NVIDIA Leadership, Kubernetes and AI Agents at KubeCon
NVIDIA leadership, Kubernetes scaling and the AI agent build-or-buy choice lead this Techstrong Gang discussion from KubeCon. Alan Shimel and Mike Vizard join Guy Currier to examine three changes facing enterprise technology teams. The conversation connects the people shaping AI, the infrastructure supporting it and the tools organizations choose to operate it. Each segment separates broad industry expectations from the practical decisions facing IT leaders.
NVIDIA Leadership and the AI Industry
The panel opens with NVIDIA leadership and its implications for the AI industry. They consider how much a technology movement depends on a prominent leader. They also examine whether customer demand can sustain momentum through a leadership transition. However, the discussion extends beyond any single executive. Enterprise buyers still need useful products, dependable infrastructure and a clear reason to invest. The panel weighs those lasting needs against the uncertainty that follows a major change. Their observations describe possible industry consequences rather than a forecast of market performance.
Kubernetes Scaling for Agentic AI
Next, the discussion turns to Kubernetes readiness for agentic AI workloads. Moving from a demonstration to production introduces coordination and scaling demands. Teams must consider how agents use infrastructure and how their workloads interact. Consequently, operational planning matters as much as the promise of the application itself. The panel explores the work needed to support these systems reliably. It also considers what cloud-native teams must learn as AI becomes another production responsibility. The discussion treats scaling as an engineering challenge, not a feature that arrives automatically with an AI deployment.
Build or Buy IT Operations Agents
Finally, the panel examines whether IT teams should build or buy AI agents. Custom development can offer control, but it also brings integration and maintenance responsibilities. Vendor tools offer another route, although buyers must still assess fit and ongoing support. Therefore, the choice depends on the work an organization wants to own. The discussion weighs those tradeoffs against the resources available to the team. It also asks how much operational responsibility remains after an agent enters production.
Recorded at KubeCon on November 12, 2026, this conversation follows three connected enterprise decisions. Technology leaders must assess industry change, prepare their infrastructure and choose where to build internal capability.
Transcript
Hey everyone, it's Thursday. We're at KubeCon, and you're watching Techstrong Gang. It's been a hell of a week, gentlemen.
A hell of a week. Has. Started early for me.
I came out here Sunday morning. I think I was up at 3:30 in the morning Sunday, grabbing a plane with the video crew, getting out here, setting up. We had an amazing Futurum summit Monday with some great discussion and coffee and breakfast and just networking.
Tuesday, kicked off this event, and it's been a blur since. I don't think, other than a short lunch break, we've been booked- Yeah ... the entire three days.
Personally, though, I thought nothing could overshadow this event. And yet, sure enough Yeah. Here we are.
There we go. Here we are. I didn't want to go there, but he did.
Let me first introduce you to our gang today. To my immediate right, the man in the middle, we've got Guy in the middle. Guy Currier.
Guy, welcome. It's good to be here. Good to have you back with us.
Especially on this day of days. Second day, right? Yeah, second day in a row.
Good. I want to do all days here- Well ... on the Gang with you, but I had to attend the keynotes at least one day.
Yeah, no, the keynotes were really good. Mm-hmm. Very good stuff.
Yeah. And then, of course, on the far right, needs no introduction, Mike Vizard. Mike.
Where's my corporate logo? Next to a group logo. I see you're in official gear.
I don't know. I feel a little overdressed, but sometimes it's better to be overdressed than underdressed, or undressed. Speaking of undressed- Mike, what have we got today?
Well, well, well, well- Any news? Any news today? Well, thanks for that dubious lead-in.
Did you have trouble coming up with any of the topics for today? Go ahead. All right.
Well, so Jensen at NVIDIA has resigned. Jensen Huang has resigned. Yeah.
Effective immediately. Mm-hmm. And it comes the question of...
I'm just going to toss this right back to you. Is that going to put an end to this thing, or are there more shoes to drop? No, unfortunately, the sharks smell blood in the water.
Mm-hmm. Politicians, police agencies, the- Reporters ... paparazzi and, yes, and media of the world are going to...
This is a froth they're feeding on. But I look at this, there's two angles to this story. Guy and I were talking here on our green room couch.
Don't get any ideas here. I'm not Jensen. But we were talk-- There's really two angles to this story.
One angle is, what effect is this going to have on the AI market, on the market in business in general? Right? And for many of us, Jensen was sort of the godfather of AI, right?
Getting out there. He was always on a stage every week it seemed. Mm-hmm.
His leather jacket, evangelizing what AI is, what AI could mean, where AI was going. He was the living, breathing face of AI for most people, I would think, more than Sam Altman or Dario or Musk or Zuck or any of them. They're all billionaire bros in their own right.
But it was Jensen that was the living, breathing AI incarnate. And so I want to kick our discussion off, which is looking at that aspect of it, which is, what do you think the effect is on the AI market in general, Guy? Nothing.
You think it's a big nothing burger? No. 5 in November of 2022.
Why? Because two reasons. One was this was just the masses, so to speak.
It was the spark. Starting with the chattering classes, experiencing natural language discussion with a computer, from their perspective. And number two, the utility of it, even though it was error prone and full of hallucinations at the time and all that other sort of stuff, the utility of it, the usefulness of it was obvious and remains obvious to this day.
That source of demand is here to stay. It's what we do with it that's really important. But no, I don't think this is going to even be a blip as far as that goes.
" It's the circumstances and the times of the moment that drive a trend or an issue. So NVIDIA would've been NVIDIA one way or another. It's nice to have Jensen, he was a great speaker, great for these conferences.
But I'm not a believer that anybody bought NVIDIA GPUs because Jensen was the CEO of NVIDIA. Not because he was the CEO of NVIDIA, but he was the Pied Piper of AI. And I think because- Well, that phase has passed.
I get it, yes. And perhaps it is. And not because of we're into agentics, and we'll soon be into robotics and all of these things.
But you look at any great technology wave, and there's usually a face, a person who was the... T. Barnum because there's not a sucker born every minute kind of thing aspect of this, but there was an evangelist, there was the guider, light of it.
That being said, I think for the market in general, Guy, you're right. No one's going to crawl back under the rocks they came out from, and then we're going to continue to consume AI. There is going to be seismic waves compression through here.
If you're CoreWeave, and the reason you can keep the lights on is because AI has basically backstopped you. If you're- Well, wait. So AI hasn't backstopped CoreWeave, right?
No, NVIDIA has. Yeah. So I'm really glad you brought up CoreWeave.
A really significant element of the news today is who's stepping in as interim CEO. The CFO. The CFO.
I think Giselle Kress. I know Kress is her last name. Yeah.
But the question then becomes they're going to get a new CEO, and so she's interim. No, but, well, so they have earnings in five days. Mm-hmm.
Okay? So that could be an obvious reason. But I think the less obvious reason is NVIDIA had its own method for maintaining dominance and growing the market, just like Intel had its method back then.
Mm-hmm. There's some similarities. Intel's was to pour marketing dollars into- With Intel Inside ...
the PC and server makers. Yep. Right, with Intel Inside.
NVIDIA's method was, roughly speaking, financing. Backstopping. Right?
So that's why it's interesting you mentioned CoreWeave, because there's this sort of circular kind of... I don't think it's overstating it to call it sort of a circular financing method where there's funding, there's stock investment by NVIDIA, and there's a return guarantee of purchasing NVIDIA. And even beyond that, NVIDIA is promising or setting CoreWeave's consumption, like guaranteeing consumption levels of CoreWeave services.
And it's not just CoreWeave, it's Nebius, it's a lot of other suppliers. It's the entire industry. Right.
That's what I'm trying to get to. So this is shock waves. Right.
They're like a bank of AI in a certain way. So- This is a page out of GE. GE opened a bank.
They operated a bank- That's right ... of finance. Today it's called Synchrony.
Mm-hmm. Right? Because they changed it in 2008, but it was the GE Bank.
Well, Dell has a bank, too. We're talking about it differently because they're guaranteeing demand. Right.
NVIDIA's guaranteeing demand of NVIDIA based services- Yep ... at CoreWeave and at other places. So yeah.
So it's circular. But the question in my mind is, who's going to step into that and become the next voice of AI or fulfill that role? Because they are big shoes to fill, and who has the chops to do that?
Well, I think NVIDIA still has the financial wherewithal. The question is, who becomes the embodiment of it? We got a rogue's gallery here.
I mentioned them. None of them, say what you will about Jensen- They've been fumbling. Yeah.
Say what you... Jensen had much better image than any of those guys I just mentioned. I guess that remains to be seen.
Let me say, I don't think it's Lisa Su necessarily. No. And I think that's to her credit She is much more methodical and- She's more of a traditional CEO.
Yeah. But she's also an engineer at heart. Right.
She was trained as an engineer. She remains that. She is- She's not an evangelist.
Yeah. You need a bit of a PT Barnum. And Jensen in his motorcycle leather jackets and all that had that flash, had that pizzazz.
But I want to take this in another direction, because behind the pizzazz, behind the leather coats, behind the billions of dollars is a man, a human person who's 63 years old and married for 40 years with children, raised a family, and has done great things in his life. It isn't Jensen's fault. I'm not blaming anybody but Jensen for this, but we all bear a certain responsibility because we put our heroes on pedestals, and we don't notice that they've got clay feet, because at the end of the day, they're human.
And you know something? " Well, you know who else and what else is non-deterministic? People are non-deterministic.
I don't think the issue is the affair. I think the issue is lying and what happened after the affair. Two affairs.
Yeah. So far. We know of.
I know nothing more, but these things tend to... One with somebody in your direct line of- Line of... Again, I'm not excusing him.
Mm-hmm. But I forgot, mama don't let your sons grow up to be cowboys or something? Something like that.
Right? We tend to hold our heroes to an inhuman standard and forget that they're human. What he did, if we went around this whole show floor, there'd be a lot of people who've done it, and let he without sin throw the first stone.
Someone said that one. That's all true, but there's a lot of blame to go around here, and so there's going to be a lot of questions that go down this path. What did who know, and when did they know it?
And they don't. And that's going to be all inside of Nvidia, plus what did people in the government know? Because from discussions out there, it feels like this was an open secret in a lot of places.
I imagine, only imagine, I have no evidence of this, that this was an open secret in certain echelons within Nvidia, but it is now documented and reported or reported and documented that the White House knew about this a year ago. I'm glad you brought that up. A year ago.
Let me just give you a hypothetical. Let's assume you had a president in office who had five children with three wives, had documented affairs while they were married, right, had paid off porn stars while their wives were pregnant. This is relatively small potatoes.
Yeah. Small potatoes. If we swallowed it for six years with the man in the White House, what'd Jensen do so bad, other than that maybe she was spying for the communist Chinese.
So do you think if the election hadn't gone the way it went for the midterms, we wouldn't be having this conversation? It's a good possibility. That's what I'm saying.
You know what I mean? And you know what? I'm glad it did it because enough of this bro culture.
I say it tongue in cheek, and what I said about what goes on at one end in the White House and the person in there presently is true. I'm from New York. I lived it.
I saw the page six pictures every day. But enough of that s**t. It's time.
I spoke to a few women here at the show about this Jensen thing, and they said they're surprised it came up because they feel like we've returned to 1992, that all of the progress we've made in the last 30 years, breaking the glass ceiling, respect for women and all this, has taken six steps backwards in this bro culture that we have enabled. And maybe as much as this is a nothing burger for AI, maybe this is the beginning of a take back of, hey, morals like this are not acceptable. This behavior is not acceptable.
We're not going to stand for it no matter who you are. Well, let me just say that I don't think 2016, if we want to pick a point when the reversion began, I don't think 2016 was all that great, particularly for women or for- No ... sexism or for bro culture or for any of those things either.
We already had a long way to go. Yes. And if there is some kind of rebound, and let me be really practical here.
I don't think we should just look at that as a social or cultural issue. It's about creativity, innovation, interaction, cooperation across more than just corporate lines and technological lines. It's with society, and it's with everything else.
That's what this is all about. That's what it's all for. Mm-hmm.
So if a story like this gets a few people to snap their heads back and pay attention, that's a good thing. Yeah. I hope it is.
So one last question about- Good thing for technology, for technologists. For society. Well, yeah, but this is a technology show.
One last thing. What impact does this have on open weight models and our relationship with China, and are we going to be suspicious that there was undue influence here? Because he was a big proponent of open models, especially from China.
You answer. I want an answer. Okay.
You go first. So- Usually you say what I was going to say ... I'm going to give you my Guy Currier impersonation.
Nothing. See? He said what I was going to say.
Right? Which is, did this woman or the company she was working for, Windspeed or whatever it's called, send chips to China that might have helped them develop some of their models? Maybe.
In the long run, that's a blip. Even in the medium run, if you ask me. But go ahead.
Yeah. And this whole show we're at, and you look how big this floor is and all the people and everything else, is a testament to the power of open source, that open wins eventually. And it's going to do so here in AI.
And again, in my book, "The Indispensability Trap," I mention that as well. At the end of the day, there'll be use cases where you want that frontier model, that closed source, black box, if you will. But at the end of the day, open wins.
Mm-hmm. And if that's the case, the whole market and a significant part of our economy has been boosted by these investments in AI, and it's Nvidia, Anthropic, and OpenAI. " And it might be- And that, my friend, is the indispensability trap.
Ah, okay. Eight trillion dollars. Uh-huh.
I think that was going to happen anyway. Yeah. Yes.
I would agree. Maybe the shape of it is a little different, but I don't even think the timeline is all that different, and you're only scratching the surface. So you're implying, maybe you're not intending to, but you're implying the benefit of let's assume that this theft took place, or so to speak.
It certainly is illegal, no matter what route you take. It was illegal for Nvidia chips of that caliber to be sold or to be sent off- Smuggled to China ... sent off to China.
Okay. But let's assume that happened, and that was critical and crucial to the development of AI in China. What did we get out of it?
What did the world get out of it? Open source models. Yeah.
Open weight. Open weight models, sorry. DeepMind.
Okay? That was- Well, not just DeepMind, but there's so many other- So cut-rate imitations of foundation models, let's just call it that for the moment. That is a benefit to AI, to the technology, and to users, and to the world, okay?
Mm-hmm. And I think Anthropic's going to be fine. " And we suffered.
" That's their right, but they're going to be fine. " They're going to be fine because they are smart people and smart managers hiring other smart people to do smart things in probably the most promising technological development we will experience in our lifetime. So I- Never underestimate the imagination of lawyers to come up with reasons to sue.
If I'm a- Totally legit. If I'm a shareholder- Totally legit. I'm not denying ...
if I am a shareholder of Anthropic, I can make a case that this harmed me And now you will see a class action suit coming from somebody- Yeah ... aimed at NVIDIA for harm to the stock prices of Anthropic. That's what you have E&O insurance for .
But what I'm referring to, that's a sideshow to the overall development of AI, and we are at the baby stages of that, and merely substituting foundation models with open weight models or whatever is just the beginning of model choice. Dude, this is a circus with 20 shows on at the same time. This is entertainment .
Yeah. The greatest show on Earth. Maybe it is about the greatest show- Exactly.
If you want to- But it's looking for a new Barnum ... sometimes the sideshow is the best show. Sometimes.
At Cirque du Soleil anyway. All right. We got to hop to our next one, though.
But look, this is quite a story. I'm sure we'll be discussing it in the future here on the gang and on all our Techstrong properties. All right.
So we are here at KubeCon, so let's talk about Kubernetes because, well, that's what this thing is really all about. Okay. And the conversation that people are having is like, will Kubernetes scale to handle these agentic AI workloads, and is it ready?
And it feels like a significant amount of re-engineering needs to be done, or at least that's one approach that people are talking about. And it comes down to one we talked about the other day, which is I can't lock up pods with AI agents, so I need a more efficient way of scaling. And then secondarily, now, this also came out yesterday, is that it looks like the control plane itself isn't going to scale for agentic AI workloads, and we need to re-engineer that as well.
These are efforts that are going to require a significant amount of time, and the concern is that people are going to be looking to deploy AI and agentic workloads. They're going to look for another platform because Kubernetes is not quite ready. Good luck.
Would you like to go first? Sure. I posted a video on LinkedIn on this yesterday where my basic take was these are all problems we already had.
The original source of the problems was web applications, file-based web applications with highly variable usage, right? The same problems now are in the proverbial saying taken to 11. The dynamism is greater.
The scale level is greater. The security vulnerabilities are greater. We're talking about scaling.
Instead of scaling out to 100, we're scaling to 100,000, thanks to agentic action. Yeah. And there will be two solutions that are going to happen at the same time.
The cloud native community is going to come up with the ultimate solution, and they're going to take their human time doing it as fast as they can. And then the other one is going to be all the proprietary and application specific, training specific, all of those solutions in the meantime. There's a second way of thinking about this, and we just had Bailey from Cosmonic on talking about it.
We could find a way to run these agents more efficiently on the infrastructure we have. So instead of using containers, maybe we start using WebAssembly, and this becomes the- So says the WebAssembly vendor. Yes.
Well, it's an argument, though, that says, hey- So WebAssembly is more, let's say a security in the old-fashioned sense of the word, like containment scenario within the container. It's not the container. I mean- It doesn't replace containers ...
it doesn't replace containers, but it is an alternative form of the sandbox, which is a big buzzy thing right now, right? Well, it all functions in a different way. Well, there's a lot of different definitions of sandboxes out here today, I'll tell you.
Fair. Completely fair. Yeah.
I'm just saying like- No, but it is ... the problem statement that sandboxes are addressing, WebAssembly can address. Absolutely.
What do they say about democracy? It's messy. It's messy.
And what else do they say? It's the best form of government we have. Yeah.
It's the best of the worst. Well, yeah. Sure.
Yes. It's not the most efficient. Certainly authoritarianism, totalitarianism is more efficient, but democracy- Long as it lasts, maybe.
What? Long as it lasts, maybe. Yeah.
But continue. But it's the best of the worst. Yeah.
And why do I always tell you that gets you p****d off? You go to war with the army you have- Yeah ... not the army you want.
You know, the guy who said that lost every war he was in . But it was like three different administrations. But that being said, as we stand here today, it would not be the AI industry that we see had it not been that Kubernetes and containers and the whole stack was there for them to build on.
Does it fit like this? No. But to go out and start something from scratch that you think is going to replace Kubernetes, you're smoking your socks.
I think we are going to, quote unquote, replace Kubernetes, but not the APIs. So whatever is underneath that will still be called Kubernetes. It just won't look like anything that we currently see as Kubernetes.
It'll just be a different platform, but the APIs will be the same. They'll be compatible. Aren't you merging the technology with the community, though?
The community is more important than the technology. Maybe. A decision the community makes to take the technology on a left turn, if it's a community decision- Yeah ...
that's what support, call it what you like. My only argument, and the same one I made yesterday, is the community needs to move faster. And even democracies, when presented with a crisis, will marshal some resources to go deal with it in a particular sphere.
This is your guys' sport. So can I violently disagree with you for probably the 12th time today? I described the two paths, okay?
When there is a need for the investment level by whoever, it could be a startup, it could be a giant enterprise, for the kind of scale, security, et cetera, that we are talking about, where cloud native is currently lacking, that will be proprietary. And they will follow that. And just like NVIDIA just contributed OpenShell to the community, at some point, there may be contributions out of that pathway.
That is the immediate pathway. It's expensive. It won't be for most applications.
And in the meantime, the community will continue as fast as it humanly can to develop these. Why? Is the proprietary not human?
No, the proprietary is human, but the community has its resources, and it's made resources. So these humans can go humanly faster than these humans can go? Well, yeah, that's capitalism for you because these humans can be funded with a boatload of money that's not- But there's only one being funding here.
Take a good look around here. You think these people are hurting for money? No.
But I do think that- I don't think money's the issue. And I think what NVIDIA did- Do you dispute the two routes? No, no.
Yes. I think you're thinking you're a child of your generation, my friend. I think today's model is an NVIDIA, a Google, take your pick of any of these big companies.
They recognize that the fastest route to market is to donate the seed of what they have in mind for an open source project, throw it into the CNCF, and you instantly have an entire community with much greater resources than any one of them are looking to put into it to foster the development and put it in like a giant Petri dish and see all the different ways it goes. I don't think that's fair. We've seen some other- It's worked.
We've seen some other- Isn't this proof? We've seen some other instances of lighter weight Kubernetes things like CIQ built with AI. I'm almost certain that somebody's playing around with this and is using AI to get to the accelerated pace that we need.
The question is, how soon does something like that get donated to the CNCF? That's the proprietary problem you have. It's going to get donated sooner than later because it takes a village to do something on that scale.
Yeah. Even what you're talking about, Mike, is not replacing Kubernetes. It's putting an API, maybe facet on it.
But the underlying functionality of what Kube does, and it's not just Kube. We tend to just think of Kubernetes. It's a stack.
Yeah. It's the mesh, it's the containers, it's the security pieces of it. You're not- I don't care how many resources you have, you're not going to recreate a clean room from scratch, optimized version.
This is going to be faster because it's already down the road. And let me just add one other thing. Okay.
Which is the premise of the question in the discussion, which I participated completely in, is unpredictable, risky, hard to control usage levels. Right? Okay.
However, a lot of that arises from the gross inefficiency in AI right now. And there's a whole other track to managing this, that has not been pursued yet, which is smaller models, less memory usage, less capable infrastructure because it doesn't need to host such- The efficiency. Yeah.
Which, that's a natural- We have a lot more- That's a natural sort of thing. Which I would argue Wasm is an example of, so. I'm with you on that, man.
I'd argue Wasm is too. All right. Hey, we've got several waiting, so I'm going to cut this one short and say let's hop to our third.
All right. Third topic should be relatively short, but there's a debate going on as to whether or not IT teams should build their own AI agents for operations, or should they rely on a service provider or somebody who has that expertise because, well, there's just a lot that can go wrong if you build your own and you don't know what's happening with these things, and they're not vetted to an inch of their life to only do a specific thing. And so are people going to wind up buying AI agents or building AI agents?
So I got to be honest with you. I know this came out of a story you were working on, an interview you did with Mirantis. Mm-hmm.
And when I got a company that's selling forward deployed engineers saying you need forward deployed engineers to build AI agents, I got to call b******t. But there's not only one of them. " That's what their lead is, right?
Who doesn't in this field? All these companies, they all have a service now. If those guys are telling you that you need to buy their FTE guy because you can't do it yourself- That's the debate ...
do you know what I mean? That's when I look out my window and make sure the sun is out or not. It's not a debate, it's a sales pitch.
I think you're being a little harsh, yeah? Yeah, yeah. Because I think the important point is, like many, many things, you can do things yourself, or you can buy off the shelf or whatever range.
Because hold on. Buy off the shelf. Hold on.
Hold on. Listen, just doing it yourself is fun and you can learn, and you can probably come up with, ultimately, if you can release, which is a big question, you can come up with something that's better, more tailored, more tuned. Whether you're using a service provider or a partner or just literally doing it yourself, this is a perennial discussion in IT.
Yep. Mm-hmm. And I would just say that probably what you, to quote someone else or someone we just quoted recently, what you don't know that you don't know about AI is probably pretty significant.
So just have a little humility before you're taking- Absolutely. That's all I would say. There is no dearth of vendors who are selling agents already.
Right. Now, the question then becomes if you do build it yourself, and I'm the IT guy or the software engineer that goes build these AI agents, and I leave the company or I get hit by a bus, do those agents come with me or do they stay with the company? Probably when you got hit by the bus, they're kind of scurrying away, I'm assuming.
Right. But forget AI agents. Do you build any soft-- For most people working here, when they sign a contract with a company, anything they develop, invent, write code, belongs to the company, not to themselves.
Well, I got- That's what I'm saying. You remember the old rhyme sheets that the rappers had, and they were stealing them from each other and all this other kind of stuff? No, I mean technical.
You have to sign. Source code technology. I don't know.
But I got my little prompt sheet, I might take it with me and my little skills, what have you. Maybe you have a prompt sheet to go do your agent. Because no, because I created the agent.
How did the prompt sheet, let's say you had your own little code repo, your own GitHub. But no, but that's my point. Well, yeah, you have your own code, but my point is, in the age of AI, I have my prompt sheet that created this cool thing.
That's why I can go to- Work with me in the end, and then I go to company B and just recreate it. You can just go do it yourself this time. But what's unique about- Faster and easier now.
Yeah. What's unique about AI agents, though, is they do have certain human-like qualities. So now let's say you replace the software engineer who built them with another software engineer.
Is he going to really want to go learn and build a relationship with those AI agents, or is he just going to build a whole new set of AI agents himself because he wants it to be personalized to him or her? What's the economic impact to me? What I just said.
He doesn't really have to if he's rapaciously self-interested in the first place. Those relationships you build are based on what? Prompts, context, harness, et cetera.
But the agents learn that. That's very portable. But the agents not only learn about the environment, they learn about you and what your preferences are and what they like about you.
I think they're short. So does my agent learn about me or about the forward deployed engineer that the company sold me? Or was the design in the first place to ensure that that doesn't happen?
To Guy's point, I think those are both valid avenues, right? You could buy off the shelf. You could develop your own.
You could bring in someone to help you develop your own. All I'm saying is when I have a vendor telling me that this is too hard, you need me to do it, I open my window and make sure if it's rainy or sunny outside because I don't believe them. My point is, many times and many things we watch on television, and at the very end of it, they always say the same thing.
Don't do this at home. Your mileage may vary. Guys, it's been an amazing KubeCon.
Guy, I appreciate you popping in here both days. I know you're busy doing- It's been my pleasure ... meetings and stuff.
Mike, as always, it's great doing these with you here. This is a lot of fun. We've got more people waiting.
Well, we ended 2:00 PM Mountain Time, 4:00 Eastern Time today. So we've got a lot more coming at you. But you're watching Techstrong TV, and you just watched the Techstrong gang.