theCUBE + NYSE Wired: Physical AI & Robotics Leaders | Diane Bryant, Broadcom, Celestial AI, Haemonetics, mmTron

Clips
More from theCUBE + NYSE Wired: Physical AI & Robotics Leaders

Diane Bryant

Chairman, CEO, Independent Director

play_circle_outline The importance of chip, software, and geographic considerations in AI development is highlighted.

play_circle_outline AI technology typically requires a decade for widespread adoption and infrastructure readiness.

play_circle_outline Custom GPUs are essential as companies seek alternatives to NVIDIA's dominance in the market.

play_circle_outline Exploring Historical Parallels: The Interplay of Hardware and Software in the Evolution of AI Technology Waves

play_circle_outline Unprecedented Growth in Data Center Investment: Exploring the Shift to Private Clouds and Sovereignty in Infrastructure

play_circle_outline Demand for optical interconnects is driving innovation in semiconductor technology at Broadcom.

play_circle_outline Leadership and focus are critical for achieving success in semiconductor manufacturing and technology.

Info
Transcript

Diane Bryant, Broadcom, Celestial AI, Haemonetics, mmTron

Diane Bryant

Chairman, CEO, Independent Director

Diane Bryant, independent director for Broadcom Inc., Celestial AI Inc., Haemonetics Corp. and mmTron Inc., joins theCUBE’s John Furrier and Dave Vellante at theCUBE + NYSE Wired: Robotics & AI Infrastructure Leaders 2025 event. With decades of leadership in enterprise tech, Bryant offers a strategic view on how AI infrastructure is evolving across hardware and software domains.

The conversation explores semiconductor innovation, the critical role of supply chain diversity and how past technology cycles inform today’s AI boom. Bryant also reflects on... Read more

explore Keep Exploring

What is the current state of innovation and challenges in the semiconductor and software industries, particularly regarding geopolitical factors and the development of sovereign clouds? add

What is the significance of AI in technology adoption and its impact on infrastructure? add

What are the requirements and considerations for developing hyperscale solutions in computing? add

What are the challenges and complexities involved in developing AI technologies and how does this impact the competitive landscape among tech companies? add

What benefits does Broadcom gain from developments in AI and cloud technology? add

What advancements are being made in optical interconnect technology and its relation to computing hardware like GPUs and TPUs? add

What factors are important for a company to successfully innovate and grow in the technology industry? add

bolt Powered by CUBE AI

Diane Bryant, Broadcom, Celestial AI, Haemonetics, mmTron

search

>> Welcome back everyone to theCUBE Studios here in Palo Alto, California. I'm John Furrier, host of theCUBE with Dave Vellante, my co-host. We're here for three days of wall-to-wall coverage around Robotics & AI Leaders. We have a leader here in the industry. Diane Bryant, who's an independent director at Broadcom, Celestial AI, a bunch of other firms, 32 years at Intel, running 99% of the server market at that time. Great to have you on. You've been a leader. You're leading companies now at the board level, so thanks for coming on. I really appreciate it.

Diane Bryant

>> Oh, thank you. It's a lot of fun.

>> So the big conversation is chip, software, geography. It seems to be the theme. Sovereign clouds come up. Countries are looking at creating data centers for competitive advantage, but the action at the semiconductor level is very hot right now, and obviously software and hardware working together. This is your wheelhouse. You've been in this industry. What's your take right now? Because there's innovation going on, it's global, the supply chain challenges. What's your assessment of where we are with the AI wave pushing the demand curve?

Diane Bryant

>> Well, I mean, my goodness, you can't go anywhere today without talking about AI. I will start by saying every technology takes 10 years to get to adoption, and there's so many proof points of that. For instance, I created the silicon photonics, the optical interconnect business at Intel was 2015, and now 2025, 10 years later, demand for optics explosion. Everyone is moving their infrastructure then compute to optical because you could only run electrons over copper at such speed and then it'll fall apart. So what is fun about AI, from my perspective, just my background, is that the entire value chain, so you can start by saying AI and all the wonderful things that can give enterprise and consumers, but then you back up and say, "Okay, well, if you're going to have AI, then you need massive amounts of compute. You need massive amounts of storage, you need incredibly high bandwidth, fast interconnect optics." Then you back up and say, "Okay, now you need custom GPUs," because the world can't rely on NVIDIA. Well, Jensen's genius, but as wonderful as NVIDIA is, people need choice, right? You're never going to get locked in. And so now you have these hyperscale solutions that are all inventing their own accelerator solutions, which goes to the Broadcom board. Obviously Broadcom is doing very, very well in the AI.

>> We had Vijay out earlier. He's now doing sales marketing strategy for the chips they're doing, Tomahawk 6 just launched, good performance.

Diane Bryant

>> Oh, excellent performance, don't say good.

>> Amazing performance. It is excellent.

Dave Vellante

>> And people missed Broadcom. We didn't because I mean we understood it I think pretty well.

Diane Bryant

>> Missed as in?

Dave Vellante

>> Missed in terms of the number two. Sometimes you could even argue number one AI company because the stuff they do is so hard.

Diane Bryant

>> Well, and that's the point. The stuff they do is so hard. Well, AI is very hard, whether you're talking about the algorithms, you're talking about building the data models and cleansing the data in order to run the algorithms and then the compute to run on it. And because it is so complex, only the really big highly talented companies can do it. And that's why you see Google was the first to have a custom TPU and Amazon and Meta everyone, right? It's complex. And then the beneficiary of all of that is Broadcom.

>> Diane, one of the things Dave and I always talk about, because we're kind of historians because we're old, but we always talk about this wave waves of technology. So the AI wave gets compared to the internet. We say it's the PC era combined with the internet times a hundred because it's got a lot of change to it. But the PC era that we lived through and you drove had processor improvement, Moore's law was kicking in. I remember the original processor, the 286, 386, 486.

Diane Bryant

>> I was at 486. I started Intel at 486.

>> I think that created the server-

Diane Bryant

>> I did the IO buffer.

Dave Vellante

>> That was a steep part of the S-curve right there.

>> That made the server market happen. I remember I was working at HP at the time, but every innovation had better Windows and better server software. So the software was matching pace with the hardware at that era. We're seeing a similar parallel now. You got NVIDIA and all the chips coming out. As it gets better, the software gets better. So that kind of relationship flywheel gets consistent As someone who's lived through that era and then you put the internet and mobile and everything and cloud on there, we're seeing new stuff come out. How should companies be thinking about this? Is it backwards compatibility? Is it ride software wave because you have that software hardware dynamic going on? What's your feeling on this? Because you've lived through that, you're seeing almost the same movie.

Diane Bryant

>> Yeah. It's a flywheel. Yeah, it's absolutely a flywheel. Okay. So the first time I got up on stage and said the words artificial intelligence was 2013, and it was because we now had the cloud. And so now all of a sudden you have the infinite, the cloud, infinite compute and storage to now run all of these algorithms and these incredibly large databases. AI was invented in the '50s and it was sci-fi because you didn't have the compute and the storage to actually do it. It was conceptual, but you didn't have the infrastructure run it. And so I had a GPU, Xeon Phi, unfortunately it's no longer at Intel, and they think of today if there was Intel Xeon Phi. But anyway, we launched Xeon Phi, and so now you have the compute power like the GPUs, and we debated how are we going to launch this thing because at the time, the only solution out there was Watson. The first solution was Watson in 2011, and IBM called it cognitive computing. So we said, "Well, we can't call it cognitive computing. Why would we help IBM and bolster their brand? We can't call it artificial intelligence because everyone thinks of that as sci-fi crazy." And so the marketing team at Intel struggled and struggled and struggled, and we just said, "Forget it. Let's just call it artificial intelligence." And got up on stage, Google Cloud, Diane Greene came on stage and we talked about now artificial intelligence is real. But that was 2013.

>> And now fast-forward today.

Diane Bryant

>> And now here we are. So it's that decade. I always say this 10 years, it always takes 10 years for when technology is viable, feasible, available to when it is truly adopted at a level that you can, and it takes forever.

>> If you can go back and do a replay at that time, what would you have done differently at Intel if you had the powers to say, "Hey, knowing what you know now, what could have happened?" Because I know Intel, a lot of stuff spun out of Intel. A lot of people had started companies, so Intel was directionally on path for this.

Diane Bryant

>> A lot of things we look back at Intel, a great company, an incredible technical leadership, and a lot of times we were just ahead of the market and then as Andy Grove would tell us, "You got to have the persistency, and I don't know, the courage to just stick with it and the market will follow because adoption technology is hard." So Xeon Phi, I launched in 2013, so we had our GPU. First year it was launched, we had 10 of the top 10 supercomputers running, not on NVIDIA, but on Xeon Phi. And then we created Omni-Path, the interconnect force. So high-speed interconnect, and then I left. And the good news is Omni-Path survived. I'm very, very proud. Lisa Spelman, a good friend of mine, she's now leading Cornelis and thank goodness they convinced Intel to not destroy the value and destroy the product line, but allow it to live outside of Intel.

Dave Vellante

>> So as John said-

Diane Bryant

>> That's great news.-

Dave Vellante

>> we're historians. We've seen all the waves. History repeats itself. It sort of echoes. So IBM of course, is famously inadvertently handed its power to Microsoft and Intel, known as Wintel of course. And now we're seeing NVIDIA, the rise of NVIDIA. They had the hyperscales come in. Of course, NVIDIA is like Wintel with CUDA. How do you see that playing out? And obviously Broadcom has found a way to compete with networking and custom silicon and open scalable and power OSP.

Diane Bryant

>> Well, I mean, yeah, so I'm a historian too. I may be older than you, but yes, exactly that. When we launched the Xeon Phi, so if I just go back in time for a second, when we launched Xeon Phi, it was immediately adopted by who? High-performance computing, scientific computing, people that own the entire software stack, they own that stack, so they're not beholden to CUDA and NVIDIA. They can port their stack over, they own the whole thing. So it's not the Wintel lock, and that's what you saw. So now fast-forward and the hyperscalers, how quickly could they move off of NVIDIA because they own their software stack, and so you see the hyperscalers and being able to instantly over 10 years, but instantly develop their own accelerator or own GPU or TPU or XPU, whatever you want to call it, because they own the software stack. That is always a lock. You're absolutely right. Hardware, software lock. But if you own your software stack, you port it and you move and you move to the best solution. Enterprise, I was CIO of Intel for four years, ran IT. That was probably one of my best jobs., I have to say. I didn't want the job. I was telling Orlini, "Please, no, please, no," because I was running the server business at the time.

>> 99% of the market share.

Diane Bryant

>> Because you'll be CIO and I'm like, "I don't want to be CIO," but it really was the best job. We had to eat our own dog food. We had to-

Dave Vellante

>> You were customer zero.

Diane Bryant

>> Yes. And so I launched our first cloud in 2011, so this is another 10 year. I'm probably jumping around and you want to take over here?

Dave Vellante

>> All good.

>> It's all good.

Diane Bryant

>> It was 2006 when Google announced the word cloud. Hopefully you remember. And it was 2006 that Andy Jassy launched AWS, right? In 2011, we were all trying to get enterprise to adopt cloud computing because this was the big unleashing of infinite capacity. And so I thought, oh, I'm the CIO, I better deploy a cloud. And I think I was the first enterprise in the world, certainly Fortune 100 company, to deploy a cloud to prove the technology. Intel is running on a private cloud on-prem with VMware and with Linux. We did two clouds. But that was 2011 and still in 2016, if you said the word cloud, there was a survey done actually. If you said the word cloud, if you said to someone, "Where's cloud computing? Where's your data stored?" They'd all look up, the cloud. Right now, enterprise is actually deploying private clouds because now it's kind of a commodity. Cloud computing, you just deploy it.

Dave Vellante

>> It's an experience.

Diane Bryant

>> That was a decade. It was a decade later. And that benefits Broadcom. Wait, but that benefits Broadcom VMware. Huge.

Dave Vellante

>> Yeah. Let's explain that.

Diane Bryant

>> Huge back to Broadcom.

Dave Vellante

>> Because a couple things. One is AI's moving to the data and there's a lot of data on-prem number one. Number two is when the cloud first came out, it actually was less expensive.

Diane Bryant

>> Oh, sure.

Dave Vellante

>> Because of the overhead of human labor. But that's gone away because you can get the same experience, substantially The same experience that I'm sure you found is when you built your own cloud with all the automation and the deployment and one click.

Diane Bryant

>> Yeah. I mean the value, it goes back to Jevon's paradox. Anytime you can make something more efficient, you're going to drive demand for it. And I remember talking about VMware and virtualization of servers and the investor, the Wall Street guys were like, "You'll never sell another server again because they're just going to keep loading the apps onto the same server." And it's like, "Oh, you guys are crazy." Now it's so much faster and easier to deploy.

>> They'll buy more servers.

Diane Bryant

>> Demand goes through the roof, right?

Dave Vellante

>> Yes, absolutely.

Diane Bryant

>> Utilization goes from 2% to 80%.

>> You were right on that one.

Diane Bryant

>> You can run a server to 80% utilization now for the cloud.

Dave Vellante

>> And let's do more.

Diane Bryant

>> And then you just keep buying more. So absolutely private cloud. It used to be very expensive to run your own data centers. When I was CIO, we had 400 data centers, very expensive. Now, you don't have to run those data centers at 2% utilization. You can now move to the cloud, run them at 80% utilization and your costs go way down.

>> Right. Yeah. Diane, we were at Mobile Congress past four years with theCUBE. Two years ago, Charlie Kawwas came on, he was talking about the chips, and they would come out with all the new stuff. This year, they didn't have a financial house meeting. This year, he didn't come on because he was doing meetings. We had a lot of meetings with operators and sovereignty private cloud is coming. So we're seeing another potential wave of private cloud on country. So this is a big deal. Broadcom, I think will participate in it with VMware because VMware Cloud Foundation is essentially an on-prem cloud.

Diane Bryant

>> That's what it is. It's a private cloud. It's on-prem cloud. The sovereign data center thing is another part of AI that is very exciting, which is that value chain. So you have to at some point have a data center. So you're building to, okay, now you need to power the data center. So you see the explosion in the market around power solutions, right? Nuclear solutions are back. You have to have the land. So the value chain, in order to make AI happen, there is a tremendous value chain that needs to be created all the way down to the physical building, being powered and cooled. And when I was at Google, we built that huge data center in Iowa, and these things are not simple. It's not like it's your living room. It's not like a plug in the server. They're very complex. And so you just see it's fabulous to watch the growth in everything along that value chain and colo data centers. We looked at that when I was at Google of do we just keep building data centers all around the world or is it better to let someone else do that? And you see a lot of data center companies sprouting out now that are responsible for the physical, and then you plop your server down in their data center instead of having to build.

>> We're very bullish.

Dave Vellante

>> Crusoe on Chase Lochmiller and others with CoreWeave and those guys. I'll share some numbers with you. The data center business, when you think about the hardware, the cooling, the power, it was perpetually about a 200 or $220 billion business. It was flat line went up some years from down. Last year it went from 222 billion to 350 billion, 60% growth in one year. I mean a flat business for decades. And that's just went boom through the roof. And it's going to continue to grow.

>> 10 year wave hit.

Dave Vellante

>> There's your 10-year wave, it'll be a trillion in 10 years.

Diane Bryant

>> There's the 10 year. Oh, absolutely. I mean, you look at just the compute market for AI is projected be between 300 billion and 500 billion in 2030, which just the compute, the hardware, the physical. And so this goes to your point, is there room for everybody? My goodness, if you have a pie that's 500 billion, of course there's room for NVIDIA and AMD and the custom GPUs and just-

Dave Vellante

>> It's not a zero-sum.

Diane Bryant

>> It's not a zero- sum game.

Dave Vellante

>> NVIDIA might get 20% of it, whatever. There's like 80% left over for everybody else.

Diane Bryant

>> Yeah, there's room for InfiniBand, Interconnect, Omni-Path from Lisa, Eltra, ethernet, optical. There's room for .

Dave Vellante

>> Now they might not get the operating margins that you were used to when you were at Intel.

Diane Bryant

>> Oh, I don't know. Technology's not easy.

Dave Vellante

>> Broadcom.

Diane Bryant

>> I don't know.

Dave Vellante

>> Broadcom will because they've got a differentiated model.

Diane Bryant

>> They have a differentiated model and they have-

Dave Vellante

>> Not sure AMD will.

Diane Bryant

>> Oh, I think it's about leadership. Hock Tan is brilliant. He's probably the most brilliant businessman I've ever worked with him. He is, I mean, software, you should be getting a 90-

Dave Vellante

>> That's saying a lot.

Diane Bryant

>> You should be getting 90% margin on software. And if you're not, you got a problem. And he took VMware the fact that it was allowed to run at such ridiculously low margins for a software company. He took it and he's a genius.

Dave Vellante

>> And he doesn't have to pay the cloud cogs. He doesn't have to pay the time. Software companies today, SaaS companies are running at a mid-70s gross margin. And he's right. supposed to run at 90.

>> Hock's got a problem with margins are too high. He's going to have to go back to Wall Street. Why they doing an 89% margin?

Diane Bryant

>> What's that?

>> His margins have gone up with VMware because of the performance numbers I've been hearing in the Valley, roughly speaking, is their biggest profit gain, I think was like five, 6 billion before the acquisition. Now it's way over 10 billion in profit, give or take, I don't know the exact numbers, but it's in that. It's different. It's almost doubled.

Diane Bryant

>> Well, there's even more fundamental reason for that, which is when the cloud was invented, when the cloud was invented, you had all these software companies that went to a SaaS model, right? And you start delivering your rather... So when I was CIO, we bought an enterprise license agreement, right? And we got Microsoft deployed, and then we paid a support and service contract over X number of years. When the cloud came along, the software model flipped and it all went SaaS. You no longer owned the license. You now paid a subscription rateable income recurring revenue, which is a beautiful business level.

Dave Vellante

>> Lifetime value is really real.

Diane Bryant

>> Lifetime is crazy. For some reason, VMware never flipped for some reason until Hock Tan came in and took over, it was still an enterprise license agreement with a support and service absurd in 2025.

Dave Vellante

>> I think it was because they were governed by a storage company.

Diane Bryant

>> No.

Dave Vellante

>> Yes, I do. I think EMC-

Diane Bryant

>> I know-

Dave Vellante

>> I love Joe Tucci. We love Joe, know him well, great executive, but he not a software mindset.

Diane Bryant

>> How long ago did it spin out though of EMC?

>> Well, he took it over in 2009.

Diane Bryant

>> Became an independent Vmware, became an independent company run by-

Dave Vellante

>> No, it was still a controlled company

Diane Bryant

>> Run by-

Dave Vellante

>> It was still a controlled company. They had the little puppet masters. They wouldn't let him spin out vSAN. They wouldn't let him be aggressive there.

Diane Bryant

>> I think you're crazy. It was being led by a CEO named-

Dave Vellante

>> All right, okay.

>> Michael Dale.

Dave Vellante

>> No, Pat.

>> Oh, Pat.

Dave Vellante

>> Joe Tucci couldn't stand up to Pat and say, "This is how we're going to run this company." And Pat won. And that's why you got the basically a low margin software business, but it was great for EMC.

Diane Bryant

>> Wait, so who are you pointing to for not converting over to a modern-

Dave Vellante

>> Pat Gelsinger.

Diane Bryant

>> Oh, okay. Now we're on the same page.

Dave Vellante

>> No, because Pat was more forceful than Joe. And Joe was like, "Ah, it's adding to our income statement. We're a storage company where we're under pressure."

>> He got a good song.

Dave Vellante

>> It's helping our margin. So it took EMC from good to a little bit better. Now Hock has taken that from-

>> Well, Hock, they thought it was a cash cow-

Diane Bryant

>> But Pat could have, in all those years, he was CEO. That was the time when software was moving to a SaaS subscription model. Didn't happen. Hock thought, had the vision, saw the opportunity. And in two years-

>> It's a nice cyclical business too, because when chips changed-

Dave Vellante

>> A hardware person that actually became a really good software CEO is Frank Slootman. He ran Data Domain, if you remember those guys, hardware boxes to do backup and he ran ServiceNow and Snowflake, high-margin software businesses. It can happen.

>> The Broadcom business model-

Diane Bryant

>> Hock did it.

Dave Vellante

>> That's right. Chip guy running that stuff there.

Diane Bryant

>> He has the hardware chip and software businesses. Both of them are on fire.

Dave Vellante

>> It's the business acumen that you talked to earlier.

>> On the focus, Hock Tan's got chips and he's got software.

Diane Bryant

>> Yeah.

>> Software's booming. Chips are booming.

Diane Bryant

>> Chips are booming.

>> But it's cyclical. Software could be up, chips could be down, chips could be up. Software or is it all growing? I mean, is it traditionally cyclical?

Diane Bryant

>> Software is, this is the point of subscription model is then software isn't cyclical. You know every quarter what's coming in and it's a beautiful business. But yes, you're right. So two different businesses growing at a tremendous level.

>> Diane, you're on a bunch of boards. We love that we can talk for hours on the history and all your experiences because they do relate to AI. But now that you're on board, you are advising and also have a hand in mentoring younger companies and also companies that are in the growth mode. There's a lot of unknowns coming. What do you do most of the time? How do you spend your time negotiating deals, advising the management team? What's the focus? Actually, the Broadcom's doing great. That's a great company to be on. So what's the day-to-day operations like for you?

Diane Bryant

>> Yeah, so it's a good point. The board position, there's no two boards that are the same what I have learned, every board is different. And certainly public board director of public boards and startup private boards, you can't find anything in common with them. There is absolutely nothing in common other than you have a quarterly meeting, you have an audit committee and you write minutes, make sure you all stay legal, but there's nothing in common. And when you have a company that is operating as well as Broadcom, as a board member, it's not a hard job. I'll say that. I hope Hock doesn't cut my pay, but it's not a hard job.

>> They're doing good.

Diane Bryant

>> They're doing good.

>> Don't change paths, stay on the course.

Diane Bryant

>> I've also been on a very large public board that we had an activist and we broke up the company, United Technologies, and broke it up into Raytheon, Otis Elevators and Carrier HVAC. That's an event. As a board member, that's very intense. Nobody wants an activist walking through the door and then Humanetics, it's a mid-cap company. It's a med tech company and they're evolving to be more tech centric, which is part of the reason I joined. But then you go to the startup land and I'm-

>> That's fun.-

Diane Bryant

>> super. Oh my goodness. There's something every day going on. And as an old woman, you can contribute quite a bit to it.

Dave Vellante

>> You have wisdom.

Diane Bryant

>> We have wisdom. We have a lot of experience. So Celestial AI.

>> Is it fun or is it more of a playground environment for you? Because I mean, because it is dynamic, so it could be-

Diane Bryant

>> Very dynamic.-

>> high velocity, but is it fun?

Diane Bryant

>> Oh, it's a hoot. It's a hoot. It's open. Look at Celestial AI. So again, optical interconnect on fire. Everybody needs to move to optics and 10-year progression. I created Intel silicon photonics business in 2015. Now it's 2025 plus 10 years. And you have companies that like Celestial AI that are on fire. They've done a magical thing in allowing people to move to optics. So step one, you move to optics is a co-packaged solution. Step two, get it closer to the GPU, to the TPU, to the XPU, and you get that much more benefit from it. So they have the magic of full integration, tight integration with the accelerator chips. It's fabulous.

>> Reliability kicks in. I mean, this is great.

Diane Bryant

>> Lower power, power efficiency, performance, everything. And so that's very fun. And you meet people that you wouldn't normally be... You have the investor community and private boards that you don't have in a public board. So the network, the people you're working with or I always say that I want to join a board-

>> Like our event last night. Like our event last night. You have to mingle and have some fun.

Diane Bryant

>> You had a lovely event. But when I joined a board, I look for boards where it's people that I can learn from that have a different background and different, because it's an intellectual. You want it to be an intellectual role where I'm learning from you, you're learning from me because we've lived different experiences.

Dave Vellante

>> How much of the discussion pie, if I may, is sort of strategy, numbers, security, AI and other?

Diane Bryant

>> Okay, strategy, numbers.

Dave Vellante

>> Numbers. Security.

Diane Bryant

>> Security. In what context?

Dave Vellante

>> It's like cybersecurity, AI and other.

Diane Bryant

>> Well, cybersecurity, get yourself a good CIO and don't think about it. I had a really great CISO when I was CIO, and that's his job. And if we have an attack, you get fired. Well, yeah, hire and fire. You've got to do that as leader. But the numbers, of course, the company like Broadcom, you have to spend a lot of time on the numbers. My goodness. I mean, you're-

>> Supply chain, private financial performance.

Diane Bryant

>> Well, and you're a publicly traded company. If you can't drill in and know every-

Dave Vellante

>> You're double-clicking.

Diane Bryant

>> Yeah, you've got to, right? And Hock's original roles were in finance, so he's got those numbers. There's no worry there. But as a board, you're a governance body. I mean, back to the board position, you're governing, you're not operating, you're governing. So your job is to make sure the numbers are all there. The AI, so the companies I'm on all have are tech. So AI is everything. It's less than AI. It's all about AI. I guess it's in their name. It's all about AI and the need to move more and more data rapidly. I'm also on the board of mmTron, which is we're just raising the next round if you have any spare money in your pocket. But they are a satellite solution. World records in power amplification and satellites. There's still 3 billion people in the world that don't have internet access.

Dave Vellante

>> Have access, right?

Diane Bryant

>> Satellite market is on craze. Had one investor tell me, there's only two places I'm investing, AI and Satcom.

>> Yeah. Great stuff.

Diane Bryant

>> So, mmTrons good.

>> Yeah. And also you don't have the pressure too. The quarterly, as a manager, you have that shop clock. If you're at a public company, you're on the board a little bit less stress.

Diane Bryant

>> Well, yes. I mean depending on the size of the startup, I mean, mmTron is very small, and so I'm the only board member, so I am the board. But Celestial AI, sure, we have quarterly meeting and watch the financials. You need to do that still. You want to prepare yourself for one day going IPO. You need to have strong financial governance.

>> Got it.

Diane Bryant

>> But yeah, it's just so different.

>> It's great to have you here.

Dave Vellante

>> I have one more question. I got to go back to Intel. How important is it public policy for the US to have onshore advanced manufacturing?

Diane Bryant

>> Incredibly important. I mean, yeah, I'd be an idiot not to think incredibly important.

Dave Vellante

>> That's the easy question. How important is it that has to be a US domicile company? And the reason I ask that is because you have geopolitical threats.

Diane Bryant

>> Yeah, it is important. And it's not something that can't be done. So it just should be done. I'm going to say the 10 year again, it's going to be 10 years from the time you break ground on a fab to when you're actually running the first wafers to when you're in high-volume manufacturing. And the world is actually, you can see the capacity impact is 10 years. So you got to get going, you got to get started, but then you need the realism behind it. And I think a lot of the government chip fact, I think the government doesn't digest just how... If it wasn't complex, you would have it today, right? You've got the leading edge is TSMC and Samsung. So of course it's not easy, but you got to get going.

>> You got to stay in the path.

Diane Bryant

>> You got to stay committed, Andy Grove.

Dave Vellante

>> But how do you feel about getting the prospects for having a US domicile company? To me, that can do advanced manufacturing are quite dim right now, Lip-bu Tan I think has at least a viable strategy. But in my opinion, I wonder if you could agree or disagree, not a viable strategy to be a number two in advanced manufacturing. I think there'll be somewhere in between global foundries and Samsung, which has its own set of problems, which are similar but different.

Diane Bryant

>> Yes. So number one, I am so excited that Lip-bu is leading Intel. I mean, that was just a huge sigh earlier. I was there 32 years. I mean, I was there when Andy Grove was the CEO. Gordon Moore sat down the aisle. I love that company so much. And to watch it degrade over time was very painful, like losing a child. And so the announcement that Lip-bu and his willingness to walk in and make it happen, God bless him. Intel's tried five times to build a foundry in the 32 years that I worked at Intel. Five times, he tried. It's leadership and it's making it a priority. Anything can be done. Come on, look at the invention that happens every day. Intel can have a very successful foundry. How long does it take to catch up and be leading edge? It's hard. Under the prior CEOs, Intel lost two nodes, two nodes behind. And if you could leapfrog and all of a sudden jump two nodes, well, then Moore's law would've been, you can double performance every one year which is you can't do it. You can't do it. So it'll take a while. But I will also say, think of how many chips there are in the world, they don't all need leading edge, right? There's a huge demand.

Dave Vellante

>> You can do six nanometers on an IoT device.

>> I agree with that. But we come back to my question, how important is it for a US domicile-

Diane Bryant

>> Oh, it is very important.-

Dave Vellante

>> to be at the leading cutting edge?

Diane Bryant

>> Yes, it's very important, but if you don't start, you'll never get there.

Dave Vellante

>> But my contention, Diane, is the only way to start is if you shed foundry and structure it in a way that brings in virtually a hundred billion dollars in 10 to 12 years. That's my premise.

Diane Bryant

>> So I don't agree that you have to break foundry apart from Intel. Look at Hock. He runs a software business. He runs a silicon business, two very different things. And he does it magically. There's no reason that Intel can't run a foundry business and can't run a product business and do it well. I just don't think that you need to do that.

Dave Vellante

>> Money and time would be my-

Diane Bryant

>> It's time and talent.

>> And leadership.

Diane Bryant

>> And leadership, absolutely. Focus, prioritization, talent. Intel isn't foundry. We grew up as a custom. We're an IDM. You integrated your manufacturing and design so tightly now you need to break it apart and have the talent over in the foundry side. You just need to do it as leadership.

>> While we got you here, I want to ask about Andy Grove since you brought him up. We quote him all the time on theCUBE. One of the books that I read a lot when I graduated business school years ago was High Output Management, classic book, how to do meetings. It was a classic book and it's a must read if he was updating that book today or AI with work from home with the leadership principles of Andy Grove. I mean, let chaos reign, reign in the chaos. We love these quotes. What would that book look like today? What's high output management today in today's culture as the speed of the game is fast leadership, first principle, you start to see cultural things?

Diane Bryant

>> I think Andy would sit here and say, "I'm not changing a thing," if you know Andy. I think Andy would say because you're right, Andy, it was back to basics, back to basics, back to basics. And so today things are moving in even faster. So if you don't have that operational excellence, if you don't have that focus, that prioritization and just disagree and commit, move, right? Andy's whole thing was just move. The world's moving so fast, you can't afford to sit around and debate.

>> Argue, complain.

Diane Bryant

>> Argue and complain. So I think Andy would say it's nothing.

>> It's still relevant today.

Diane Bryant

>> He'd be a little mad about it.

>> No one uses emails called Slack.

Diane Bryant

>> Well, he'd be a little mad about Moore's law of dying for Intel, not for the industry.

Dave Vellante

>> But your point is like-

>> Jensen doesn't even do one-on-ones. Andy would have one-on-ones. He'd had a management-

Diane Bryant

>> Who doesn't do one-on-ones?

>> Jensen said he doesn't do one-on-ones. Jensen Huang, he said on Dave-

Dave Vellante

>> His performance reviews are in front of everybody. All his direct reports.

>> Good luck with that.

Dave Vellante

>> It's efficient.

>> What's real.

Diane Bryant

>> But I will say Andy would walk in and say there's too many goddamn meetings. Right? Too many one-on-ones. So I think Jensen, it's actually don't sit around in a conference room.

>> Only the paranoid survive. So he would probably be very paranoid of not missing a node.

Diane Bryant

>> Oh, we would've never, I mean, even if were still alive, we whatever. And so BK dealt the first deadly blow Intel by killing Moore's law that he gets to take that to his tomb. And then Pat just put Intel on life support and now Lit-bu's going to save the day.

Dave Vellante

>> Yeah. I think his plan is imminently viable. Pat's was not, and I'm sorry we wrote about it for-

>> Diane, it's great to have you on and really appreciate you sharing your thoughts and commentary. Congratulations. I love the work you're doing. And again, thanks for coming on theCUBE. Appreciate it.

Diane Bryant

>> Yeah, it was my pleasure. Thank you.

>> The leaders are all coming on theCUBE. I'm John Furrier, Dave Vellante of theCUBE plus the NYSC is part of the Wired community that we're building open source community. It's open to everyone. Come join us. Thanks for watching.