The EDGECELSIOR Show: Stories and Strategies for Scaling Edge Compute

Tech Jam Session: Virtualization, On-device AI and the Future of Silicon with Tech Legend Bob O'Donnell

November 28, 2023 Pete Bernard Season 1 Episode 10
The EDGECELSIOR Show: Stories and Strategies for Scaling Edge Compute
Tech Jam Session: Virtualization, On-device AI and the Future of Silicon with Tech Legend Bob O'Donnell
Show Notes Transcript Chapter Markers

Hold onto your headphones as we venture into the world of Edge Compute with tech guru Bob O'Donnell, esteemed analyst and founder of Technalysis Research. Bob's profound insights into the tech industry will take you on a journey through the virtualization of computing, unveiling its transformative potential. But don't be fooled, our tech expert is also a rock n' roll enthusiast, leading his band, the Headliners, offering a testament to the diversity of passions in the tech realm. 

Get ready to immerse yourself in an engaging discussion about the role of software and virtualization in reducing risks and increasing flexibility in modern network infrastructure. We'll navigate the debate of Open RAN versus Virtual RAN and delve into the intriguing prospect of vehicle virtualization. And it doesn't stop there, buckle up as we fast-forward to the future of risk and customizable silicon, examining the competitive edge of big tech companies in crafting their own custom chips. This episode is steeped in tech exploration and teeming with fascinating insights, so grab a seat as we embark on this tech adventure.

Want to scale your edge compute business and learn more? Subscribe here and visit us at https://edgecelsior.com.

Pete Bernard:

When you ask people what Edge Compute is, you get a range of answers Cloud Compute in DevOps, with devices and sensors, the semiconductors outside the data center, including connectivity, ai and a security strategy. It's a stew of technologies that's powering our vehicles, our buildings, our factories and more. It's also filled with fascinating people that are passionate about their tech, their story and their world. I'm your host, Pete Bernard, and The ED GECELSIOR Show makes sense of what Edge Compute is, who's doing it and how it can transform your business and you. So let's get started. Cool. Well, bob, thanks for taking the time. Yeah, really appreciate it. We met at Mobile World Congress at that kind of weird Italian restaurant. What was that? I didn't know the name of the place.

Bob O'Donnell:

Gettros.

Pete Bernard:

I had those monkey paintings on the wall.

Bob O'Donnell:

Yeah, things have been used by people for CES and other Vegas convention meetings for literally decades, not just years. Oh, I was going to say generations.

Pete Bernard:

Yeah, well, that probably true. It seemed like it was a little time capsule in there, but anyway, no, it was great to meet you there and that was actually my first show as kind of an Edge Celsius person and some sort of waiting into a lot of this analyst and consultancy business. But I know you've been doing it for a long time. I was reading up on your bio and in terms of the time you spent at IDC I think 14 years there you spent 10 years with your own firm. You know you've been a big influencer for virtual reality and AI and all these things, so it's kind of goes without saying you've been on the cutting edge of a lot of this work for a long time. Yes, maybe give me your version of the intro, since I'm kind of just regurgitating your website.

Bob O'Donnell:

Yeah, no, I mean I have been a. You know, I started out in tech related press. I originally started doing music technology stuff. I'm a musician and that's still my passionate hobby of a band. I play guitar, I play trombone, I got keyboards all around me and then I got from that into computer publications and from computer pubs I got into being an analyst. And yeah, I started at IDC a long time ago and and that was there for 14 and a half years and then decided to do my own thing and it has one month from now it'll be 10 years at Tech Analysis Research. And then, along the way I did, I did my own Colin radio talk show on an ABC affiliate here in the San Francisco Bay area and I've tell you I write for USA Today and Forbes and a whole bunch of other places. I've done a bunch of TV for all the big networks Bloomberg, yahoo Finance, cnbc, et cetera, et cetera.

Pete Bernard:

So awesome. So yeah, I mean we're only recording audio here, but if you saw Bob, you might recognize him from those talking heads things. But and you mentioned also, so I play guitar. That was one of the things I had been talking to Leonard about was kind of I don't know if that's a thing like people that are into tech or playing music.

Bob O'Donnell:

There's a lot of people in tech who are also musicians Absolutely.

Pete Bernard:

Yeah, I saw your website for the headliners.

Bob O'Donnell:

Yes, the horn band.

Pete Bernard:

Yep, that looked really awesome. Yes, thank you. Impressive. So it sounds like you guys have some pretty regular gigs too.

Bob O'Donnell:

Ah no, we're doing all right they. There's a gig actually coming up in a couple of weeks. Unfortunately, I will be out of town, so we had to get a fill in for me.

Pete Bernard:

but but yeah, but it's yeah.

Bob O'Donnell:

I started the band over 20 years ago specifically to be a Chicago cover band, because I grew up in Chicago so I was a huge Chicago fan. And then Blues Brothers and then we did. We branched off into all kinds of stuff, but it's all horn band based stuff. I've written original tunes for us as well, and so, yeah, we have a lot of fun. It's a great time Cool.

Pete Bernard:

I've been playing up here with some guys for about seven or eight years. I mean, I've been playing in bands since, you know, I went to school in Boston in the eighties, nineties, so. But these guys have been playing with for a while. It's been a lot of fun, it's it's a great exercise for a different part of my brain and it's creative and I don't know. It's some kind of good visceral thing you know. Get into it Absolutely. Get out from behind the keyboard every so often. That's cool. Hey, so I saw that you published a document recently. A paper around ran.

Bob O'Donnell:

Yes.

Pete Bernard:

And not not to get too much in the weeds here, but it was interesting because you you were talking about and I did a little, I did a little tick-tock recently on sort of D? Ran, c ran, v ran, o ran, the whole evolution, and it's kind of like we're on this track or train of the virtualization of everything.

Bob O'Donnell:

Yes.

Pete Bernard:

And maybe because, as semiconductors are getting faster and networks are getting faster, you know, in general people are able to now do things in software and in different sort of layers of the cake Right Then back in the old days, you know you pretty much had to hard code everything to a custom piece of breadboard. You know that you put together and now it's all becoming virtualized. I mean, do you see that as a as a as a meta trend? That's happening not only in Telcob and other places.

Bob O'Donnell:

Well, yeah, I mean, it started in the computing world, right? I mean, if you think about it, we went to, I mean, even mainframes and terminals, for gosh sakes is a sort of a form of virtualization. You're virtualizing the computing resources of of that mainframe. You didn't really virtualize it back then, you actually physically shared it. It was not virtualized, to be fair, but Right, but yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah, yeah. Quickly it was realized that hey, there is all these computing resources. How can we use them more efficiently? Like nothing worse than having all these compute resources and then finding that it's only being used 20% of the time. Right, so how do we improve that? And so virtualization became a thing to, to leverage that. And virtualization has, you know, become the thing and that's how almost everything is done in the computing world.

Bob O'Donnell:

And, frankly, the telco industry is about 10 years behind on figuring out this idea of virtualization. Because the thing there there's a couple of things that are different, and that's why part of the reason why telco has moved at the pace that it has Number one, I mean, first of all, telcos are utilities. They're not really tech companies. A lot of people tend to think they're tech companies and they certainly have tech aspects, but they're a utility, and I mean that in the sense of they need to be there and be reliable like a utility. You, you need to have that service all the time. Computing has gotten significantly more reliable than it used to be, but you know if, if, if you know somebody's server goes down, you're going to live right. You'll come back when telco networks go down. Literally people's lives were at stake, so it's just been a much slower moving process because of that.

Bob O'Donnell:

And the truth is there were a number of companies had dedicated hardware the Ericsson's and Nokia's and Samsung networks and Huawei's of the world that did a good job of doing what they were doing. But there was this recognition hey, we can move this stuff into software, as you, as you mentioned, pete and and then really get into um leveraging. You know what they call Cots hardware, commercial off the shelf hardware, meaning you could buy a standard server from a, a Dell and HPE, a Lenovo, uh, and then run this software and then achieve the same things and then, oh, by the way, not only does it duplicate the functionality, you can do it a little bit more efficiently from a power and cost perspective and, most importantly, you can automate it, and you can automate certain processes, um, and what have you? And you could leverage software written, you know, for one place in other places. So there were a lot of reasons why it makes sense to do it. Yeah, it's just taking a long time.

Pete Bernard:

Yeah, and I think I mean Telco, like you said, is, um, it's a tend to tends to be kind of a risk averse business. Yes, Um, they had an existing system in place for a long time that worked 24, seven, yep, you know 365.

Bob O'Donnell:

That's exactly what they need.

Pete Bernard:

You know getting in there and, uh, kind of blowing it up and starting over again. It was probably not something anyone was really looking forward to doing. Yeah, but that, and that's why it probably has taken a while. But. But like you said, it's uh. When you get into software and virtualization, you get into, you know, a lot of risk mitigation. You get into a lot of uh, you know vendor choice and uh flexibility, yes, and you get into, you know a lot of you know theory. Explain the agility in terms of adapting, uh, uh, adapting, you know, new services and new capabilities on the fly without reinventing the hardware. So it's got a lot upside, but it's probably a lot easier to do it when you're just getting started, as opposed to you've been running a network for 50 years.

Bob O'Donnell:

Well, and that's yeah, and and exactly to your point. I mean, that's what we've seen people uh do. The most rapid adoption has been in these what are called green field networks. You look at Dish here in the U? S. You look at what Rakuten did in Japan, uh, and some other places where they didn't have this huge existing infrastructure that they had to work with and modernize. And you know, in the U? S and other parts of the world you've got telcos that have been there for decades, um, and modernizing that is a lot harder than building from scratch.

Bob O'Donnell:

And so you do it piece by piece, bit by bit, and that's what we're seeing happening.

Bob O'Donnell:

It's taking a while, but it is indeed getting there and it's opening up new possibilities. And, and some of the advanced capabilities for 5g, for example, you know, are going to be easier to uh deploy when everything is software defined, sure, um, and then eventually, as you said you know you talked about the various O, various ran prefixes oh, ran is the other big one open ran versus virtual ran, and open ran takes virtual ran even a step further by saying, hey, we can interconnect, yeah, we can mix and match. Uh, mind you, I think what a lot of people are finding is. It's one of those things that sounds great in theory and not necessarily as good in practice, cause, again, because of the reliability and what have you um, there's this sense of hey, I need one throat to choke if something happens and goes down, and if I'm intermingling too many different vendors and too many different parts, you know everybody's going to point fingers at each other and be like sure you're not quite sure what's going wrong.

Bob O'Donnell:

Yeah.

Pete Bernard:

Yeah, no, I agree, it's a. It looks good on PowerPoint. I mean, and you're seeing some of the telcos, the challenger telcos like dish and other folks embrace it. Um, I wouldn't say because they have nothing to lose. They have a lot to lose, but it's at least they. They maybe are more willing to take chances and try to get the competitive advantage than maybe the, the, the, the, the, the geo leader in the telco space. You know whether it's a Vodafone or whoever that's, maybe you? Know, probably a little more cautious on their solution.

Pete Bernard:

But eventually it'll get there. I think there is a lot of work to do to make sure stuff works together and the right performance and the right quality, yep, and they can solve that problem and then then maybe that'll accelerate. But uh, and I would say also the virtualization I was curious what your thought is on this too is around uh vehicles. You know software defined vehicles and other traditional metal bending business where they're now thinking about, wow, software is important, um, and you know virtualization and we can get into the architecture, but ultimately going from a bespoke manufacturing mode into more of a software mode where things are mixed and matched on the stack, you're seeing those pains happen in the automotive industry as well.

Bob O'Donnell:

Absolutely, and and you know it's because people are looking for more capabilities in their cars. I mean, cars are, some have argued, is sort of the next mobile device. Uh, in a, in a smart, smart mobile devices, your car, and you know, and there's some truth to that. Obviously, you know, later generation cars do have some pretty impressive tech in them at this point, and people look to them for all kinds of activities, um, and information and what have you, and so there's been a lot more effort to do that. And and then the other thing is because the, the guts of cars have gotten unbelievably complicated, and so there is this sense of down the road, there needs to be some level of simplification, and so if you can do multiple functions in sort of one, what's called ECU or electronic control unit, that's thought, that's the the heart of how cars are constructed these days. Um, that can you know, save you time, money, energy and effort, and so companies are starting to do it.

Bob O'Donnell:

One of the big problems and almost nobody really understands this is that the way car companies are actually organized, like the structure of the company, is that they were structured around these EDUs, so there was groups of people whose job was a particular EDU. So when you talk about virtualizing EDUs, you're basically talking about taking people's jobs away because they're like well, wait a minute, if the functions of my EDU get subsumed into this other thing, then I lose my job. I don't think that's a good idea. So it's been very interesting to see that transition process and that's part of what we've seen with. To be honest with you, as we move to electric cars, which a lot of them are software defined some of the challenges in terms of how they're constructed and how they're built and what have you, and that's caused some of these labor issues.

Pete Bernard:

Yeah, yeah. Well, I think today apparently the UAW got to some agreement with GM and other folks.

Pete Bernard:

But I think that you're right, it's a cultural issue. A lot of times when companies are adopting technology, there are underlying cultural issues that can really affect the adoption and the speed of adoption and the success. Frankly, because you have all these antibodies in the organization, Like you said, maybe you're having different motivations around the tech and even though it all looks good on paper, all makes sense, the CEO is scratching his head or her head about, like why isn't this working? It's like, well, your employee base is freaking out about job security and relevance and stuff like that. So, yeah, now it's interesting with all these, the virtualization of things, the clarification of things. When it hits the ground, where the rubber beats the road, that's where it gets really interesting.

Bob O'Donnell:

Yes absolutely.

Pete Bernard:

Hey, quick shift. Use an automotive metaphor, we'll shift. You were at, were you at the Snapdragon Summit? I was, I was, I was in Hawaii. I think Leonard said he was going to be doing some jamming with you guys. I don't know.

Bob O'Donnell:

We did. We actually played a little guitar well into the evening after a few cocktails. But yeah, we had a good time Awesome.

Pete Bernard:

Yeah, I actually talked to Megadaga last week around. She's a Qualcomm IoT kind of AI person, senior director, so we had a good chat about edge AI and on device AI. Yes, qualcomm's doing, and I know that was kind of a big theme there. There's a lot of big themes going on, but one of the interesting things was, of course, was the Orion processor announcement and sort of. You know, qualcomm really again used my automotive metaphor pedal to the metal with their race to build, you know, super high performance semiconductors for PCs and other compute devices. And I saw there was a lot of chatter about AI PCs. Right, and you know I think you've been around the block a few times as of I, I worked in Windows for a while. I've actually worked on the original Windows, on Snapdragon project, but you know you go back to AI PC. You know there's like cloud books and net books and like is AI.

Pete Bernard:

PC. Do you feel like that's a thing, or is that just a hashtag mashup or what's?

Bob O'Donnell:

Well it's, it's a little bit of all the above, to be honest with you. I mean, some of it is is clearly a hype thing, but look, the fact of the matter is there are a lot of good reasons why doing some of this generative, generative AI type of stuff that we're starting to see on on the device makes a ton of sense. One of the things that I mean again, jenny has just hit the world, you know, with an unbelievable force and people are scrambling to build, to buy as many GPUs as they can and build as many big, you know data farms and server farms with stuffed with Nvidia GPUs as they can in order to power this stuff. And if you project out, there's not enough power and a not enough GPUs in the world to power these things. So, realistically, you have to think how do I divide and conquer? And one practical way to do that is to do some of the work on the device. And the problem was everybody thought, oh, it's, we're not going to be able to do this, it's, it's, it's too hard to do, there's not enough compute power on these devices.

Bob O'Donnell:

But what we've started to see and Qualcomm's been a part of it, but so is Intel and AMD and other folks are talking about how hey we can actually build, you know, these SOCs, the system on chips that incorporate multiple components, most notably the what's called the NPU or neural processing unit, and actually do some of these things on the device, and that all of a sudden changes the picture. And so what was interesting at the Qualcomm event was they had this new partner calling the Snapdragon X Elite, and it has two main new features. One is the Orion CPU you mentioned, which is pure raw CPU performance, but the other big thing was they doubled up on their hexagon NPU and they're talking about the level of performance it can do for these AI applications on a PC, and we've already had some of that on phones up till now, but we haven't really had much of it on PC. So this is an interesting new opportunity to do it, and because a lot of the big applications being used right now are text based, it makes more sense on a PC, where we're used to using a lot of text, and even the image based stuff image creation and editing is often done on PCs too, so the idea of doing more of that on a PC is actually very appealing.

Bob O'Donnell:

Now there's a lot of challenges you worked on the first. You know arm windows on arm stuff. You know compatibility. That problem is not going away, right? All those x86 apps that are not going to get ported to native arm are still out there, and how well they work is one of the questions that Qualcomm's asked us to answer. But for now they just had this blowout of hey, look at the raw performance we have both on the CPU and on the MPU. This is, you know, a new thing and, like you said, we're putting to the pedal, to the metal and we're going to try and reshape the PC market, and it's going to be interesting to watch.

Pete Bernard:

Yeah, no, I think it's fascinating. I mean, I think a lot of the x86, the long tail x86 apps with emulation usually just work fine, I mean, especially with that much horsepower.

Bob O'Donnell:

Yeah, well, and that's what everybody's going to be waiting to see. But you're right. I mean, like you were talking about music, none of that. There's like no music, apps that are going to get ported to arm for a long time, and drivers for devices and everything else. So you know, will those things work and will these other specialized apps work? That becomes the big question mark.

Pete Bernard:

Yeah, the specialized apps. Yeah, especially that's kind of the long tail of stuff is going to be challenging. Yeah, I thought it was interesting. I mean I actually did a paper and a couple of videos on this. I called it AI's environmental Armageddon, so it's sort of like the the appetite for AI workloads and, as you said, like building data centers and powering data centers and cooling data centers. Yep, like it just doesn't. It just doesn't add up. Yeah, like there's just not enough ground water and electricity, right, to do everything every all the cat poems that people want to generate with chat, gpt, right. So if we can move some of those workloads to the interim edge or the far edge or the PC, then you know why not. Yeah, with Qualcomm doing it in a much more. You know, I call it inferences per second, per watt. Yeah, you know, I don't know what that is, but that, that that kind of metric where you're looking at horsepower per power used. Yeah, I think Qualcomm probably has an edge over most companies in that space.

Bob O'Donnell:

Well, that's obviously. That's certainly what they're claiming. And then and the question will then we'll be getting support Because, as we all know it's you can have the greatest you know chip in the world. If there's no software that works for it, doesn't matter. So it's all going to come down to not only delivering great hardware and a great platform, but then convincing developers to do what they need to do. They certainly have the advantage that Microsoft is very much behind them, is pushing hard and will be delivering a AI capable version of Windows that runs on these devices. They've been talking about that publicly for a while now, so that's going to be interesting to see what that does and how that happens.

Pete Bernard:

Yeah, no, I think I mean Microsoft should be able to do something interesting there with, especially with some of the native apps and teams and some of the other things that they have as far as third party developers. Yeah, I mean TBD it's. We'll see what happens, right.

Pete Bernard:

I think, and speaking of. We'll see what happens. I think in like an hour or so is the Apple event Yep, like five o'clock. So they'll talk about the M three, which my bold prediction is it's way faster than the M two. That's my prediction, and so that'll be interesting to see how that now stacks up against what Qualcomm talked about last week?

Bob O'Donnell:

Yes, exactly, and they have a great way.

Pete Bernard:

It's speaking of music and other apps you know for content creation and stuff in the in the Apple universe. They have a great handle on developers there. So yes we'll see where the right, where the real AI kind of breakthroughs show up, on what platform first. But yeah, makes for interesting times.

Bob O'Donnell:

Yes, indeed Indeed.

Pete Bernard:

So other question I had for you, and I don't know if this is kind of in your bailiwick or not, but you're talking about risk five, yeah. You know we're just talking about the new via core, which is ARM based, and Qualcomm, and all that good stuff and all the interesting politics behind that. And then there's this thing called risk five. You know, you know which is this kind of alternative architecture that's, you know, license, free license quote, unquote, royalty free, royalty free.

Pete Bernard:

But then you know we've got some startups trying to do some things in that space. A lot of folks I know is like NXP and Schneider and a few other companies are getting together as a consortium to do some risk five stuff. And then there's companies like sci five and who had sort of a very unpleasant implosion last week and kind of laid everybody off, or the vast majority of folks. And so it's sort of like well, where is risk five really fit in here relative to? You've got obviously, intel and x86 and AMD is doing some cool stuff in x86. You know you've got ARM which is now IPO'd with all that whole ecosystem, whether new via as part of that ecosystem or not, it's unclear. It's sort of like the same instruction set but whatever. And now you've got this risk five and like is there enough oxygen for risk five to even turn into something there? I don't know.

Bob O'Donnell:

Well, look the way I would are. Risk five, I think, has a lot of opportunity on the low end, on microcontrollers. I mean, a lot of people don't know this about ARM, but there are. Arm actually has three product lines the a series products, which are the high end stuff that goes in the smartphones, the R series and then the M series, which are the microcontrollers. And risk five is primarily being is challenging ARM on that low end and the microcontrollers and that's, you know, that's in everything from you know, a camera to a toothbrush and all these kinds of all these things now have little microcontrollers and so billions yes, exactly, and that's where risk five is a really interesting option.

Bob O'Donnell:

It becomes a lot harder to do it at a high performance scale and again, it's not necessarily they can't design a chip. Somebody might be able to do a risk five implementation that, from a pure you know numbers perspective, looks impressive, but it will literally take a decade to get the kind of software that you would need to do anything useful on a high power device, and that's why it's not really an option for the high end. It's a tremendous option for other types of devices and for microcontrollers, and I think that's where you're going to see a lot of traction. So that's my take. I mean, yes, there is opportunity for risk five when we think about the entire universe of computing devices, but probably not in the rarefied era of the x86 and the high end stuff.

Pete Bernard:

Yeah, no, I would say that's fair. And yeah, the microcontroller space. By the way, the R course for arm are real time real time.

Bob O'Donnell:

Yes, I just skipped over that.

Pete Bernard:

I'm digging into that and you'll see, like, like the Nvidia Oren, which is a couple of big a 78 cores paired with an R core, right what they call a safety island, so you can do like hard real time deterministic stuff as part of your app processor thing.

Bob O'Donnell:

Yes.

Pete Bernard:

Fascinating how things are getting mushed together like that.

Bob O'Donnell:

It is.

Pete Bernard:

That's a whole other discussion. But, yeah, I love R cores. I think they're cool. So, yeah, no, I think I think it'll be interesting to see how that pans out. One of the things, like you said, the software. But also I think there's an interesting kind of deconstruction going on with semiconductors in and of themselves, right, there's things like chiplets and other kind of building blocks where you're seeing some of the big industrial folks you know doing their own custom silicon. Yes, right, and I wouldn't be surprised if maybe we've got Tesla in there but other automotive folks building their own silicon over time and not relying on the latest whatever, and I think it might, might be getting in that direction. I'm not sure if you have any take on that. Do you think that silicon will start to get a little more fragmented and customizable per vertical?

Bob O'Donnell:

I would argue that it already has. I don't know that we're going to see everybody become a silicon vendor, because it's hard. It's really really hard, not just to do it but then to maintain it. So I honestly don't think we'll see too many car guys do their own semiconductors. They'll customize some arm cores, they'll do some other things perhaps, but that's about, I think, as far as most of them will go. But if you look at other big tech companies Microsoft doing their own chips, google doing their own chips which has already happened Amazon doing their own chips, which has already happened so when you look at the big tech companies who have a vested interest in the silicon in a way even more than the car guys do, I think you're going to obviously we've already seen it and I think you're going to see an expansion there.

Bob O'Donnell:

Who knows, maybe we'll see some of the car guys start to do their own silicon. I'm sure they're thinking through it and trying to figure out what's the best way to do it. But I also think they know their limits and they have to be very cautious on that, because they could end up spending a lot of time and money on an effort that in the end, isn't that useful.

Pete Bernard:

They're still trying to figure out how to do software. So they are. Get that first. But you're right, silicon is a five or 10 year thing. You need a lot of scale to make it pay off. The ROI. I do think that the cost and the overhead will start to come down with things like chiplets and other things.

Bob O'Donnell:

Yes, well, yeah, exactly. I mean, the other thing is there'll be a different layer or a different level of semiconductor company where they can literally sort of do a Lego block style thing. So you've got companies who've done traditional chip design, like the cadences and synopsis of the world, that have these applications now that are designing systems, not just the individual chips. That's what they used to do, the EDA tools, as they call them, where the inner guts and workings of a chip, and now they kind of expanded out into systems. And ARM is doing this too. They have this thing called ARM Total Design where they're helping companies piece together a variety of technologies but they can pick the Lego bits that they want to get their finished model completed, and that, I think, is going to be a lot more common for companies Well, and also more fab capacity coming online.

Pete Bernard:

I mean all the stuff they're building in Arizona and other places. They got to keep those full capacity, so you might see some good deals out there for fabbing a million of something or a million of something else, I don't know. And then also, I think, with EDA, how do you infuse and I'm sure this is happening already AI and CISM EDA process? They are already doing that pretty quickly and just put the blocks together and spit it out.

Bob O'Donnell:

Yeah, they're doing it for better designs. They're doing it for more power, efficient designs. But yeah, they're deploying AI and generative AI stuff already and it's having a huge impact. Fascinating yeah.

Pete Bernard:

I had a kind of a meta question for you. Being in the business and doing all these things, like you have a broad portfolio of things that probably throughout the day, you're talking about, thinking about, writing about. How do you kind of organize your head.

Bob O'Donnell:

I love what I do and so all this stuff is interesting and fascinating to me and I love making connections across things, and that's just kind of how my brain works. And part of the reason I left IDC as much as I enjoyed my time there, which I thoroughly did was I wanted to expand to bigger and broader things and tell broader stories that connected more industries, more people, more companies, because I thought there were a lot of things out there that weren't being told. Because with the big analyst firms you tend to be very specific and very deep, which is great and there's absolutely a need for that. But I thought there was also a need for a different kind of perspective, a broader perspective, and that's what I've focused on, and so I just it's kind of how my brain thinks and I get bored easily, so that's good and I like to jump around and I'm able to do that and then figure out how this stuff fits all together.

Pete Bernard:

So yeah, no, I hear you on that. I mean, one of the things I always like to do is look at the combination of technologies. You know, how does 5G and AI and Edge, you know, come together? And it's like the chocolate and peanut butter analogy.

Bob O'Donnell:

It's the tech buzzword bingo. You know to think get all those things working together, but yeah.

Pete Bernard:

Yeah, but then it's like they all have impact on each other too, and so sometimes like we were talking about AI and now EDA and chiplets, and now all of a sudden things get combined together and enables a whole new way of, you know, building and delivering tech. Absolutely true, good Well, bob, again thanks for your time, you're welcome.

Pete Bernard:

Happy to have joined you you seem to be extremely busy, and so I appreciate you carving out a little bit here. Yeah, and I see your guitar in the background. I don't have my guitars in my background.

Bob O'Donnell:

It's all good.

Pete Bernard:

There's a Martin back there. But yeah, hopefully we'll run into each other again at one of these events and you know that's good.

Bob O'Donnell:

I appreciate you having me on.

Pete Bernard:

Thank you very much. All right, thanks, bob, you care. Thanks for joining us today on the Edge Salsa Show. Please subscribe and stay tuned for more and check us out online about how you can scale your Edge compute business.

Edge Compute's Transformation Potential
Software and Virtualization
Future of Risk and Customizable Silicon