Telemetry News Now.
Welcome to another episode of Telemetry News Now. We are in a strange place today because we are in sort of like the post Valentine's Day, post holidays, but, like, prespringtime and summer. In upstate New York, this is like when the great gray blanket just sort of covers all of the northeast, and we all lay low in our homes with the lights off and the sheets over our head until the springtime finally emerges. That's where we are today.
Also, our good friend Justin Ryburn is not with us. He is traveling both this week and next. So we have a special guest with us, Scott Robohn, who you've probably seen plastered all over the Internet as the, cofounder of the Network Automation Forum among other things. So, Scott, it is so great to have you here today.
And I don't wanna say, you know, standing in place of Justin for the time being because I think this is a very special possible.
Well, no. That's not what I meant. What I meant is that this is a very special occasion for me. So thank thanks for joining today.
Happy to be here. And in your gray blanket that you're experiencing in upstate, we're sunny and sixty in the Shenandoah Valley, here in Virginia.
And today happens to be, my wife and my wedding anniversary. So, how else would I be celebrating that anniversary than coming on Telemetry News Now, Phil?
I mean, that's that's the best thing. I mean, of of all the options, the celebratory actions that one could take, that that would be at how many years if if I can ask? Thirty six years. Oh, congratulations, Scott.
That's great. That's great. Yep. I'm looking at, twenty two with my wife in, about six weeks or so, almost two months.
You're good. Don't blink. Yeah. Don't blink. Alright. Well, let's get into the headlines for today.
So the first article we'll tackle today is from SDX Central on February nineteenth, titled, Should Broadcom be Eyeing Intel?
Now when I came across this the other day, I kinda felt the earth shake underneath my feet, and here's here's why. You know, Intel has been such a mainstay of semiconductor manufacturing in the US.
They've had some really tough times the last few years, especially in the last twelve months.
And seeing this potentially huge move of some or many assets coming away from Intel into the Broadcom empire. We've all seen Broadcom swallow up other properties. We focus a lot on their software acquisitions these days, but, this could be a really interesting move. Phil, what do you think?
Well, I mean, swallow it up. I mean, isn't, like, VMware one of them that kinda, like, was a cornerstone of IT operations for many, many years and and still is, of course. Correct. You know, I I have to imagine that there are gonna be some serious, issues with government approval, like we've seen with, like, HPE and Juniper, right, in recent days. Sure. I have to imagine that there's there's gonna be something along those lines with this as well. Right?
It's possible, especially with the, you know, the wind blowing in the direction Philip manufacturing in the US. Exactly. And Intel Intel being one of the one of the largest, if not the only, that still has significant silicon, production presence in the US. So, you know, Chip's act notwithstanding, let's, let's see how this gets carved up.
I actually had a conversation a few weeks ago with a friend of mine who was at Broadcom for a few years and had significant interaction with Hock Tan. And, they're it's all business. They want to bring assets on board and get productive revenue out of them. So what Yeah.
And the issue there is that, you know, TSMC is not an American company, and Right. Hence the conflict with the Chips Act and why I think this could be a concern with regard to government approval and the complexity to execute that kind of an acquisition.
That being said, Broadcom has also been, like, in debt reduction mode for quite a long time. So are they even in a place financially to do that, or is this just like they threw this out there in the news to, like, generate some sort of a news cycle?
I don't know. No. That, you know, the article points to them actually being successful three quarters going and I think reducing debt by nine billion. Okay.
So they are making progress on that front. You know, it's it's like you teed that question up perfectly, Phil. Thank you. You know, we'll we'll see.
It's funny that, you know, you started out talking about upstate New York.
One of the CHIPS Act properties, I've kinda tried to keep an eye on is a Micron proposed build, north of Syracuse, New York. And I'm really looking for some successes from CHIP Act money. I'm not sure that I can point to any yet. Mhmm. It's hard when the cost of manufacturing is high and you can do so much better, you know, in other countries, Finding ways to incentivize and actually do this domestically again, it's not trivial.
I'm not saying it's impossible, but Yep.
So next, you enjoyed the movie. You might have enjoyed the, multiple TV franchises called Stargate.
That's not what we're talking about. The next article we'll tackle is from Futurium on January twenty second. Stargate has been announced, and here are our questions. So Stargate is a proposed new loose program and cooperative effort that involves some of the big players like OpenAI, like Oracle, like other foreign investment, organizations to incentivize the development of data centers in the US for AI capacity.
As with anything like this, you know, you first you treat it with skepticism. Right? Okay, great. It's a lot of talk. Could just be marketing splash.
But you see the players coming together and talking pretty seriously about this. And knowing that we're in an era of deal making, whether you, you know, like the current trend or not, It's something to watch. And, over the next few months, we'll see if anything really, you know, comes from it.
Yeah. Well, what does that mean? What does comes from it mean? Because, like, I've I read through the same article, and I and I obviously linked or rather it's been a few weeks now.
This came out in, like, late late January. So I'm I'm you know, and we even mentioned it on this podcast. So Yep. I'm aware of the initiative, but I'm still actually not aware of what act like, what are they doing?
Like, what is this supposed to be? Like, we're going to build new data centers. We're gonna train new models. We're gonna support private industry.
That's, you know, that's that's kind of an important thing. So I do wonder if this is, you know you said the word loose, and I think that's an apt word here because I think that's what the strategy is. It's very loose and uncertain as to what's going on. We're talking about a lot of money here, but we're also, you know, missing some of those big players in the AI realm right now, like Microsoft that was conspicuously absent from the whole Stargate conversation, NVIDIA as well, although I'm sure they're related.
You know, they're gonna be connected because of the the the chip manufacturing role that they play. But that's, you know, that that's concerning to me. Some of the web scale companies, the cloud companies, Amazon, AWS. Right?
Yeah. They were not not part of that, discussion.
No. So I don't know what this means, and I think that that uncertainty, is on purpose. I think that this is an initiative that's maybe largely political. You know, OpenAI is all over it.
SoftBank is all over it. But we did see remember that tweet from Elon Musk where he's like, SoftBank has no money. Get out of here. I don't know where that stands today because that was a few weeks ago now.
Correct.
You know, I think that's actually one of the interesting aspects of this. Knowing that, Elon Musk has a lot of connectivity with the current administration, but the current administration is also willing to have conversations with his rivals, like Sam Altman at OpenAI, and I actually think that's a pretty good thing, you know, in the big picture. So Sure. Sure.
The the the metrics are as much as five hundred billion investment and as many as a hundred thousand jobs. We can pretty much be assured that those metrics won't be hit, But, I I hear you on the uncertainty around this. The flip side of that coin is I want the market to figure it out. I would like those players to come together and figure out, okay, what investment makes sense not to be prescribed, from a two hour drive away from me in Washington, DC.
Mhmm. Mhmm. Because make no mistake that the federal government on its own, apart from this initiative, is doing work in this realm and in this field. You know, whether it's monitoring troop movements across the world and applying models and algorithms to understand and do some sort of predictive analysis, they're already doing this kind of stuff. So what is it that is it just the vested interest of of bolstering the United States economy and then partnering? As as we've done for generations and generations, right, where we have relationships, all sorts of big companies to the extent that some of those big companies like Lockheed Martin and and others, the the vast majority of their revenue is from the government.
I do think that that is likely considering the importance and the impact of artificial intelligence on society in in general at large, that it could be a very similar partnership between private industry and the United States government the United States federal government, just like we had with, like, defense contractors and, manufacturers in the past. I think that artificial intelligence is gonna fall right in line with that. Perhaps that's OpenAI, perhaps it's not. But I I do wonder if the Stargate, program and initiative is the first step in that direction, where we're gonna see that kind of relationship.
Well, what if we could have, you know, an Apollo program like this, but with some of the best competing and cooperating technology providers, that's interesting. You know? That could be really interesting, and I know I should take the rose colored glasses off right now. But, you can see that cooperation. It doesn't have to follow an exact template of things that we've seen in the space race or in in, defense.
Yeah. Past results don't guarantee future. I don't know.
Whatever they may turn.
Yeah. Your mileage may vary. YMMV. Right?
I have really good consulting clients that are involved in government contracting, and I'm swimming in this right now. And people are trying to figure out what's the impact of this. How do we align to the very loose, promises that have been put out so far? And it's it's a wait and see.
Right? We gotta we gotta see what develops. Yep. Down the list, next item for discussion.
From SDX Central, February eighteenth, twenty twenty five, CompTIA introduces Cloud NetX Certification for IT Professionals in Network Architecture.
So I'll state upfront. I've never taken a CompTIA test or cert. I've always been a vendor cert person. You know, I was early on in the Cisco program in the nineties, then Juniper did some Nokia certs later on in my career.
Uh-huh. I've always felt like the context of implementation in a vendor environment was really helpful. Never saw the value in the vendor neutral certs. What about you, Phil?
Have you ever have you done any of the CompTIA certs?
I did one. I did the Network plus years and years and years ago, and that was the very first certification that I earned. And it was like a watershed moment for me in my personal life because I was transitioning from one career into technology.
When I passed that, I think I took it twice. When I passed that at the local Pearson VUE center, it was a big deal to me. And I remember learning, you know, the basics of TCPIP, what twisted pair is, what, crosstalk is, and I think there may have even been a section on crimping, cables and and terminating ends. I'm not sure. But that stuff was meaningful to me. So I do think that there's value there for sure. And I think moving forward, there's gonna be more value to vendor neutral certifications.
Now maybe it's not certifications per se as in, like, a formal certification. Although, I think there's absolutely room for that because it it's something to latch on to that's structured.
But, really, the spirit of what I'm saying is with vendor neutral training and professional development in general, especially as we kinda go beyond You know, for many, many years, Cisco did a fantastic job of of having this education program. They still do, by the way. And and so in that, they were able to develop, like, generations of people that knew Cisco that were brought up on Cisco, such as myself and you you mentioned Cisco and Juniper as well. And that's fantastic.
And and it is a very ubiquitous technology, so it does make sense. It's not that it's impractical. But that is starting to change now, especially as we look at other, networking vendors in the data center, in in the security realm, in the wireless space. And so it is it is important to branch out to multiple vendors that understand the fundamentals.
But where do you do that? Where do you learn those fundamentals? And I think this, initiative from, CompTIA, CompTIA, however you say it, this is specifically more focused on folks that already have a background in networking and want to kinda, progress, go to the next level in their systems architecture role, whatever you wanna call it.
Right? Right.
You know, that that's good.
And I and I do and I do understand that, but I will say, if you're a very if you're an experienced network, professional already, are you gonna look at those exams? Are you gonna look at a CompTIA course or track? I I wouldn't. I wouldn't.
So maybe I'm wrong, though, and I'm just being cynical.
I am curious enough to go take a look at the details and see Alright.
That's a good one.
So I might I might do give myself some homework here. I'm, you know, I agree with you in that I have shifted toward I see more value in the vendor neutral certifications, and I I think my involvement in Network Automation Forum has really influenced me on that the most in that we see more open source everything and vendor neutral tooling.
And so there's more of a place for, just pick on Python as an example. Right?
Acquiring skills in Python programming for network automation and being able to demonstrate them reliably, that's gonna be that's that is more and more a useful tool. And the last I checked, nobody nobody corners the market on Python.
So Right.
Right.
I think I think this is moving up into the right for sure.
You know, that's something I was thinking of is that in order to go to that next level, if you're already an experienced network engineer and you're trying to stay right there at the forefront of what's going on in technology, presumably because you need to, but, you know, maybe because it's an interest and you wanna learn, The technology is varied and diverse. It's not just like learning the next routing protocol more deeply and, like, how do BGP communities work? It's it's really kinda branching out into other technologies that are networking adjacent or maybe not even networking at all, but you kinda gotta know it these days. So I don't wanna say it's a concern, but it does make it more difficult for any one vendor to jump on that and say, we got the end all, you know, certification track Yeah. And body of coursework to get you there.
Well, very selfishly, I'll say, you know, what does it mean to be a next generation network engineer?
It's not subnets and spanning tree anymore, and there's just lots of other stuff that you need to you need to learn.
And not just Python, Linux skills and more. Right? They're again, they're very vendor vendor agnostic.
Mhmm.
So next, hot off the press from the Cisco press room this morning, February twenty fifth. Cisco expands partnership with NVIDIA to accelerate AI adoption in the enterprise.
It took us this long into the episode today, Phil, to actually start really talking about AI. So, we should congratulate ourselves for that.
That I do congratulate myself because that's something that Justin, like, kinda sends me private chats while we're recording. Like, really? Another AI article? Come on. So I agree.
So, you know, my basic reaction to anything I see from a vendor press room and perhaps this vendor press room and AI, my radar is up automatically.
I kept reading. It's like, Nope, this is, this is interesting. Building of common architectures between Cisco networking products and NVIDIA GPUs, even in our operation with NVIDIA's Ethernet networking products and technologies. There seems to be some there there.
It's interesting timing, you know, with NVIDIA GTC just coming out in a few weeks. That's probably no accident.
So I'd encourage people to take a look at this and say and and see what they think. I think it's also another, notch in the Ethernet belt, if you will. Right? Absolutely.
Giving a major hint here that in Nvidia sees the future, and the future is colored Ethernet.
Mhmm.
So, Phil, your reactions.
So here's a quote from Chuck Robbins. Right? Enterprises are under immense pressure to deploy AI quickly and effectively. This is all from Chuck Robbins. Direct quote. And many leaders struggle to justify the investment while balancing the risks. Together, Cisco and NVIDIA are partnering to remove barriers for customers and ensure they can optimize their infrastructure investments to unlock the power of AI.
So we're talking about the, Spectrum x product. Right? The idea from NVIDIA is that this is gonna give you what they say. Okay?
This is from their website, nearly perfect bandwidth at scale. That's important for AI workloads at scale. Extremely low latency end to end stack optimization, whatever that means. Advanced, RoCE extensions for scalable AI communications, deterministic performance, that's important, and, which is interesting, built on Sonic and NVIDIA Cumulus, which are open network operating systems.
Right? So, yes, I agree. Obviously, this is an Ethernet focused, you know, initiative, pretty big scale, but they use the term enterprises.
And so I assume enterprise is doing some sort of serious AI workloads.
Is that, like, true?
I know some enterprises from working with customers that there are some that are fine tuning models and training models because of the nature of what they do. Very large financial. Like, you would recognize their name, you know, on a global scale. Sure.
And certain service providers where they have the resources and the initiative and the and the need to be able to apply this technology to do fine. Okay. But by and large, I do believe this is gonna be a more consumable technology where folks are not gonna be running workloads in that sense in your typical enterprise. Yes.
They may leverage certain AI technologies, even running certain things locally where they do need to have, you know, workloads bouncing back and forth between some pods in the data center, but not to the extent where you're talking about how do we reduce job completion time for training our model over these next three or six months. So that's why I wonder if this is really the most relevant thing in the world or not. Because, you know, I know from my own experience when I went through the Cisco certification processes and stuff, and I'm reading the textbooks, they would identify an SMB, a small and medium business as, like, a company with ten thousand people.
And in my brain, I'm like, what? Yeah. It's like that's a big company.
That's a company with resources and multiple c level and departments and Juniper has almost ten thousand employees.
Right? Fair enough. Not a small not a medium business.
Yeah. Exactly. You see my point that, like, if we recognize that most enterprise organizations out there are not ten, twenty, thirty thousand people or more, right, whatever they consider large enterprise, how are folks gonna be utilizing AI? And I don't think it's gonna be by, you know, running heavy duty workloads in data centers that are that are going to require a lot of this technology. Now that being said, the amount of revenue you can generate from a couple of hyperscale customers that are doing that is significant. So it still makes a lot of sense for Cisco, which I personally believe have been, until very recently, been conspicuously absent from the AI conversation with regard to networking.
We've had other vendor names popping up a lot, not Cisco, and that has changed in recent days.
I don't feel any pressure to to assign superlatives to this. Is this the most interesting thing? I don't know. What I do know what I do know is it's an important juncture, and it looks substantive.
It it's like, who who wants to be partners with NVIDIA today?
Everybody.
Everybody wants to be part of some NVIDIA program. This is smarter than your average bear. Right? This looks like a substantive, cooperative development of architectures. And if for anybody who wants to go read this, piece, read all the way to the end. Like, the most interesting details are in the last, five, six, seven bullets of the item. You have to you have to wade through a lot of PR speak to get there, but stick it out, and you'll see you'll see what I'm talking about.
So also from the Cisco newsroom from a couple weeks ago, not even a couple weeks ago, a little more recently, Cisco, again, just unveiled the n ninety three hundred series smart switches.
Now this is a new class of switch, wait for it, AI optimized data center networking hardware.
What? Pretty cool stuff. Now the new switches are gonna integrate AMD, Pensando, DPUs, and Cisco Silicon one, which they say this is Cisco says, says enhances scalability, security, and adaptability. I'm not a hundred percent sure what they mean by that last one. And again from Cisco, these high capacity, multifunctional switches are meant to simplify AI data center designs.
So how do they do that? By embedding networking and security services directly into the switch fabric. So the idea there is that, therefore, I guess, you reduce hardware sprawl and operational complexity, you know, by integrating everything into one platform and one to one fabric. So the the first major service is Cisco HyperShield, and it's gonna provide AI native hardware accelerated security.
The idea is that it creates what they call a micro perimeter around workloads, and that's interesting, to enforce autonomous segmentation policies and update security, automatically.
The first n ninety three hundred smart switches, they're gonna have twenty four one hundred gig ports. They're gonna ship in spring of twenty twenty five, so that's right around the corner, and, with a top of rack model following later this summer. And we did just talk about NVIDIA's Spectrum X, so here we are.
So I've kind of lost track of Pensando over the years. Like, they've been up and down and up and down.
And now to see, okay, they're finally getting still a a branded reference in a new architecture here. This is this is pretty interesting.
I do wanna understand what that secure micro perimeter really looks like.
That sounds Star Trek to me. I I know that you'll that'll get a comment out of you if I if I put it in Star Trek terms. But, yeah, we'll we'll see what this really looks like. I'm trying to envision what does it look like to have a bunch of DPUs housed in a switch. I'm sure there are some good pictures that I can take a look at that'll help clarify.
I mean, and the security conversation is now picking up a lot of steam because there are folks that are talking about, alright. You have this dataset that we're looking at. We have this large language model that's interrogate. We have this rag system. We have this agent system, whatever.
But wait a minute. Should that particular agent have access to this data? Should this workload be able to speak to this workload? So we are now you know, at first, we were very excited about the the potential of the technology in a broad scale. We didn't really think about security. I don't wanna say that security is an afterthought, but you and I are networking people. Security is an afterthought.
Yeah.
I have a friend, TJ. Security is not an afterthought. That's where he's at at first. But for us, it was, like, all about, like, connectivity and how how do we run this workload and get this amazing result.
But now that we're starting to realize the implications of those security ramifications where we have, like, do we need, like, read only access to this particular chunk of data? And maybe we need to have role based access control for agents, not for people, not human beings, but for agents. It does make sense that we are now cordoning off sections of data or, the infrastructure or workloads, whatever that means. I'm not exactly sure what they mean by workloads. Is that a series of DPUs and GPUs, whatever? That does make sense with the direction of the conversation in the industry right now. So I get that.
We have a good model for this. It's it's from the cloud world, IAM. Right?
Who gets access to what? You know, what processes should be able to have connectivity to each other?
And in the AI agent discussion, we talk about collaborating with agents like colleagues.
I think that model can directly apply too. Like, not all agents should have access to all types of data, all other agents, etcetera. Yeah. So I'm not saying it's a perfect model, but seems like it's a good place to start from.
Yeah. And that's exactly where we're going. So okay. Moving on, from TechCrunch, February twenty fourth.
For us recording today, that was yesterday. So this is recent for us. Perplexity, you may have heard of, teases a web browser called Comet. So Perplexity, which is an AI company that's been in the news lately for other things, they just recently announced plans to launch its own web browser named Comet.
Now we don't have the details on its release date or its features yet, but I personally think this is a pretty logical move for them. Now they are already an AI company specifically built to use large language models. I'm not exactly sure how they do it. To search, the real time Internet.
Right? So it's already an AI powered search engine. So to me, this announcement, like, it makes a lot of sense, at least at least in theory.
And the thing is that the browser market this is why I say in theory. The browser market is already I don't wanna say saturated, but it's very competitive. And it's not like Google or Alphabet, whatever. And Microsoft, they aren't you know, I mean, they are actually adding AI technology to their browser functions. So it's not like they're not doing it, so there's no other options.
The reason I say that Perplexity has been in the news is because just recently, I read about how they launched, something called Deep Research, a deep research tool to specifically compete with OpenAI, Google, XAI, all that. They got quite a bit of press for that. And, and I think they also put out something for, like, an AI assistant for Androids, and and that kind of thing.
So all of that's interesting to me. And I I'm a big fan of Perplexity. Fine. They have a very high valuation now in the nine plus billion dollars. Cool. But there's a larger discussion here, in my opinion, that goes beyond, like, a new product announcement.
Internet search has been, like, what we've known it for decades now, where you, like, go search for a thing, and then you have links to other things, and you're in charge of the search. Sure. I get it. There's ads, and there's, like, you know, algorithms that put certain things on on page one instead of page two. I I do understand that. But in theory, it's links and and hyperlinks or whatever that you can kind of explore and go down the rabbit hole of information.
That's not the case if we go down this road with, like, AI assisted search and and and search engines and and browsers that have that incorporated into it. Scott, have you looked at, like, when you, like, Google something or I don't know what search engines you use, but you get, like, that automatic AI response right at the top.
And I almost automatically ignore them. Like, because I'm Okay. And and that's muscle memory. I'm not saying I do that deliberately.
It's like, I've been searching for stuff in browsers this way for this long. I'm not gonna change now. Sorry. Mhmm.
But the parallel here is I've been typing stuff into a window on my computer and expecting results for a couple decades.
Now I'm just asking an AI tool. I'm asking it questions, and it's giving me results in English language without me having to go hunt through the URLs that are provided.
Now let's caveat emptor. Right? Let the buyer beware. You know, you've gotta verify a lot of that stuff.
Mhmm.
But, there's there's value in the the new model. And I my Google hacking is now prompt engineering.
So Mhmm.
I agree with you that there's value, but there's also value in owning the search process.
Sure.
I get to choose what I put in that prompt, which which you would do with a, you know, AI engine anyway. Right? AI search engine. But then it just gives you the answer.
We read it. We move on. Right? If I'm searching it out and researching it out, which I I understand requires effort, the current model of search and search engines, it requires me to vet certain information to kinda, like, bounce around and understand the complexity of an argument or the complexity of an issue or of a topic.
Whereas getting a response back, I I understand you just ignore it, move on, but I don't think most people do. And I sort of struggle with that where I'm like, alright. Let me just read this thing. That is a very different paradigm of how we, you and me, human beings, like in society.
You know, get get rid of the engineer part. But, like, just, you know, people walking around, consume information, and learn new information when they're trying to look it up. You know? And so what particular engine is behind or rather what particular, AI model is behind a particular search engine when we're using it, and why is it producing the results that it's producing?
And, you know, that those are all concerns for me. So it it changes the dynamic of how we find, search, consume information, and perhaps our understanding of what's going on in the world around us because it's been colored by that search engine.
So you don't have to abandon all hope, ye who enter here. Right?
And I actually forget which engine it is, but the default is that you put in a query however long and detailed, and it'll give you URL references that it actually used and the results are returned to you. Okay. So so there are mechanisms by which you can do some fact checking if you have any concerns, and maybe you should do more fact checking even if you don't have concerns.
And there are ways to use the chat tools. Like, one of the number one use cases for me is I'm trying to learn about x. I know a, b, c, and d.
Point me to five or six other areas that I need to look into that I haven't already covered on this topic area.
The tool will return, you know, usually pretty good suggestions on things to go look at. And I can either query further in the AI tool and or I can go use a traditional browser and just look for links on those topics. And I do I do both pretty routinely. So you're in control. You don't have to let it do it all for you.
Yep. Thank you for the, reference to the inferno.
That was, that speaks to my heart as a former English teacher, high school English teacher from the beginning.
Shooting for that. Now I know. Now I know.
Excellent. Yep. Absolutely.
Okay. Moving on from Reuters on February twenty fourth again. India's IT sector should overhaul business, create own language models, and the, and that's from the HCL Tech CEO.
So from an article out of Reuters, Indian IT firms must rethink their business models as generative AI reshapes the industry. That's a warning from the HCL tech CEO, C Vijaya Kumar.
So speaking at an industry event in Mumbai, he explained that AI driven transformations are fundamentally different from previous shifts that we've seen in tech. So think things like cloud adoption. Right? This shift requires companies to be much more proactive in developing new revenue streams.
Vijaya Kumar stressed the need for India to develop its own large language models, I get that, to reduce their reliance on foreign tech companies, which I I assume he believes would mitigate geopolitical risks.
And he also warned that open source AI models may not remain freely accessible. So think about the open source models that we use locally all the time now. And that would make it absolutely essential for India to invest in cost effective AI training themselves, right, their own infrastructure themselves.
I mean, if we consider that AI companies that we normally discuss, right, on this show, right, they appear in the news regularly. They're almost all American companies With a few exceptions, we we have been talking about deep speed DeepSeek quite a bit, recently.
They're not necessarily models coming out of academia. I know that MIT and Stanford stuff, they have folks that are working on this stuff. But when you hear the news about the latest model, the new the latest technology, it's coming from, you know, private companies.
And and maybe those companies are now partnering with the government, but, nevertheless, they're they're private companies. So they don't necessarily have that unspoken rule that academia has to share information for the sake of science or the betterment of humanity. Right? Like you'd have if you were putting something out there on, you know, archive and you work for Cornell or whatever.
I get his position at least on being somewhat concerned about, like, one nation's complete reliance on another nation's technology and a technology that's incredibly impactful to their economy, to their society at large, and that really will direct the progress of GDP and and their access to information and all that kind of thing. So I I do understand that from a, like, a geopolitical and by the way, countries that might have competing interests as well. So I do understand his position. And, you know, I'm I'm speaking about this article right now from Reuters, and it is talking about specifically India, but substitute any country in there.
Sure. It makes sense on a broad scale, I think.
You know, this raises more questions for me than answers.
I think one of the dominant factors here is the high availability of model training material in English. That's probably a huge factor here in what we've seen for development of models so far. I'm very curious if there's an analogy that follows. We know that different spoken and written language can cause you to think differently about problems and issues. Mhmm. And I wonder if there's work done in model development and training in other languages that there could be different or interesting results, you know, on top of the importance of the the GDP economic related issues you've called out.
Interesting.
Yeah. You know what? I'm glad you brought that up. I wasn't thinking about that. I was really thinking more in terms of, like, geopolitical stuff.
But you may know I think my the listeners may know years and years ago, I was very interested, in pursuing a advanced degree and a and a focus in my life on semiotics, which is the idea of how does language because we think in a framework of something, and it's language. You know, for me, it's English. My mother is Italian. And that's really how we negotiate with reality outside of us.
And I do believe that there is an existence of an objective reality that we can know and that exists apart from us, and we get we, like, interface with it through language.
That presupposes some interesting things about large language models.
Do the large language models, which don't really contain facts per se, they're vector databases at huge scales, do they then understand, quote, unquote, and present knowledge, whatever that means, differently because it's a different language. And I do know that, you know, when you look at some vendors that are not US based, their Python is in English. Right? So there is something to say here about, you know, countries around the world looking at, alright, all these these American companies and a few, Chinese companies developing these large language models and these predictive analysis algorithms and all these kind of things. You know, what can we do that's in our best interest? And I think that's gonna be a major a major concern for folks moving forward as artificial intelligence becomes such a driving force to, you know, a nation's economy, to the nature of the the makeup of their society, and and so forth.
Everybody has homework now. The weekend after you listen to this episode, go watch the movie Arrival with Amy Adams and Jeremy Renner, and you'll you'll get a hint of what we're talking about here from, how language and communication can impact how problems are solved. Yep. Absolutely.
Alright. So, moving on to upcoming events. I have a long list for us today. We're gonna start off with the Florida networking user group that is, a networking user group put on by the USNUA.
That is gonna be on March five in Tampa, Florida. If you're not familiar with the USNUA, you can always visit their website USNUA dot com. Click on groups and events and find a event near you.
The Indiana networking user group is on March thirteenth in Indianapolis. The Texas networking user group is on March thirteenth in Dallas, Texas. There's quite a few going on. We're we're getting into the warmer weather. GTC, an important event, March seventeen through twenty. I don't have in my notes here where that's located. Scott, where is it?
San Jose. San Jose, California.
Yep. Networking field day thirty seven, near and dear to my heart, will be this March nineteen through twenty, nineteen and twenty in the Bay Area, also live stream. So make sure to check that out, tech field day dot com. The Kentucky network and user group is gonna be on March twenty in Louisville or Louisville, excuse me for mispronouncing that for those over there.
DCD Connect, data center, focused event, is gonna be on March twenty four and twenty five in New York City. I'm gonna be there. Scott, I believe you are as well. Right?
Yep. We'll both be there.
Excellent. The Virginia Networking user group on April thirty in Reston, Virginia.
Scott, I believe you are the event leader for this one.
Right? I am. It's, near and dear to my heart, and it's the April third, not the thirtieth.
Oh, excuse me. I said April thirtieth. Okay. So April third in Reston, Virginia.
Last but not least, this is way out, but because Scott is here, I thought it would be right to talk about it. AutoCon three, May twenty sixth through thirtieth in Prague.
The deadline for that is also coming up in just a few days on February twenty eighth for submissions for presentations and workshops.
So, of course, if you're interested in attending, you go to the oh, actually, Scott, why don't you take it from here and tell us how folks can sign up and how they can submit a talk?
Yeah. Go to network automation dot forum. In the upper left corner, you can click on future events, and you'll see submissions for talks and submissions for workshops. You'll also see the registration link for AutoCon 3. And if you are listening to this on February twenty seventh, you got about twenty four hours to get an idea for a talk or a workshop in. We've got, quite a bit of incredible content already submitted, and our work's gonna be cut out for us the first couple weeks of March, going through all the proposals, you know, the the whole advisory board together with Chris Grunemann and myself. So I did a site visit.
I love the venue that we're going to. I want to make you all as jealous as possible. It's the best venue, I think, that we've been in yet. This is our fourth event. So, hope to see you there.
Excellent. Thanks very much. So for today, those are the headlines. Bye bye.