, the model right now is, is it's very, very, uh, price intensive, as you mentioned, Mike.

You know, it's a, it's for every transaction. It's two, it's two tokens, right? So, uh, but that needs to be, uh, addressed pretty quickly.

All right? I am not a Wall Street investor, and I do not invest in those stocks because, well, that's a bit of a conflict of interest in my point of view. But John, um, your opinion here, is there a disconnect between what Wall Street is seeing and maybe bordering on the other side of irrational exuberance?

And, you know, I'm an IT guy and my IT reality sees something different. Yeah, no, you're, you're absolutely right. This is kind of a, this is a sensitive point, I think for me as well as for you is like, this Wall Street really makes no sense.

And, you know, and in fact, the way it reacted to NVIDIA's results was, was kind of puzzling up, but not surprising. Uh, whenever Apple makes an announcement, the stock goes down because they, there's all these, these, these, these over valuations, or underestimations, whatever, what you would wanna call it, wall Street Vex is from one extreme to the other. And that's what's so maddening, I guess, to a lot of the CEOs are under this 90 day shot clock, and they, they try to adhere, they try to appeal to the street, and they, the street will always find something, there'll always be a number with, with Nvidia is the ROI, right?

It's not good enough, or the margins, oh, they're, they're slightly down. Meanwhile, they're blowing out revenue numbers and then profitability. So, you know, it's just, it's, I with a grain of salt, so to speak, because Right, the reality world, the IT side versus the, the trading of, of, of stock are two different worlds to me.

Well, there you go. Well, when the stock goes down, that's the time to just buy more. Okay.

Because it's gonna go right back. Yeah. But it'll have nothing to do with the fundamentals.

I'm, we are not a stock picking firm here. We will never tell you what stocks to pick. But, um, I would just say if you're an investor, ask better questions.

We'll be back in a minute. All right, folks, we're back and we're talking about our second block for the day, and Sergio Brynn was out this week saying that Google developers are not making as much use of gen AI as they should be. And he showed them an example of some capabilities.

Um, I found it interesting. Um, but it's a bigger question in my mind, and I'm gonna launch to Paul first. Are developers just too conservative when it comes to these kinds of new and emerging technologies?

I mean, essentially Sergey Bri is chatting developers who work for Google, and that's the best among the best in class in the world. So, you know, what's the average developer thinking? Yeah, Mike, you know, I mean, I think it, uh, you know, when we look at these, these, uh, approaches for utilizing, um, you know, AI in your development cycles and such, and utilizing, uh, advanced agile kind of processes and methodologies, um, there's a clear distinction in my mind when, when a, when I do kind of overlay that, that that market challenge to our research, right?

Um, when I look and talk to developers, uh, there's a, there's a clear, uh, push for organizations to release code faster, okay? Uh, in a stu recent study that we've done, uh, we saw that over, and the study was over 800, uh, respondents. We saw that 24% of the respondents want to, uh, the business KPI is to release code on an hourly basis, yet only 8% are able to do so.

And part of that reason is because, um, they, they're, they're not using in this, the, the methodologies, the tech stacks and the, and, and everything available to them to release that code very rapidly. Now, when we look at, you know, overlay that data with the amount of applications that are being, are projected to be created, what we see over the next three years is there's going to be over a hundred percent of new applications being created with the same or fewer resources available today. That's what our re that's what our research shows.

Um, the only way in my mind that this will be attainable is to utilize, um, you know, co-pilots and AI in order to get you there. Now, bringing it back to today, Mike, um, I I think you hit on something very clear. Um, there's a maturity level here.

There's, uh, understanding what organizations are doing today, how to get, basically how to get through the day, right? 'cause the developers have to get through the day to release the code and get their, their day jobs done. But they also have to understand the new tech stack in order to confidently deliver what they want to, you know, what they, where they need to go.

Um, as I said on previous episodes, the accountability is going to lie with the organization, putting the code out the door. So developers don't wanna lose their jobs, they don't wanna lose their business reputation. They don't wanna lose credibility by putting out, um, you know, autogenerated code that's not going to work properly.

Um, you know, that, that there's a whole slew of angles this can go, which is security as one app option, uh, testing as another option in the ci Everything in the CICD pipeline needs to be taken into consideration. And we're trying as an, as a, as an industry, as a, as, as you know, an organization that trying to do this is to automate that as much as possible. But that maturity is, is, uh, in some areas are growing pretty fast, and, and a lot of areas, in most areas is going pretty slow.

Lisa, you'll notice soon that there's a theme starting to emerge here. And the next question is, so based on everything that Paul just said, I feel like maybe the AI companies have blown past the marketing motion that says, let me educate my customers how to use this stuff and, and maximize all this stuff. Because basically I feel like, you know, people are writing blog posts and having trouble parsing the word is and talking about things that are, is the way they'll be three years from now versus today, and we're not kind of doing the fundamentals of bringing developers along and end users.

Is there, is there something to miss here in your mind on the marketing side? You know, what struck me with surrogate brand's comments were how, if you juxtapose them to what Matt Garmin said, that, and that leaked business insider information from a fireside shed a couple months ago, where he's saying, I think in the next 24 months, we're gonna see a, a massive shift in, um, a lot of developers not coding anymore. So there's interesting messages going on out there that are different.

Google seems to be a, a little bit on the risk averse side. Um, I think, you know, you can only do so much education. I think what they, what what we're seeing Google Face is search pressure.

We saw open AI's search GPT just a, you know, a couple of months ago and put pressure on them. So I think what serges are articulating is there's risk averse there. They were afraid of showing mistakes in hairs.

They, and they said, you know, we did, we did that. It was embarrassing. But he's saying we have to be moving past that.

I think that's where marketing can help. Um, but I think there needs to be more proof in the pudding. I think it's there, and I do think marketing can help with that, but ultimately what they need to be able to do is compete well.

And I think we're seeing some cases where they've fallen behind, and now they need to kind of pick up the pace and help those developers understand where they can be upskilled, where they can, can to Matt Garment's point be more innovative. And that's kind of the direction that I see this going, is more on, on that side. Elisa, I think one of the pieces here that you kind of hit on is ups, you know, upleveling and, and skill, um, is, is is a part of that maturity curve, but also, you know, organizations are, are starting to create api, you know, AI specific KPIs, right?

And they're trying to, you know, put these things in place as to kind of force that motion forward. And I, I agree with you that like, there's, there's definitely a need to up, up level and move quicker, but the other side of it is it has to be done at a pace that it's, that is digestible to the both consumer of the, of the application, but also the audience building the application. It can't just be overnight things just shift over, because that's where a lot of mistakes are gonna happen, Right?

You know, there's this, this, this weird dynamic going on, Mike, sorry to to, to jump in, but there's the, the end users, they're kind of teetering between excitement and dread when it comes to ai, right? And often crops up. And I, I just want to mention the example of Congress recently banned a Steph from using Microsoft's AI copilot over security concerns.

And there's a, a Gartner report that just came out. And the reason why I know about this Gartner report is because executives out here are pointing me toward it, and they're pointing me toward it at, for a competitive advantage against Microsoft. Basically, the guard report warns of amplified risks around the 365 copilot, such as severe data exposure, misinformation and data decision making hazards, uncontrolled content, sprawling reading from this report, risky extended data integration.

So there's that, there's that element as well, which, which, uh, I think has been played out. And it's also starting to pick up, and it's starting to pick up from this time of, of year where we're seeing all these AI announcements, and we talked about this before, there's Apple Service now, uh, Salesforce, slack, uh, Oracle. They're all making announcements, and they're all trying to gain an edge, not just in, in certain, in terms of lifting their, their products, but also mentioning the hazards of the pitfalls of others.

And Microsoft co-pilots have been getting a fair amount of that grief and flack, Right? Some of the descriptions of how development is gonna evolve in AI and the age of ai, I'm gonna use a technical term to describe this, are bonky. And here's the Bonky part.

So Whoopi, I am not gonna type my code. The machine is gonna type code for me. Well, that's just freaking awesome.

I still have to figure out how to design the software, architect the software, make sure it scales. This is called being a developer. And there are classes of developers, some are more professionals than others, and there'll be a lot more citizen developers, but they are all developers.

And it's not like AI is going to magically take away the notion of how to build an application. You have to understand how those things work. So the fact that, you know, we have automated, the typing part of creating code is not, in my mind.

This is like, you know, major leap for that's gonna automate the need for developers. And now I'll get off my soapbox, but Paul, am I crazy? Yeah.

The, the, the thing here that you have to, I, look, I agree with you, Mike. I mean, uh, you know, the AI is the kind of e equivalent of using a hand screwdriver or a drill, right? It's a tool to get the job done.

The, the, the fact of the matter is, is our data shows, and I wanna echo a couple of points here and I'll, I'll, I'll, I'll compliment the, uh, report that came outta Gartner, but our future in research shows that 18, I'm sorry, nine months ago, 18% of production applications were running AI in production then ran that reran that report nine months later, and that, that number jumped up to 54%. AI is not going away, right? AI in production applications is not going away.

It's going to be an enabler to make these applications. It can create these applications faster, um, in theory, cheaper, uh, you know, and, and more repeatable. Um, and, and, but, but the fact is, is you, you know, it's an enabler.

I agree with you a hundred percent. It's, it's a tool in the toolbox. It's not, it's not, it's not gonna build a house for you.

You still need to have the plans. You still need to have all the things to make it work. And again, I will go back to my stance all along, which is organizations are accountable for security, they're accountable for the, the quality of the application.

They're also accountable for the impact. So their end users of what they produce, if, if they're using AI to create it. And just because you can write code faster.

So, you know, if I can have it write code and it took, you know, 20 minutes to write the code versus, you know, uh, three days to write the code, it's still producing a, a product and you're still accountable for it. So that's, that's where stance is with it. And, and, you know, by using the data to back it up, it is going to continue to grow.

It's going to continue to be part of the developer ecosystem and part of, of the tech stack. It just needs to be used appropriately. Mm-Hmm.

Lisa, I'm gonna come back to you because, um, I'm not a theme as, you know. Um, so does this make any sense to you where a company stands up with their executive leadership and tells, uh, maybe the 70% of their customer base that's actually responsible for consuming their product, that they're gonna be automated out of existence? It just seems like, you know, the wrong phrases are being used here to describe what's gonna happen and, uh, you know, are these words that they're gonna regret because they weren't chosen more carefully?

That's a great point. I think they need to be really careful in the words that they are using to relay the current status to allay concerns. Um, it's always interesting when an executive, like Circuit Bri stands up and, and says what he says, or Matt Garmin says what he says with prospective developers and, and, and AI productivity.

I think that, um, it's incredibly important to, to message correctly the opportunities, the challenges, um, but going forward, where, what's the intersection? What's the balance of, of managing the opportunities with the challenges so that organizations ultimately benefit and learn how, if things change, if there's a big sea change, where is my opportunity? Where can I grow?

Where can I expand? Where can I innovate? That messaging needs to be there.

And I think I, I don't see it yet, but it needs to be there. Yeah. I mean, Amanda, I don't know what you see, but more and more I talk to people and you use their phrase ai, and you get an eye roll, people are like, yeah.

You know, I mean, Whoopi, and they're kind of like, you know, um, either they're a under impressed what it can do today and not overly convinced them what it might do tomorrow. Yeah, absolutely. They're a little tired of hearing it.

They feel it's, um, being overused in too many areas right now. And I think it comes down to the fact that they are still hesitant about the quality and the security, like we've mentioned here, the quality, security, reliability. And as you said, you still have to have a knowledgeable person who knows about coding and checking all of it, because I mean, like a regular person such as myself who does not know how to code, couldn't just say, okay, AI put out all this code and it's gonna be great and perfect.

I I somebody needs to be there to check it. Um, so there's, it still has a ways to go. Mm-Hmm.

John, is this just the latest example of the left coast being kind of divorced from the rest of the reality in terms of how their products are used? I know we've talked about this. Oh, yeah.

Yeah. We live in a bubble here. There's no doubt.

It's like being, it's like being in the bellway or being in, in New York at this immediate capital, the world of financial capital of the world. We see things around us, which may have been adopted earlier among some of us, and we assume that everybody else knows, and sometimes we have no idea what we're doing, which is often the case. So we presume that we have the final answer.

We have the best answer, and that's 50% of the time it's wrong. So it's a, it's a process. And, um, this is no different than what happened with cloud internet smartphones.

Um, we, yeah, I, I I, I, I cop to it. We're, we're divorced from reality at times. Yeah.

I, I sometimes think the Valley has the same problem parsing the word is that Bill Clinton had during his impeachment trial, But wow, that's, that's a bad gr that a bad optic, that's a whole statement. Wow. But, but to be fair, right?

And let's, yeah, let's talk about where AI is going, and I'll go to Paul about this, but, um, right, the reasoning engines in these LLMs are getting better by the day, and so we will be able to orchestrate different processes that we're seeing the rise of these agents that are trained on a narrow set of data. So I think that, you know, we're on a journey here. I think part of the issue though, is we're setting up a level of expectations between, you know, what's possible today using these tools versus what's happening tomorrow.

Paul, you want to hazard guess this time next year, application development? Where are we? Yeah, I think that it goes back to the beginning part of this discussion where we were talking about, um, you know, organizations, um, restricting the use of things like Microsoft Copilot, right?

Um, I think that what we'll see is, um, the advancements with these, um, these AI tools, these enablers to make these things faster, will be, um, really focused on private learning models, right? Private instantiations and, um, protection of data is going to be key, right? So if we don't have the protection of that data, the, you know, I can, I can completely understand if you're using, you know, a, uh, you know, a system out there, uh, you know, say an Office 365, uh, you know, uh, instantiation of copilot trying to use your information and trying to build your presentation, and it goes, the information gets pushed into a public domain that's proprietary to the company.

That's pretty scary to the company, right? You don't want to do that. You're releasing your information out there.

So a year from now, I would, I would say hope, but I would think that we would as a, as, as a, as an industry, put those, uh, checks and balances and guardrails in place, that these technologies that are being built are gonna be built and used in a way that are more appropriate, that comply with, uh, you know, compliance regulations and governance within organizations. If we don't, then unfortunately, a lot of these things will go the way of, uh, say, Napster and these other technologies that were out there, right? If we were remember Napster.

But, um, but anyways, yeah, I just think that a year from now, this will all be harmonized and, uh, really great technology that's in place right now, but we need to put those compliance regulations and guidance in place. All right? Many, many years ago when I first started out on this business, I had the pleasure of meeting Paul sfo, who was a futurist, and he shared a, a bit of cowboy wisdom that he always views during his analysis of things.

And he said, son, never mistake a clear view for a short distance. Think about, we'll be back in a minute. Discover Textron Group, the epicenter of tech innovation.

We are your go-to, for reaching IT leaders and practitioners worldwide. Our secret impactful content that sparks awareness, engagement, and top quality leads with us. You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more.

Join our satisfied clients. Let's revolutionize your tech journey. Contact us today and tell your story to the world in the most powerful way with Techron Group.

All right. Our final block is this growing disconnect that seems to be occurring between Big Tech and Washington, and for that matter, every other government around the world. But, um, Lisa, we had some, uh, theater this week involving 60 million people or so, watching a couple of people run for president.

Um, part of the discussion that's going on in that political climate these days is, you know, is Big tech a friend or a foe of the average individual? Are they too much involved in, uh, editing content? And what is their, you know, what's misinformation and what is not?

Um, I hear this conversation coming from Washington. I'm not sure if Big Tech is hearing it, though. They seem to be kind of either deliberately ignoring it or they just don't understand it.

But what's your take of what's going on here? 'cause it seems like it's getting a little heated, Uh, it, it is getting heated. You're right, Mike.

I think there are a couple of things going on. I think that, that what we've seen Washington demonstrate for years and years as we see tech leaders and CEOs like Mark Zuckerberg going and testifying in front of Congress, that there is a lack of education. You know, the, the, the government leaders are experts at what they do.

The tech companies are experts at what they do. I don't think they've come to that medium yet where they really understand each other. I think what we're seeing with some of these, um, anti PR campaigns are companies that are really, um, deliberately approaching PR in a, in a way that involves kind of not going the traditional PR route for, to, to embrace controversy, to if knowledge flaws, to disrupt typical messaging.

So we're seeing that go on. Um, and, but what tech leaders really need to do is they need to maintain their credibility. I think what would be great is if we see this collaboration between a diversity and stakeholders, meaning tech partner with government organizations, non-government organizations, helping leaders become more to that middle understanding.

So that ultimately, I think if they can do that, and that's a big if right now, tech leaders could help influence regulatory changes and really facilitate an education there, that just seems to be a gap that we're not seeing filled. I don't see a lot of progress towards that. I did see Mark Zuckerberg the other day, um, and he was talking about how he sent a letter to the, uh, how Republicans apologizing for censoring misinformation around COVID-19.

But he actually said, I I, he was kind of done apologizing, um, that he would definitely, um, push back if something like that happened again. So I think there's still a disconnect there. There's discord.

I don't know how close we are to, to, to solving that, but it's, it's there, it's apparent and it's interesting. Both sides need, need mediation, really to, to help bridge it. Mm-Hmm.

John TOSA's point, The Valley again, seems like, you know, there's this unique culture out there, shall we say. Yeah. You know what, you're reading my mind.

Yeah. You know, exactly. I, I know exactly where this is gonna go.

So my dad's an engineer. My son-in-Law is an engineer. And this area is very engineer focused, and engineers, as we all know, are terrible communicators.

They're not very good at talking to other people, learning from other people. They, they tend to think they have all the answers. I'm not saying this about my father and my son-in-Law, but I, the attitude in general that I get is that we think we know better.

And this is not just applied to the government, it's applied to the entertainment industry, to the media industry. Uh, and, and the attitude, and Zuckerberg, I think is as guilty as anyone was. Trust us, we will self-regulate ourselves.

I know you don't understand Congress, what we do, so let us do it for you. Now, that has evolved over the years because Congress finally learned that with the case of social media, they felt badly burned. So they're trying to get ahead of this issue with ai.

And one of the things that we've talked about is there is, um, a movement towards more of these partnerships between the governments and the private companies. And there is a memo coming out, and we've talked about this as well, and I'll be very short about it, that that plays off of the, uh, executive order about ai. And there is going to be an effort to combine government agencies with folks at, say, Google, meta OpenAI, maybe even Xai, to try to come to some sort of agreement on the responsible, ethical, whatever you want to call it, use of ai.

So we are going in that direction. But, um, it's, it's, it's also being complicated because the, the big tech companies are very good at lobbying. So they are as, as influential as any in, in trying to to move around.

And so we're gonna see big tech kind of try to assert itself where the, the big five or seven companies are going to try to set the rules for everyone else, which creates even more issues for the rest of the industry. And I, I just feel as if this, this keeps repeat history keeps repeating itself. We're not learning anything from here.

And I think it all has a lot to do with the kind of God complex that exists here. I know, and I'm maybe going a little too far, but I do think that that definitely exists on, we'll solve all the world's problems. You know, we, uh, bill Gates that we can get, we can eradicate that, we can solve this.

And it's not, it's not true. Amanda, you're deep in the heart of Texas. And so, you know, I'm here in New York.

Paul is in the North Carolina. Is that where you were? Right.

Lisa and John are in the Valley. I'm, what are you hearing from folks about tech? I mean, are they engaged in this conversation in the middle of the country?

Are they kinda, uh, concerned or are they just kinda like, you know, it's just something happening elsewhere? Well, I think I'm hearing both sides. Um, as we've talked about, I see a lot of my friends and, um, business leaders out in the community embracing technology.

They wanna know all they can about AI and all the other newer technologies. And then you have the other side of it saying, AI stay away from me. They don't want anything to do with it.

Um, it, it's confusing to them. They don't see the point, and they have this idea, it's like some sort of evil terminator, you know, in their mind when it comes to thinking about ai. So I see both sides.

It's also kind of a problem with the, the valley itself. We're very schizophrenic. So we have Google who, who didn't want to be evil, wanted to do good things as, as a counteract to, to Microsoft.

Uh, we had Apple, we had Steve Jobs who genuinely wanted to change the world, but then we also have people like Elon Musk, right? Who, who are the worst example of what is going on here. I mean, that's, that's an extreme example.

So in a sense, we're kind of our own enemies too. There is a tendency among engineers though, and, and it doesn't matter if they're tech or whatever, but, and, uh, and maybe it's shared by developers who are problem solvers. I mean, I feel like there's this issue where, you know, I was good at solving one issue, so therefore I am I major insights on every issue in the world.

And it's like, you know what? That's that kinda reality. But Paul, I don't know what you did.

Yeah, the PETA principle does definitely apply here, right? I mean, if we, uh, if we just elevate people because of their, their past and what they succeeded on, one thing doesn't necessarily make them successful in others. You know, look, I think that there's, uh, there's definitely, uh, kind of the tone of the conversation here.

It kind of flows. There's definitely a, um, kind of almost like a bifurcated split here on, uh, you know, belief on, on how to use things. I know that there's, it's, it's, it's funny when I talk to my, my peers, colleagues, customers, prospects in this space, um, I was just having a conversation with a, somebody from a, a big chip manufacturer based in the, on the, on the, on the left coast.

Uh, and that, that, that, he's a, he's a fellow. He is been there for 30 years. Um, and he was talking about how he is definitely not an early adopter to any of these things.

And the more people in the tech world that I talk to that work in the tech world, um, that express that they either don't use what they're building or don't let their children use what they're building, um, it's, uh, it's concerning, right? It's concerning. 'cause there's one side of it.

It's like if they're building it and they're the ones that are in charge of doing this, yet they don't want to use it themselves, and they don't want to use, uh, let you know, they're, they're closest people use it. Uh, that in my mind is, is it should raise your eyebrows. It should think like, oh, well, maybe I shouldn't do that.

Right? Um, but the other side of it is, is we also have to have the leading edge, bleeding edge people to kind of test and break and try this. I mean, in the developer world, it's open source, right?

You, you know, that's why we want to have open source so we can harden and bang on things as much as we can, so we can expose the issues, uh, that could, that come out of these, these projects that are out there. But when you, you know, when you're building these tech stacks and, um, you know, the, the, the owners of the tech stacks are questioning it, uh, that, that raises my concern as well. All right, we're gonna play round robin here 'cause that allows me to circle through and get everybody's opinion.

But I want your best advice for kind of closing this divide. And I'll start with John, who's kind of at the heart of one of it, but Schwartz is our, our guy in the valley. So, Well, my frustration is, this goes back, I even think back to the late nineties, and I, I was working at the Chronicle in San Francisco and I did a story about Diane Feinstein.

So I talked to her about tech because she was a, a very effective senator, but the one area that she had a blind spot was tech. And the tech industry did itself no good by, by not reaching out to her. Uh, it, there definitely seems to be progress, and I'm very hopeful and optimistic that, um, members of Congress, especially out here have, have gotten up to speed, have gotten a lot better about understanding how tech works.

Even to my local congressional race. I talked to the two candidates here a couple of months ago, and they both understand the technology much better than their predecessors. So I have hope there in terms of education and a meeting of the minds.

Alright, Amanda, you run Textron, do ai. For those of you who dunno, Amanda Ani, um, what's your advice? I think it's all about transparency, openness, education, collaborating with a variety of different people and focus groups, and just allowing people to learn about use cases and play around with the technology so they become more comfortable with it.

I mean, this same issue happened back when the internet first came available. Everybody was very concerned about the internet and the repercussions of it. So I think it just takes time and education and just being open about it.

All right. Paul Nadi is the practice lead for application development for the Vitor group and has a unique perspective on this, but, uh, I don't know, Paul, developers just need some sensitivity training and that'll solve everything. Oh, I don't know if that's really what it is.

I think it's a, I like what Amanda said, uh, transparency, but I also think it's a trust issue. Uh, I think that there has to be a building of trust. Um, there's too much, uh, you know, too much, uh, in play right now where people put, or vendors or organizations and industries just put things forward.

And it's not vetted out, it's not tested, it's not, there's no, call it fact checking. I hate to use that right now, but, um, but basically there's not a lot of, of, uh, you know, full understanding. And, you know, I think it's once burned twice shy, right?

I think if you use something and it doesn't work, right, um, it, it, it causes concern. You know what, what, what I find in, um, uh, in our research is, uh, only, uh, a couple years ago, only 29%, 2022 data to only 29% of organizations were doing continuous testing in the CICD pipeline. When I reran that study in 23, it jumped up to 60, 66%.

Um, and, and part of that reason is because there's problems with the quality of information, like, you know, of, of applications and things that come out the door. If you're burned by the quality, you, you're less likely to use it. I mean, how many, just looking at all of us on the call here, if you, if you're, you know, watching a movie and it, it continues to stream and, you know, buffer, you're probably gonna find an alternative source to watch it, right?

That's the same with any application, right? So, uh, that's what I think it's, it's a trust issue. I think, uh, again, I'll back up what Amanda said there on, on, on, uh, transparency that needs to kind of come through, but I don't think it's about sensitivity.

I think it's more about understanding and education. It really does come down to maturity. Right.

Lisa, I'm gonna give you the last word on this subject. You're the CMO advisor for the Turing Group, and you speak to all these people who are crafting these messages. So what's your best advice?

I really like what everyone said so far. I, I'm gonna kind of, um, lean on what John said. I think it's encouraging that we're seeing, um, some of the, maybe the, the, the newer, um, political folks leaning into technology and, and working to understand it.

I like what Amanda said as well, in terms of the collaboration, the partnerships, the education that really needs to be there. I, I, I mentioned, um, optimism about the, the collaboration between stakeholders, you know, government, non-government, industry peers. I think that can go a long way to enhance credibility, but also what I think it can do is really help build those relationships.

Marketing can definitely help there. And I think once those relationships are stronger and there's a little bit more understanding and, and maybe a little bit more kind of coming to the center, I think it will lead to critical conversations that can really determine the future much better. So I'm encouraged what John's seeing out here on the left coast.

Um, but I think it's all about that open collaboration, communication and really trying to come to an understanding. It's not gonna be in the middle, but they need to get closer. And I think that that messaging can help with that, but it's gonna take more than that.

It's gonna take relationship building. All right, folks, back before Congress was this, uh, swampy thing where everybody just kind of argues all the time, regardless of reality, and their political party matters more than the thought process. They used to have things called fact missions.

They used to actually go places and learn things and figure out what was happening with those things. I think maybe it's time for those folks in Washington to get out and talk to the people about tech because well, everybody seems to be at polar opposites, and that's not good for anybody. Hey, I'm Mike Biard.

Thanks for watching the latest edition of Techstrong Gang. I want you to stay tuned and watch the rest of the episodes for Techstrong TV that are coming up right behind us. Of course, you can find this episode on websites and all our favorite streaming sites.

Until then, we'll see you next time.