Tech Barometer – From The Forecast by Nutanix

Tech Barometer – From The Forecast by Nutanix

By Tech Barometer – From The Forecast by NutanixNewsTechnologyBusiness NewsTech News
Download on the App Store

Tech Barometer – From The Forecast by Nutanix episodes

  • Finding Clarity During EUC Industry Disruption

    In this Tech Barometer podcast segment, End User Computing veteran Ruben Spruijt cuts through chaos from recent changes around virtual desktop infrastructure and desktop-as-a-service offerings.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Transcript:

    Ruben Sprujit: It’s like a massive plate of spaghetti. Where do I start? What’s the challenge?

    What’s the idea you have?

    Jason Lopez: Ruben Spruit is the field CTO and VP of product at Dizion, a company that provides desktop as a service and cloud computing solutions to enable remote and hybrid workforces. He helps companies understand how to deploy end-user computing technologies, but not only companies. He’s passionate, hands-on, and ready to help anyone in the EUC community.

    Ruben Sprujit: First, you listen to peers, to customers, to prospects and understand, hey, what is their idea or challenge? And many of us listen for a couple of seconds and then, okay, this is the problem, this is the solution, this is the vendor. But it’s good to step back and listen a little bit more.

    Jason Lopez: Here at The Forecast, we’ve interviewed Ruben numerous times for articles and Tech Barometer podcasts to learn about end-user computing, virtual desktop infrastructure, and desktop as a service. In this interview at .NEXT 2025 in Washington, DC, he talked with The Forecast editor-in-chief, Ken Kaplan, about changes around EUC, such as new vendors and partnerships in the wake of Broadcom VMware’s spinoff of its EUC offering, which became Omnissa. And he talked about AI and methodologies for helping IT teams build and evolve their EUC strategy, and offers tips for building a career around the EUC ecosystem. But he starts with the question, why are people using desktop as a service?

    [Related: How End-User Computing Leaders Make IT Work for People]

    Ruben Sprujit: It could be students with their own Chromebook want to access SPSS, nuclear power plant designer and builder using high-secure, sensitive data with design apps to design a nuclear power plant. It could be a retailer or using a handful of Windows apps in their retail office. It could be government agencies using office productivity. So it’s different reasons to use virtual apps and desktops. But it all starts with, OK, I want to have a simple access to apps and desktops. I want to be workplace independent, need to be secure. And simple is the key. How can I keep it simple so the operational complexity is sort of gone? And that’s important these days because knowledge and people with skills is challenging nowadays. Keep them is challenging. Find them is challenging as well. So building a platform that’s easy to run as an admin and easy to consume as a user, like both we are also end users, consumers of solutions, these are the main ingredients of a good platform.

    Ken Kaplan: I met you seven years ago at Nutanix. It was explained as it’s streaming your desktop experience from a server, from the data center, from the cloud. That idea, just how we would watch a Netflix video, it’s streamed to you. So your apps aren’t necessarily on your device. And there are lots of reasons for doing that. And it seems to be almost the status quo now. So what has changed in those seven years and what are you excited about today?

    [Related: From Skunkworks to Leading Hypervisor Alternative]

    Ruben Sprujit: A lot of things changed. Like seven years is a long time. So if we go back maybe seven years, Nutanix acquired our company, Frame. We saw Corona and a surge of workplace independent work. That’s one of the reasons why we grow like crazy also within Nutanix. So that changed. That’s new normal nowadays, right? Hybrid cloud is new normal. So supporting on-prem, supporting NT2, supporting public cloud natively. Unified communications is normal as well, supporting that. High-end graphic apps, all kinds of apps supporting that. But still with a clear strategy, keep it simple. Like how can we make sure that admin experience and user experience is as simple as it can? Everyone is using today a browser. 70% of the work we do as end users is browser. And then the other 25% often is like Windows apps. So how can we deliver both web and Windows apps in a seamless way to users? What is happening as well, we spin out of Nutanix, a little bit more than two years ago. So Nutanix can focus on infrastructure, hybrid cloud, and we can focus on end-user computing. So Dijon acquired Frame technology. Look at Frame as the car engine, while Dijon is more like a sort of managed services company. Coincidentally, both companies, so Dijon and Frame started roughly in the same time, 2012. Dijon with focus on Horizon, nowadays called Onisa, build their own control plane to orchestrate and automate Horizon deployments. In the same time frame was built as, okay, how can we build in cloud native solutions to stream apps and desktops to end users? So fast forward today, 2025, Dijon is using the engine, the Frame engine, to stream desktops and apps, Windows and Linux, personal and pooled, that’s technology, but also add a managed service on top of that to help customers when needed, not required, but when needed to add admin services to local admins with service we offer. It’s a services company with a very strong engine called Frame.

    Ken Kaplan: The ecosystem, you touched on it, can you talk a little bit more about the constellation that you’re seeing out there, companies and changes that are happening?

    Ruben Sprujit: It’s a very exciting space, end-user computing. So end-user computing, I always explain as a massive umbrella, and end-user computing, in essence, is the solutions we, you, listener, are using to get work done. That could be a virtual app, virtual desktop, that could be unified communication, that could be an AI agent, it could be a web or SaaS app, it could be a mobile device, it could be cloud storage, all the things I just explained equals end-user computing. And virtual desktops and applications, my expertise, or one of the expertise I have, is a fundamental element in end-user computing, but end-user computing, as I mentioned earlier, is much broader than that. So if you look at the bigger picture of what is happening in end-user computing, AI is part of that, AI agents, SaaS tools, old tools but then sort of AI-infused is happening, unified communications, but still Windows apps, SaaS, software as a service, desktop as a service, application streaming, all these items combined is what is real today. People think, well, Windows apps are dead. That’s not the case. They are kicking alive. Yes, it’s slowly degrading from 100% usage, 15 years ago, 20 years ago, to maybe 70% or 60% nowadays. Yes, SaaS and web did pick up the other like the bucket, but it’s a combination of SaaS, Windows, mobile applications which gets work done for us. So that’s happening. The whole industry is evolving to as a service. If I zoom in a little bit into like the VDI, the virtual desktop infrastructure, and DAS, the desktop as a service space, that’s evolving as well with private equity taking over like the Citrix part, same for Omnissa, spin out of Broadcom, VMware, refocusing, finding new alliances, including with Nutanix. That’s good for the industry, gives more choice. If you look at from a vision perspective, all these developments are helping us big time because a lot of customers, I would say 60% of the conversation I had in the show for past days, are customers, partners, scratching their head, price increase, 2x, 3x. Some vendors are refocusing on like only top segment of customers, and they are not in that segment of their customers. So they are scratching their head, hey, what is next for us? And next could be a physical PC. Next could be alternative for Citrix or Omnissa. That’s happening big time. It’s a very interesting space right now. Combination of like hybrid cloud, on-prem, public cloud, workplace independent, building a secure workplace, security is super critical for many. So if there’s no data on the endpoint, which is the case with virtual apps and desktops, that makes life much easier, not easy, but easier to build a secure workplace. That’s happening in our space.

    [Related: How AI is Shaping the Future of Data Storage Strategies]

    Ken Kaplan: There’s not even a perfect storm, but it’s certainly a storm of changes out there, from the vendor changes, the price hikes, to just the innovation that’s coming across, the options that are out there. You add to that AI and people are thinking differently. How do you characterize this time in your career and what you’re seeing? It seems like there’s a lot of partnerships going on, a lot of working together. There’s some urgency.

    Ruben Sprujit: Yeah. On one side, it’s super energizing, but it’s also complex. It’s like a massive plate of spaghetti. Where do I start? And to help customer with that conversation, and there’s nothing to do with vision or product. My background is consulting, and I always try to explain the bigger picture and understand the why. What’s the challenge? What’s the idea you have and you want to sort of solve? Okay, then what kind of concepts fit with that, and then which vendors can help to fill in that concept to help with the idea or challenge. So what we did last year, we built an EUC Hexagrid. The EUC Hexagrid is a physical representation of our end-user computing space. It has main pillars, six of them, and 22 sort of sub-pillars. And the goal is you start, okay, what is EUC? What is end-user computing? What’s the purpose of these six main pillars? What’s the purpose of the 22 sub-pillars? First, you listen to peers, to customers, to prospects and understand, hey, what is their idea or challenge? And then you use the Hexagrid to plot maybe two of the main pillars and three of the sub-pillars to what their idea or challenge is. Before you talk about how the vendor landscape looks like, and many of us listen for a couple of seconds and then, okay, this is the problem, this is the solution, this is the vendor. But it’s good to step back and listen a little bit more and then plot six pillars, 22 sub-pillars, and then talk about the vendors, what they can do. That’s what the EUC Hexagrid is all about. So it’s an infographic, a graphical representation, 220 vendors are in that picture. It took us three months to build, and there’s a white paper as well. So EUC Hexagrid, if you go to edition.com, you will find the resources EUC Hexagrid. Just fill in the email address, you’ll get the Hexagrid white paper. And that helps in understanding what’s happening in this space from a higher or medium level and paints a picture what vendors are in this space and what are the relevant vendors in this in this space.

    [Related: Performance Engineering in the Age of AI]

    Ken Kaplan: So how do you describe where we are today versus when we met seven years ago, for people who might be interested in focusing their career in this area of IT?

    Ruben Sprujit: Opportunity, I would say. Like, look at AI. But people are also scratching their head, hey, what does it mean for me? What does it mean for my business? What does it mean for products? I’m responsible for products. How can we leverage AI to make life simpler as an admin? How can we solve issues before the end user sees them? Would be great if that is possible, and it is possible, that’s something we’re building. So that’s one angle to understand, hey, what the possibilities are. The beauty of cloud is that you don’t need to spend a crazy amount of money to get started. And that applies to, well, Gemini, JetDVT, that whole theme of AI, that’s where I would spend, if not spend already, time in understanding what the value is in your personal life, but also like with agents, communicating with agents, what it might look like in two, three, five years from now. No one knows. It’s tough to predict. Also with this whole, like, global thing that’s happening, tough to predict, but nevertheless, don’t be shy, don’t be sort of confused by that. It’s not something you can control, and if it’s not something you can control, just set it aside, focus on the things you can control. That’s what I would do. And one of the things is understanding what’s new. AI is newer, but still like the as-a-service cloud, hybrid cloud, I see customers moving from public cloud back to on-prem for different reasons, understand why, understand what kind of tools and vendors are key in that space. It’s really exciting, exciting time.

    [Related:

    FinOps Flicks on AI to Optimize IT Resources]

    Ken Kaplan: You’re also so well known in the community because you’re very involved. You’re on webinars. I remember you’d be at your desk and talking with your hands in the air, or you had your home desk set up just perfectly for these kinds of interactions that you have with people. Where can people go to join the community or learn more about the community? How do they get in to follow what you do?

    Ruben Sprujit: Well, the beauty is there are many different communities. So I would say my community is all around end-user computing, about GPUs, about remoting protocols, about VDI and that. So that’s my community. One of the communities is World of EUC. So World of EUC is a massive community. There is a European community called E2EVC, not connected to any vendor. It’s crazy guys like myself organizing it. And there are also online communities as well, like on BlueSky, on X it’s relatively easy to find the right folks and connect with them. So if you’re new or newer to this space and want to learn, my experience in the past 20 years in that community is that I would say everyone, 99% is here to help. So if you’re new and maybe new to writing or new to presenting or new to technology, just reach out. I’m still Ruben with two feet on the ground. I’m smiling, right? You can see.

    Ken Kaplan: Even just to get you to sit down and sit with me, I waited in line because people just want to come and talk to you.

    Jason Lopez: Ken Kaplan is editor-in-chief of The Forecast. He interviewed Ruben Spruid, field CTO and VP of product at Dizzion, a company that provides desktop as a service and cloud computing solutions. Learn more about end-user computing at Nutanix.com. Go to Nutanix.com slash solutions slash end-user-computing. I’m Jason Lopez. This is Tech Barometer produced by The Forecast. You can find many other technology stories at theforecastbynutanix.com.

    15 min
  • FinOps and AI Optimizes IT Resources

    In this Tech Barometer podcast segment, Mayank Gupta, director of product marketing at Nutanix, explains how intelligent FinOps tools automatically detect optimizations and take corrective action to manage cloud costs, track carbon footprints, and meet regulatory requirements in the AI era.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Transcript:

    Mayank Gupta: How can we make FinOps itself a lot more intelligent, now that we have Gen AI, can intelligence be built into the system where it automatically understands usage patterns and takes action to bring it down?

    Jason Lopez: Generative AI is transforming how organizations manage cloud and IT spending. Although it brings new costs and it also affords new opportunities for automation. This is the Tech Barometer podcast, I’m Jason Lopez. On today’s edition: FinOps, bringing together finance, engineering, and business teams to manage cloud computing costs. This is increasingly important for optimizing IT resources in the age of AI.

    Mayank Gupta: A lot of companies are getting into a hybrid multicloud environment as they’re using GenAI.

    Jason Lopez: Mayank Gupta, Director of Product Marketing at Nutanix, tells The Forecast editor-in-chief Ken Kaplan how FinOps is evolving with better observability and automated cost controls.

    Mayank Gupta: They are realizing that they have to actively track these dollars they cannot just use and run this forever without having some sense of accountability.

    [Related: AI is Hungry for Data, But Can IT Infrastructures Keep Up?]

    You used to just track your CPU storage. Now you’re tracking one more resource, which is your GPU. Similarly, when it comes to training and inference, whether you’re running on prem or whether you’re running on the public cloud, a lot of these FinOps tools have to have insights into your training cost and inferencing cost. Training models can be a very expensive proposition. So we are building that in the tool, and we’re building these new features into our cost governance models, where we are seeing something which has come into the industry, going to be there for the very long haul. Let’s bring in those models and have that. The other very interesting side of this business is, how can we make FinOps itself a lot more intelligent now that we have Gen AI? Can intelligence be built into the system where the system automatically, through agents or through intuition, through learning, understands, hey, this is the usage pattern. And if it’s deviating from them, let me take some action automatically and bring it down. How can we make the job of an IT admin or a FinOps admin easier, getting all these signals from different places? And now we have AI to help. We can distill these signals in something which is easily digestible. Even on the actions that need to be taken off, the user can say, hey, take these actions based on a set of best practices, and now Gen AI can help with them. So we are looking at, how can GenAI make it more effective?

    [Related: Performance Engineering in the Age of AI]

    Ken Kaplan: Let’s talk about maybe the biggest challenges of your customers, the biggest challenges that come to mind and you hear all the time.

    Mayank Gupta: Certainly I think what’s happening a lot is customers have their on-prem infrastructure, which is their data centers. They might have stuff on the edge. They definitely have stuff on the public cloud. How do I have visibility across public cloud, edge core data center visibility is the first step in you controlling your cost. The second big step is having an accountability and doing the due diligence of let’s put this set of dollars to this account. Let’s set this dollars to this account. Who’s using what and where? That’s the second important step. Once you have the visibility, once you have the accountability, how do you make sure you’re optimizing those dollars? How do you set constraints and limits on those dollars? There has to be a sense of checks and balances that, and this is the industry term for this is a showback. We’ll show back to you how many dollars that you’d use. Right? You might not pay me for that, but at least these are the resources that you’re consuming. And then we optimize this, right? If you are wasting dollars, let’s bring those resources down. If something is not needed, let’s turn it down. That’s the third very important step where we make sure that nothing is being wasted.

    [Related: How Ivy League Dartmouth College Moved to a Future-Ready IT Platform]

    Ken Kaplan: You do a lot of webinars, and you talk with professionals and also customers. (3:12) Are there some stories that come to mind about how they’ve applied FinOps?

    Mayank Gupta: I was talking to a Philadelphia utility and the cloud bill just from one vendor had 160,000 line items of different services each costing something else, and they were getting overwhelmed because they could not A understand, and then B, how do I optimize this? So having a solution, having a product, having a tool that can ingest all this information, tag it appropriately, and then automatically give you suggestions on, Hey, you know what? These are the resources that can be reduced or eliminated or optimized. That is something which everybody needs, and I think that’s just one use case. And across the board, every single time they started using our product and they realized that it was a lot easier when we tagged things under services and we said, Hey, this is your storage. This is your load balances. These are your VMs. These are your containers in a graph format, and this is where you can optimize.

    And they were able to click and right-size things. So that’s one example. The other case where I’ve talked to customers, 80%, in fact, I think it’s 83% of Nutanix customers have not just their on-prem system, but they also have multiple hyperscalers. So these are multiple public clouds because all of these vendors have different clouds. The ability to see all of this in one single place is a great benefit for these vendors, and that’s what finops tools provide them. You have the ability to see every single thing in one single place. That’s a fun conversation for me to have. If you had this visibility, what would you do? And they say, oh, you know what? That would make me so much more empowered. That’s what I see every single time.

    [Related: Search for VMware Alternatives That Meet Existing and Future Needs]

    Ken Kaplan: Do you see FinOps, does it help with other parts of budget control, like sustainability issues, energy consumption? And do you see FinOps maybe evolving to other places such as that?

    Mayank Gupta: Yeah, a lot of our customers are seeing that, hey, we are a little mindful of their dollars. People are being mindful of their carbon footprint. And this is all coming together. So you bring a great point, Ken, that yeah, people are bringing it together and people want to see, hey, this is another evolution in me seeing off my costs, my carbon footprint, and all of those things. Yeah, absolutely.

    Ken Kaplan: Yeah, it feels like maybe it starts to move into this regulatory area where you have to report out certain performance of your business and maybe FinOps helps contribute to those pure reporting that needs to be done.

    Mayank Gupta: Yeah, no, absolutely. (5:50) There are regulations with some time-bound periods where these reportings are going to be as a part of your regulatory planning. (5:57) Just as you file your quarterly Q1 reports and 10K reports, this will come in there, right. (6:01) So a lot of our customers are getting ahead of the curve and they are asking these things, not just for FinOps, but also for their carbon footprint. (6:09) And that’s why a lot of vendors are actively working to build that into their dashboards and their metrics that they track, yeah.

    [Related: Composable Data Centers That Power Enterprise AI]

    Ken Kaplan: Yeah, I mean, anything you want to say to people who are just getting into FinOps? (6:21) How do you see this evolving.

    Mayank Gupta: A lot of people think that, Hey, you know what? This is a rum part of business, I kind say is that this is the kind of business which your CIO, the CFO, and don’t be surprised, even the CEO would actively track. So it’s a great line of business to be in because this does touch the bottom of the company, the dollars that you save, other dollars that can be applied in other parts of the department. So it’s a great field to be in. A lot of companies are getting into a hybrid, multi-cloud environment. As they’re using genai, they are realizing that they have to actively track these dollars. They cannot just use and run this forever without having some sense of accountability as the regulations for carbon, as the regulations for cost come even more mandatory than in that regards. It’ll be a more exciting field to run. There’ll be always more to be done, and there’s always be an opportunity to be in the limelight because you’ll be saving your company dollars and saving your line of businesses dollars, and you could be the hero.

    Jason Lopez: Mayank Gupta is Director of Product Marketing at Nutanix. He spoke with The Forecast’s Ken Kaplan.  Learn more about managing FinOps with Nutanix Cloud Manager software. Search Nutanix NCM Cost Governance. This is the tech Barometer podcastI’m Jason Lopez, Check out our other stories at the forecast website which you can find:

    nutanix.com/theforecastbynutanix

    10 min
  • Dartmouth College IT Director builds modern IT platform for AI

    Video interview with Dartmouth College Director of IT Infrastructure Services Ty Peavey, who tells how his team chose the Nutanix Cloud Platform for managing virtual machines and container orchestration to support the university’s growing need for AI capabilities.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Transcript:

    Ty Peavey: Some people have the thought process that AI may replace what they do every day, and then there’s others that look at it to elevate what we do every day. And I kind of am in that camp. I really believe in treating AI as a tool. It’s a way to do things easier, faster, better, but at the end of the day, the human really matters.

    Jason Lopez: The 2025 Enterprise Cloud Index Report found that AI applications are driving productivity, automation, and efficiency across industries. At Dartmouth College, an Ivy League Research University in Hanover, New Hampshire, that trend is playing out in real time. Ty Peavey Dartmouth’s, director of Infrastructure Services, explains how his team uses Nutanix software to modernize IT operations and support researchers who increasingly rely on AI in fields such as medicine, neuroscience, and environmental science.

    [Related: Orthogonal Advantages of Cloud Native Technologies]

    Ty Peavey: We have some of the best faculty in the world that really have some challenging needs. They all want access to some sort of ai. Some of ’em have small budgets, some of ’em have large budgets, and it’s really been a challenge to try to help them adopt LLMs into their research these days. And so as we look at tools like Nutanix and what we’re seeing there, we’re hoping to see that simplified somewhat and maybe commoditize and get access to LLMs on-prem early. It does take a lot of effort to help translate what they’re looking for and really appreciating the kind of resources both from people and hardware of how to deliver that. There’s also a lot of need to access AI with our vendors like Azure and with AWS. And so we try to facilitate those conversations as best as possible. We have a research group that also does infrastructure, and they’re kind of the pioneers. They’re the early adopters. They really paved the way for more of the generalists, like my team, to learn about AI.

    [Related: Public Sector Adopting AI, But Serious Gaps Remain]

    One of the things that we’ve done recently as more of a team building is we’ve worked with them to use some of the hardware that they’ve gotten under research to build our own infrastructure LLM. Really the goal is to try to understand the vocabulary, understand the platform and the architecture. What are the layers that go in to bringing in your own LLM? And really training it is the other thing. We have this fun approach that we want to take all of the documentation, all of the tickets we’ve processed over the years and train our LLM so that we can ask it all kinds of questions that we do every day. So it’s a lot of fun. We’ll do it for a while and we’ll take it down and then we’ll move on to the next thing.

    [Related: Legacy Health Turns from Broadcom VMware to Nutanix]

    We’ve traditionally had really siloed roles dedicated to storage, Linux, windows, all of these job descriptions that were very unique. And so when we got away from three-tier technology and moved into hyperconverge, we were able to break those walls down. I’m very proud of that, and I think my team has adopted it quite well. We no longer have a Windows sys(tem) admin. You may be on my team and work on Kubernetes in the morning and Nutanix in the afternoon. We brought in Kubernetes about a decade ago, really kind of some niche places that we could roll out containers. Today we have over 400 containers, over four Kubernetes clusters. Two being in Nutanix with NKP and two being in EKS. We find the adoption is great. I think we’ll always have a fair amount of full VMs, but we anticipate as our application location start supporting more and more containers. We’ll see that number grow as well. And I think what we’ve found in that is really good job satisfaction. People like what they’re doing. It is challenging at times a good challenge. And Nutanix has really helped facilitate that for us. I mean, we’ve dodged the Broadcom bullet a bit and we’re in a really good place and we’re really happy about that decision.

    [Related: Search for VMware Alternatives That Meet Existing and Future Needs]

    A number of things have shaken up the market and shaken up the technology. Definitely the acquisition of VMware from Broadcom is a first for me in my 30 years. Supporting infrastructure, adapting to that and trying to come out from the other side with an equal or better technology stack for our users has been really important. At the time, Nutanix and I think still does, you can buy it with AHV or you can buy it with VMware. And we literally took a vote in my staff like, okay, you’ve, you’ve seen VxRail, you’ve seen Cisco’s product, you’ve seen HP’s product, now you see Nutanix. What are we going to do? And we unanimously said, yeah, let’s go Nutanix and let’s go AHV. Let’s go all in. And it was a bit of a shock. I didn’t expect it, but it was unanimous. And the transformation had, I don’t want to say easy, but it wasn’t bad. There was a lot of similarity to concepts that we already knew. A lot of the fears were just in our head. So I would say it was an easy adoption, if you would. My team will probably hate that I said that, but it is. I think we were better for it, and I think Dartmouth is better for it.

    7 min
  • Spark Systems COO Built Powerful IT System for FX Trading

    In this video, Spark Systems Chief Operating Officer Chaitanya Peddada describes how migrating to a modern IT platform powered by Nutanix’s hyperconverged infrastructure technology boosts the FinTech firm’s high-speed foreign exchange trading capabilities.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Transcript:

    Chaitanya Peddada: It’s an extremely competitive space, and the FX markets aren’t like the stock exchanges which start at a particular time and end at a particular time. It starts up Monday morning and goes throughout the week right up to Saturday. So maintaining that is extremely important. Customers expect to see the best prices exactly in real time, no matter where they’re sitting in the world. It’s sub-milliseconds, which means it’s microseconds.

    [Related: AI Ambitions in Financial Services Tempered by IT Infrastructure Challenges]

    Jason Lopez: The largest financial market in the world, the FX market, is a global decentralized network where banks, financial institutions, hedge funds, corporations, and traders exchange the world’s nearly 180 currencies at negotiated prices. Singapore-based Spark Systems is a financial technology company that creates and runs high-speed, ultra-low-latency FX trading platforms. Chaitanya Padada of Spark talked to The Forkast about its goal to double average daily trader volume.

    [Related: The State of IT is Moving to Trusted Vendors]

    Chaitanya Peddada: In order to do that, we have to ensure that we have all of our basics right, which means the software services on our side, the customer support on our side, and most importantly the infrastructure, because that’s where it takes a lot of lead time in order to set up the infrastructure, which we maintain. And this year we made the transition to move our entire Singapore data center onto the Nutanix stack, which allowed us to not spend money on our SAN storage devices, which would have to be renewed this year. So we used Nutanix Move to move from the legacy system to the new stack, and we also were able to add in a couple of extra servers to have a much larger stack, and we are ready to service clients or add customer servers up till the end of the year.

    We are in a very tightly regulated space, so there are not just regulatory requirements but also requirements imposed by our clients, especially the banking clients, right? They have tight requirements, and these are not just security requirements but even data storage requirements. And that’s the reason why we host our own hardware. We don’t host customer data in the cloud. And because we have to maintain our own hardware stack, it’s very important that we, apart from our own software, which is Java-based technology, we partner with the right partners right from the OS level to the VM level, as well as the entire infrastructure stack.

    [Related: How Ivy League Dartmouth College Moved to a Future-Ready IT Platform]

    We host our own software in data centers, which are maintained by us for the customer, which means as soon as the client signs up, within days, if not weeks, they want to start up and start trading right away, right? So, having the right platforms in place, being at the right locations, which are the FX hubs around the world, is extremely important. Apart from that, latency plays a very key factor. We need to be in the sub-millisecond space, right? And those are the sort of benchmarks we use while testing any third-party product. So, whether it’s the OS or whether it’s the VM solution that Nutanix provides, these are some of the basic test cases that we run through, apart from load testing and other FX parameters.

    [Related: AI’s Rising Tide Confounds IT Decision Makers]

    We also consume a lot of data, and per day, there’s an immense amount of market data that’s going through our systems, as well as the client’s proprietary trading data, orders and trade data. All of this data has huge potential in order to leverage for building TCA systems or AI-enabled TCA systems, right? What we are looking at doing at the moment is to use this data in order to give our customers trading signals before time. We also have a library of algorithms which have predictive capabilities. Now, when we get into the predictive space, it’s very important for vendors like us to be able to assist our clients in order to provide new strategies for them, because in terms of predictive AI or algos, we need to really partner with the right customers. The traders are the people who come up with the ideas, but we need to be ready to implement those ideas as soon as possible. So this also allows us to leverage on the cloud technology, which we haven’t done so far, but even over there, I think Nutanix is probably the right partner for us to have on-prem services as well as cloud services using the same technology stack.

    [Related: Legacy Health Turns from Broadcom VMware to Nutanix]

    This was the year we transitioned onto Nutanix in a big way, right? We had initially started off in 2022, experimenting in our London data center, which is a smaller footprint compared to Singapore and Tokyo for us. And this year, we took the decision to transform the entire Singapore data center onto a Nutanix HCI stack. The most interesting thing about this transformation was, as I told you, we have only a small window to do a migration per week. So it’s one and a half days per week. And over a period of six weekends, our team managed to transform the entire stack onto a Nutanix HCI stack. We’ve gone issue-free so far, which is a first, probably in my career. We haven’t had any migration issues because there was also a lot of support from not just from Nutanix, but also from Nutanix partners to practice and prep for this migration process.

    7 min
  • Making of Nutanix AHV Hypervisor a Top Alternative

    In this Tech Barometer podcast, former Nutanix engineers Mike Cui and Greg Smith describe the genesis of the Nutanix AHV hypervisor and the confluence of industry changes feeding its growing success.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Transcript:

    Mike Cui: Well, this is where we have to go into the story.

    Greg Smith: Maybe Mike, since you were there before me, you can provide some of the context and the lead up for what happened.

    Mike Cui: I came up with an idea and I convinced a whole bunch of people we should do this. Not that we should do this, but we could actually do this. It’s possible.

    Greg Smith: As I remember it, when I joined, I thought I was joining the Stargate team to work on storage and performance tuning, and Mike pulled me aside and said, well, we need this one change in Stargate, but then you should really go and work on making KVM a product.

    Jason Lopez: Even in IT, life happens. Innovations often emerge when you don’t plan them, and that’s what happened to Mike Cui and Greg Smith. Today, Mike is a principle engineer at Annapurna Labs and Greg works at Foxglove Technologies. In 2013, they were at Nutanix and back then the market wasn’t expecting another hypervisor.

    Kanchan Mirani: It was very much VMware who created that category.

    Jason Lopez: Kanchan Murani is senior director of strategic marketing at Nutanix. She said VMware was a pioneer of the hypervisor and grew into an industry leader.

    Kanchan Mirani: At some point, VMware was so far ahead. It was really the only hypervisor out there that any enterprise would seriously consider.

    Jason Lopez: Until AHV.

    [Related: IT Resiliency: Running Two Different Hypervisors]

    Mike Cui: I had an idea for something like AHV.

    Greg Smith: I guess we were the creators of AHV. Yeah.

    Mike Cui: This is where it’s not like we had a grand plan. It is tied with the story of how we accidentally got our first customer and what we had to do to solve the problem for this customer, and then eventually evolving into a full-blown product.

    Jason Lopez: Before delving into the story of how AHV came to be, let’s look at what it is. Nutanix AHV, or the Acropolis Hypervisor, is virtualization software that enables it to run VMs. It’s a core component of the Nutanix Cloud platform which is ever evolving. With the latest update, as of mid 2025, AHV is at version 10.0. Here’s how Kanchan describes it.

    Kanchan Mirani: It’s a pretty core infrastructure software piece, which helps IT teams manage all their resources, essentially. Think of an IT team that’s getting requests from all different departments and lines of business who want to run their specific applications and they’re trying to manage all their resources. Now, imagine if they had to put in actual boxes for each request that each line of business came to them with, versus having a sort of bank of resources, which they’re able to allocate really easily without actually doing physical connections and so on and so forth. So virtualization helps you do that because it makes it software based. You can give somebody a machine, which is a fractional part of a server and say, here, go develop on this, and you can do that all with software through the hypervisor.

    [Related: Legacy Health Turns from Broadcom VMware to Nutanix]

    Jason Lopez: Back in 2011, AHV didn’t exist, but at the time, Mike was on a team pushing to get the company’s first version of its Power App out. Power App was designed to provide a simple and efficient way for users to create, deploy and manage virtual machines.

    Mike Cui: We shipped it and then throughout 2012 things were mostly chill. Like people can finally relax.

    Jason Lopez: A year later, Nutanix decided to expand support beyond VMware to other hypervisors such as Microsoft’s, HyperV and kernel based virtual machines or KVMs, the open source virtualization technology for Linux operating systems.

    Mike Cui: We installed it. It was our first attempt at making something work in Japan. The customers were asking us, how does this KVM thing work? Where’s your user interface? I’m like, we don’t have any.

    Jason Lopez: Mike thought the customer wanted KVMs because they knew how to use them, but they didn’t know anything about them.

    Mike Cui: We’ll go back and work on it, and I tried to build something during the same trip. The night before I was flying back, we had a Friday all-hands meeting. We just did a deal with a three-letter government agency, and this deal was going to be all KVM and this is me still sitting here in Japan, asking: like what? I just had to explain to our first customer that we have nothing.

    [Related: Search for VMware Alternatives That Meet Existing and Future Needs]

    Jason Lopez: Mike saw another customer was eager to run Nutanix on KVM and knew it was high time to create that GUI.

    Mike Cui: Right about when we closed the deal, they realized they didn’t have any money left to buy ESX licenses.

    Jason Lopez: The customer in Japan could no longer pay for the VMware hypervisor because they had spent their budget on buying new Nutanix software and new flash memory powered hardware. As Mike recalls, they held Nutanix in such high regard. They believed it was possible to move from ESX to the KVM. They were all in on Mike and his team to get them up and running on the new high hypervisor.

    Greg Smith: There’s no way that any customer in the modern era would’ve accepted this as a solution, right? But because they had no experience with virtualization, we got away with a lot, and so this really scrappy, scripted, hacked up solution. How much time do we spend on that? A couple of weeks, right? 

    Mike Cui: It was all you. I don’t know (chuckles)

    Greg Smith: Yeah, it wasn’t more than a couple of weeks. The first week I spent on doing the Stargate code, a snapshot from NFS to SCSI, and then I think it was two or three weeks working on the scripts and then you went on vacation for two weeks.

    Mike Cui: Oh, yeah. I went to try that.

    Greg Smith: So I’m brand new at the company. Like here I’m defending this…thing that did not feel like a product at all.

    [Related: The State of IT is Moving to Trusted Vendors]

    Kanchan Mirani: When Nutanix came along, we essentially created AHV from an open source base to begin. At that point, we were trying to optimize what’s possible in the field of software-defined computing. We had the storage piece. We wanted to extend it to computing, which is where the hypervisor came in, and over time we developed it into a pretty world-class hypervisor that was very competitive with what was out there.

    Jason Lopez: AHV became a real alternative to VMware. Over time, this evolved with steady but slow adoption, but the latest chapter in the AHV story started when Broadcom acquired VMware.

    Kanchan Mirani: We’ve always supported freedom of choice. So in our platform, we support ESXi, which is VMware’s hypervisor as well as our own AHV. And we’ve seen a steady growth in the percentage of our customers who are using AHV. We’ve seen a steady growth since the beginning, but it definitely got a bump and continues to grow after the acquisition, so that’s an indicator of where the customers are putting their choices now.

    Jason Lopez: Between August, 2023 and April, 2025, the adoption of AHV as a percentage of cores grew significantly, which means customers run more of their Nutanix powered IT operations on the AHV hypervisor. Greg said simplicity and optionality are built into the core of AHV.

    Greg Smith: Part of what made VMware such a great product to use as an engineer was like all of these nerd knobs that you could just tweak out to the nth degree, but it led to a very complicated experience, especially when you had to consider things like upgrades. So we decided very deliberately to try and sidestep all of that by presenting a very simple interface that we expected never to change. That was a huge part of our philosophy. For a long time, we thought that the surefire win for AHV as a solution was like test and dev. We had made it work for test and dev internally, and we said: Look, we know that this works.

    Jason Lopez: Greg and Mike say there was some luck, too. You can have the right philosophy but the wrong timing, and in their story, they didn’t think they were building a full-blown marketable product, but a solution for a single customer.

    Mike Cui: I did not expect AHV to have that level of success. I just feel like we were building something and we got it to work. That was already a huge accomplishment.

    Greg Smith: It just had a knack of being in the right place at the right time with the right things going on all around it, various things from the open source world that we were able to take advantage of.

    [Related: How Ivy League Dartmouth College Moved to a Future-Ready IT Platform]

    Mike Cui: Oh yeah.

    Greg Smith: Like VMware trying to stop us from shipping ESX, if that hadn’t happened, we wouldn’t be shipping AHV as broadly as we are today. It wouldn’t have had as much exposure.

    Mike Cui: I think things just happened to align when I went to Japan to set up the first customer. At that time, there was sort of no concept of vSwitch on Linux.

    Greg Smith: Oh, yeah.

    Mike Cui: You use Linux bonding if you want to bond two interfaces, and then you would apply a VLAN-tag on top of the bond. Then the issue is, well, if you use both bonding and VLAN tagging at the same time, now all your performance is crap for some reason, right? Like that year is when (Nutanix) AOS went mainstream and all of a sudden it worked. We did everything we needed to do. It worked.

    Greg Smith: Really well.

    Jason Lopez: Greg Smith works for a robotics company, Foxglove Technologies. Greg Cui is a principal engineer at Annapurna Labs, provider of custom chip software and accelerators. This is the Tech Barometer podcast. I’m Jason Lopez. Tech Barometer is a production of The Forecast. Check out The Forecast on Nutanix.com for a more detailed print version of this story, “Birth of a Hypervisor that Unleashed a Hybrid Multicloud Era.” Again, at The Forecast by Nutanix.com. And once you’re there, check out some other technology stories at The Forecast. Thanks for listening.

    10 min
  • Nutanix CEO on 2025 IT Trends

    In this video interview, Nutanix CEO Rajiv Ramaswami discusses 2025 IT trends, including migration to trusted vendors, the rapid evolution of AI technologies, and mastering the long-term state of hybrid multicloud environments at scale.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Transcript:

    Rajiv Ramaswami: Now it’s time for us to look at the most modern or modern applications, AI. As the year progressed, we saw AI, generative AI evolving at a rapid clip. Here we’ve expanded from just doing simple inferencing to providing you a full set of age agent frameworks and models.

    Jason Lopez: At the .Next 2025 conference in Washington DC, The Forecast caught up with CEO Rajiv Ramaswami after his keynote to briefly chat about technology and market trends.

    Rajiv Ramaswami: If you’re a customer, I think everybody wants to know about what’s going on now? What are the trends? What’s Nutanix doing? How can we help these days? Of course, top of mind topics for everybody is what’s their future end state when it comes to a vendor that they can partner with for the long haul as they migrate away from their current solutions. So that’s a very top of mind topic for everybody. And then we quickly get into a discussion of trends and where things are going for the future and how we can help them. For the longest time, VMware was the established infrastructure vendor for many of these companies. They’ve had 20-year-plus relationships with them, and I was in Japan, and I think there was a growing awareness two years ago. They weren’t particularly concerned. Last year, they started getting concerned.

    [Related: Nutanix CEO Stokes Surge in IT Ecosystem Partnerships]

    This year they’re actually moving. For example, here at the show, we had Toshiba talk about how they’re planning the migration away from Vme Nutanix. And you’re seeing that, right? I mean, some of this is also from certain cultures. There’s a lot of faith and trust built up over the years, and they want to work with somebody they can trust. So we are seeing a lot of that certainly happen over the last year. We’ve gone even in terms of what we’ve been able to bring to market and made tremendous progress. And it’s still not mature. There’s a lot going on every day, and I don’t know what it’s going to be. It was regular AI, generative AI, then agentic AI, I don’t know what’s coming next. Reasoning models came in recently. So this is rapidly evolving. I think hybrid cloud is pretty much established at this point.

    [Related: Growing a Multibillion Dollar Data Security Company in the AI Era]

    Everybody realizes they’re going to be operating in a multicloud environment with some workloads running in the data center, some at edges, some in the public cloud. So that model is very well established. And almost every customer I talk to has made these comparisons. They’ve looked at the cost of running in a public cloud, whether on-prem, they’ve looked at the security issues, all of those things, and they’ve decided that they’ll figure out some things that were done in the public cloud, some and on-prem. So that’s well established at this point. Now it’s a matter of more executing on that at scale.

    3 min
  • Cohesity CEO Sanjay Poonan AI-powered Data security

    In this video, Cohesity CEO Sanjay Poonan tells The Forecast about advice he got from Nvidia CEO Jensen Huang that led him on a journey to build a multi-billion-dollar, AI-powered data security company.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Video transcript:

    Transcript:

    Jason Lopez: Cohesity is a cybersecurity company led by Sanjay Poonan. He told The Forecast how he came to Cohesity on advice from Nvidia, CEO, Jensen Huang.

    Sanjay Poonan: He said something that really rang with me, which is, you’ve done the big company stuff, 20 billion, 10 to 20 billion at SAP, six to 12 billion at VMware. You haven’t taken a small company and made it big, so you should try that. And for me it was a little bit of like, can we take something that was small and make it an iconic company? SAP and VMware were already iconic when I got there, and we of course made it even better. That’s what I saw as the opportunity of Cohesity. It was a world-class company. It was about a $300-million company when I joined, with fantastic technology and an incredible founder. In fact, many of the customers told me it was the best tech they’d seen since VMware, so there was no tech risk there. We needed to scale the go-to market to get it to be a multi-billion and the leader in the space. We were number seven. So I saw a tremendous opportunity at the center of this notion of data security because of ransomware attacks and all the pressure on data to create a leader, and that’s what we’re off doing today.

    [Related: Composable Data Centers That Power Enterprise AI]

    The entire product is based on notions of machine learning and AI. Before generative AI, the founders were Google folks who built a lot of key capabilities of ML and AI into the product. A lot of these cybersecurity data detection of entropy and anomaly detection are AI algorithms. So that was from the get go, built to be a key part to the proposition of the product driven by Generative AI. We started working with Nvidia. Nvidia made one investment in this space. They looked at all the tech companies and picked us to do an equity investment because we cracked an important problem of how you could use retrieval augmented generation, also known as RAG, directly on secondary and backup data. We invented that. We in fact patented the idea. Nvidia loved it, and we were featured at GTC in both 2024 and 2025. Jenssen’s words were Cohesity backs up the world’s data and they’re building their RAG application called Cohesity Gaia on top of the Nvidia platform. We’re working with them and Google and Microsoft and Amazon who are also those four companies, I think pioneers in what the world of AI is going to look like. You can imagine a problem like this where hundreds of millions of PDF documents sit in our backup. You can easily query that, summarize it, get insights into sort of flying a flashing of torch light right into your data. That is going to be a tremendous advantage to not just keeping the data secure, but also getting insights into it.

    [Related: Raising Ransomware Resistance]

    Today, the number one problem is data resilience. People’s data is under attack from nation state actors, cyber criminals, and we work with 13,000 customers. The largest of the largest 85% of the Fortune 100, 70% of Global 500, and it’s the name brand and financial services, tech, telco, healthcare, you name it. The public sector are our customers. So their number one priority is keeping their data safe from ransomware attacks. And we’ve designed something that’s got the fastest cyber recovery and enormous amount of advanced security. Now, what they want to be able to do is, let’s take that a step further and using AI tools, get insight into that data, and we have a very good sort of two-pronged engine approach to how we approach our company’s ambitions. Number one, we innovate to be the best product in the industry, and that’s been always the pride of Cohesity, I hope, and we will continue to do that.

    [Related: Protecting Against Ransomware at the Data Level]

    Number two, we obsess about our customers. We get an anchor customer, particular vertical. We study their needs, we drive what they need. And then every customer, that vertical becomes automatically like that anchor customer. You pick the largest bank, make ’em successful, pick the public sector agency and the federal, the biggest telco, the biggest tech company, the biggest hospital. And when you do that, you can repeat that. That’s the playbook I learned at SAP and at VMware and now you get to apply it at a smaller company. We’ve gone from 300 million now with this major acquisition we’ve done at Veritas. We’re now on the path to being a 2 billion company and the biggest space, number one, but the enormous amount of innovation and we’re pouring a lot of engineering including into our exciting new work we’re doing on Nutanix, which I’m very excited about.

    5 min
  • IT trends impacting IT departments

    In a video interview, Nutanix’s Lee Caswell tells how IT customers need to meet changing needs as they embrace AI capabilities.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Video transcript:

    Jason Lopez: In between press briefings, Lee Caswell tells The Forecast about some of the biggest IT trends, challenges and innovations explored at the 2025 .NEXT event in Washington, D.C.

    Lee Caswell: We’re really interested in that customers get value out of AI. And this is one of the things where a lot of customers today are a little bit stalled trying to figure out, well, hey, I don’t want to spend too much. I don’t want to spend on the wrong thing. How do I get started, right? And how do I also make sure that the AI that I bring to market has the same enterprise level resilience, the same day-to-day operations, the same security and privacy that I bring for all of my traditional apps. How do I make AI just the next enterprise app? And so we’re helping customers basically with choice. So the idea is, what are the fast-changing parts? GPUs are changing pretty fast, so we’re going to give you the access to the GPUs or CPUs with acceleration, for example. LLMs are changing fast. Last year, we gave you almost an app store, if you will, into LLMs from our partners Hugging Face and NVIDIA. This year, now we’re supporting agentic workloads, which is an iterative cycle where it’s not enough to just get an answer. Now, I want to iterate on that answer and give you more relevant results, maybe through guardrails, maybe through re-ranking, maybe through embedding. And so we’re giving you all of those access points so you can get started today and have a production-ready enterprise experience.

    Jason Lopez: Caswell talked about how enterprises are onboarding AI and managing data privacy.

    Lee Caswell: Well, I think first off, we’re helping customers understand the fast pace of the AI market. So most customers today realize that large language models, or they don’t have to be that large, it could be SLMs as well, will be developed either in the public cloud or by some of the largest customers because of the incredible capital expense of training a model. So now customers are looking and saying, well, all right, I’m probably going to get access to a model, but how do I make sure now that my data remains private? It could be, for example, that your model becomes more sensitive than the data itself because the model now starts showing you’re inferencing your ideas, your insights into the data that you have. And so that idea of saying, it’s going to be important to access large language models and change them out. I want to have access to that. We provide that capability. And then I also want to have this ability to go and have more responsible results and be able to go and manage the infrastructure so that it’s actually an enterprise-level experience. That’s how we’re helping customers basically get started with our Nutanix AI, enterprise AI products.

    Jason Lopez: He tells how Nutanix software helps enterprises build and run a future-ready IT operation.

    Lee Caswell: Yeah, it turns out, right, you know, the quality of our engineering has been really something that’s just been delightful to see as we’re bringing all the elements necessary to bring AI into the enterprise. And so that could be, for example, like the storage, right? You have to ingest that data. You’ve got to run the model. You’ve got to archive the data. How do you do all of that together? We provide an offer for that. In addition, by the way, many of these models are containerized. And so as you bring containers and Kubernetes, for many customers, these are relatively new concepts into the on-prem world. How do you leverage what you know already today? And here we acquired a company called D2IQ, and we have a terrific set of engineering leadership resources there who are now taking that, what was a production shipping product, and now adding Nutanix data services. It’s an incredibly compelling offer and part of the complement of products you need to make AI successful.

    4 min
  • Liqid CTO Sumit Puri Composable IT infrastructure

    In this video interview, Liqid CTO Sumit Puri explains why dynamic IT infrastructure that taps into pools of GPUs and scale-up memory can quickly and efficiently run virtual machines, containers and AI workloads on-premises and at the edge.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Transcript:

    Jason Lopez: Liquid is a software-defined infrastructure company that enables more efficient use of high-cost, high-power resources like GPUs in data centers. Instead of installing GPUs directly into each server, Liquid creates centralized GPU pools and dynamically allocates them to servers based on workload demands. This composable approach maximizes GPU utilization, reducing waste, and improving performance.

    Sumit Puri: We saw this vision of GPUs being important in the data center many years back, and then all of a sudden this thing called ChatGPT burst upon the scene and made this front and center in everyone’s mind. And so we’ve been focusing on pooling and sharing these resources for a long time, and now all of a sudden AI is forcing itself into the mainstream, especially in the areas that we focus on, things like the enterprise, and now all of it’s coming kind of market for us in a very, very good way. And so we ended up building the product, and the market ended up coming our way.

    [Related: Nutanix CEO Stokes Surge in IT Ecosystem Partnerships]

    Jason Lopez: Liquid addresses the growing demand for AI inference workloads, especially for enterprises that want to keep their data on-prem instead of moving to the cloud.

    Sumit Puri: What Liquid focuses on is very much on building power-efficient, cost-efficient solutions for inference. And it’s interesting to think whether it’s on-prem or it’s in the cloud, a lot of the data, which is what a lot of this AI is driven on, it lives on-prem. Eighty-three percent of the data is actually on-prem. And so one of two things must happen. We either must move the data into the cloud, or we must bring the GPUs on-prem. We think there’s a lot of customers who are not willing to wholesale move their data to public cloud, and so therefore we want to build efficient solutions to allow them to process their data on-prem.

    [Related: Swarms of AI Agents Powering Businesses and Daily Life]

    Jason Lopez: There are advantages of pooling and sharing GPUs instead of deploying them directly in each server.

    Sumit Puri: There’s three primary benefits why somebody takes this journey of pooling and sharing the GPUs. One is around performance. I need that server to have a lot of GPUs because I need it to run very fast. We’re not limited by two or four or eight in a box. We can compose 30 GPUs to a server and give you the fastest servers on the planet. That’s one reason. The other is cost. If I have to deploy 30 GPUs, do I want to buy four servers, put eight GPUs in every server, buy a bunch of networking? Or do I want to buy a single server, deploy 30 GPUs, reduce my cost, reduce my power, and have a more efficient way of deploying these resources? And the third reason is agility. It’s very hard to predict, will my workload need one? Will it need two? Will it need four? Will it need eight? Do I use an H100? Do I use an A100? Do I use an L40S? There’s too many choices, and so we try to take the guesswork out of it. Let’s put a centralized pool of whatever device type we need and pick the right tool for the right job at the right time.

    [Related: Search for VMware Alternatives That Meet Existing and Future Needs]

    Jason Lopez: The AI infrastructure landscape is evolving from training to inference, especially for companies that are not building foundational models.

    Sumit Puri: The first chapter of this entire AI journey was very much focused on training and building these foundational models. And the way that we see that going forward, there’s probably only going to be 10 companies on the planet that can afford to build these massively large 100,000, 200,000 GPU clusters to build the foundational model. The other 100,000 customers that are out there, they’re going to take these models, open-source models like Llama, they’re going to bring them on-prem, they’re going to fine-tune that model, they’re going to do RAG, they’re going to do inference, and that part of the journey now is just starting. We think by the end of the decade, inference actually is going to be a larger piece of the overall AI pie than is something like the training portion of it.

    Jason Lopez: Liquid makes AI inference more accessible and efficient for enterprises, especially with Kubernetes and model deployments.

    Sumit Puri: So NVIDIA has a big push for something called NIMS, NVIDIA Inference Microservices, which is basically a container. And what they’ve said, it’s very, very difficult for people to get all the layers of the stack perfectly right to deploy these models, and so we’re going to containerize these models, and that is the way that enterprises are going to go off and deploy this. We have a plugin for our solution where what we do is we suck the container in, we probe the container, we figure out what type of GPU and the quantity of GPU in the backend, and we connect that GPU resource to the specific server in the Kubernetes cluster, then we deploy that container on that machine, so you have this perfect matching of hardware to container, and we automated the entire process. We’re at a point now where we have one-click deployment of inference. You say, give me Llama7b go, and we will automate the entire process on the backend, and within two minutes, give you a model that you can speak with. When you’re done with it, you hit the delete button, we’ll rip those GPUs off, put them back into a centralized pool so that next container, that next model that you’re looking to deploy has resources to use. We think that’s how you get there, is you have to make it easy for enterprises to deploy these things. They don’t have the large armies of data scientists to do this on their own, so the more that we can automate this, the easier it is for those companies to consume, the more that we can make it more efficient, and the way that we think about efficiency is tokens per dollar and tokens per watt, because that’s what the enterprises are limited by, they’re limited by power and money. If you’re a hyperscaler and you have unlimited power, unlimited money, we can’t help you. But if you’re an enterprise, that is the metric you need to think about. Tokens per dollar, tokens per watt, automation, ease of use, those are the things that are needed to get AI to scale.

    [Related: Bracing Data Centers for Wave of AI Workloads]

    Jason Lopez: The company supports virtualization and dynamic GPU allocation in enterprise environments.

    Sumit Puri: We are a platform for a variety of different applications. One of the applications we are very well suited for is virtualization. If we think about VMs for a second, it’s very difficult for enterprises to predict which VM they’re going to deploy at what time and what resources that VM might need. We deploy infrastructure for three to five years and making that long-term prediction is very difficult. We’ve partnered with Nutanix where we can put a centralized pool of GPUs inside of a Nutanix cluster and depending on the requirements of a specific VM, we can match the GPUs on the fly dynamically, hot-plugging GPUs into servers to meet the requirements of the VMs on Nutanix.

    8 min
  • Seeing Swarms of AI Agents Powering Businesses and Daily Life

    In this Tech Barometer podcast, disruptive technology investor and analyst Jeremiah Owyang explains the rise of AI agents and a future shaped by a multiplying AI-first mindset.

    Find more enterprise cloud news, features stories and profiles at The Forecast.

    Podcast transcript:

    Jeremiah Owyang: The biggest industry right now in AI is the AI agent space. It’s destined to grow 10 times every five years. Right now it’s estimated at 5 billion, which will go to over 65 billion, and then it’s estimated to be 500 billion just within the next several years. This is the hot market growing within the AI space right now. People use the internet. You physically go out and find different websites to get information. You fill out tasks and you order flights, you order e-commerce, you get things done even inside of your enterprise. You have to fill out expense reports and timecards. All of these things are wasteful and not great experiences in the future. Your AI agents are going to go out and do those tasks for you. The information will be reassembled in the way that you want when you want in a multimodal way, whether it’s text, video, ar, vr, whatever time of day that you want, and in the amount of information that you want. It’s no longer up to the web designer. It’s no longer up to the website or the app designer. It’s up to you.

    [Related: 4 Trends Defining the Future of Enterprise AI]

    Jason Lopez: What you heard are excerpts from a keynote speech by Jeremiah All Yang, general partner at Blitzscaling Ventures. He also leads an event called the LAMA Lounge and AI community of hundreds of AI company founders. We recently interviewed Jeremiah to tell us more about agents.

    Jeremiah Owyang: AI agents are dependent upon large language models.

    Jason Lopez: In just a moment, Jeremiah explains the importance of LLMs to agents. But let’s start here. An agent is software code that allows a device to understand its environment process information then acts autonomously to take actions to achieve specific goals. AI agents can do more than common AI assistance that respond to specific user prompts. An AI agent can self task, it can learn over time and even recruit other agents to help. They operate across multiple apps, often working silently in the background even while we sleep. A smart home thermostat or robot vacuum or autonomous vehicle, or just a few examples of AI agents in action. In many ways, AI agents are less like tools and more like digital organisms adapting and evolving in a world built from code.

    Jeremiah Owyang: As they become more advanced, they operate in sequence, in combination with each other. In concert, which is called the agent crews or fleets or groups or swarms. In many cases, they are like low level human employees.

    [Related: Managing Enterprise AI Sprawl]

    Jason Lopez: So how does an agent know what to do?

    Jeremiah Owyang: They get it from large language models. They’re dependent, so it’s not if they’re a replacement of large language models, they’re actually executing the tasks. So think of large language models as your brain with all that knowledge and the things to do, but the actual clicking and typing and doing the actual physical tasks like your limbs, your appendages, your body are the AI agents, and you need both in combination to be an effective quote creature.

    Jason Lopez: From the point of view of an enterprise, you might be wondering if IT infrastructure needs to change as companies shift from just using LLMs to deploying AI agents. The answer would be yes. An IT team needs to prepare for how agents will interact with the company’s systems. This means choosing the right tools or partners and deciding on what permissions and access an agent should have for the tasks it could do.

    Jeremiah Owyang: It could fill out your expense report after looking at your credit card. It could create meetings or meeting summaries. It could build your monthly reporting deck and grab data from disparate places and aggregate it into one location. It could be like a virtual colleague. It could be just the two of us on a call. We could have three, four, or five assistants that are not just recording, but actually talking, collaborating, taking notes, taking actions, and generating diagrams in real time.

    [Related: Enterprise AI Reality Check: Implementing Practical Solutions]

    Jason Lopez: Many companies are already using AI assistance for customer service, handling basic questions using company data, and Jeremiah says, companies are using Gen AI in all sorts of interesting ways. He said 10% of Pfizer’s marketing content is being generated by ai. This can help workers be faster and more productive.

    Jeremiah Owyang: Now, imagine if that worker who just wants to focus on making good decisions or the relationships with customers or relationships with internal stakeholders could allow the AI agent to do all that busy work for them. Now at your company and many companies who has an executive assistant, typically it’s only provided to those at the top of the hierarchical pyramid. Now, imagine a world where every worker has one assistant wait, two assistants, wait, wait, wait, 10 assistants, 10 assistants that help with data or meetings or scheduling. Just all of those things. Imagine the level of productivity that can increase for those knowledge workers, and I think we’re on the cusp of that because if you interface with just an AI agent, it means you just don’t need that many enterprise apps anymore.

    [Related: A New Generation of Data Centers Spreads Use of Enterprise AI]

    Jason Lopez: This isn’t to say that an AI agent is just a simple addition to a team. Some companies are exploring what it would be like to rely on AI led teams.

    Jeremiah Owyang: So it’s really critical that you as a worker lean into AI and lead that within your company and for your career, or you might have to be updating your LinkedIn profile. So I think it’s critical that we lean into this technology.

    Jason Lopez: We’ve reported before as we head into the age of ai, there’s a skills gap. Companies like Nutanix have AI teams that can build AI into products or create apps that boost business productivity, but using AI as a skill that will be needed across different business units to enhance employee competency. Many companies onboard AI capabilities like Microsoft Copilot or Glean or services like writer.com. Many workers take upon themselves to acquire knowledge and certifications for AI expertise. They turn to YouTube or platforms like Coursera or continuing ed from colleges and universities.

    Jeremiah Owyang: Every worker is responsible for their continued growth. Now, for executives, there’s a new role. There’s a chief AI officer leading the charge within the organization to use these tools. Most of the innovation is happening with the young startups, many of them who spun out a big tech or spun out of big companies and they say, I’m going to move much faster.

    [Related: Business Steps for Using Agentic AI]

    Jason Lopez: Jeremiah says, small teams are capable of building a global enterprise AI software suite in one year with rapid adoption.

    Jeremiah Owyang: Crew AI only has 16 employees, and they’re already one of the dominant enterprise agent leaders because they’re using AI for everything.

    Jason Lopez: Crew AI is a San Francisco based firm, which makes development tools to build AI agents that do tasks in apps like CRM or ERP systems.

    Jeremiah Owyang: They are what is called an AI-first mentality. So let’s talk about an AI first mentality. This is a common mental framework within Silicon Valley, amongst the AI leaders. If you have a problem in your life, you first see if there’s an AI that exists off the shelf, whether it’s an app you download or an enterprise app that’s an existing enterprise software. If it doesn’t exist, then you try to build it and step three, if it doesn’t, you can’t build it. Then you hire somebody to build it. So you follow that lineage to think about, how do I move fast? And it’s always about leveraging AI first. So that’s the AI-first mindset.

    [Related: AI and Cloud Native Skills Reshape Careers]

    Jason Lopez: Building on the rise of enterprise AI platforms and the growing availability of ready-made templates. The next logical step is even more transformative. As tools become more accessible in development, more streamlined, the creation of AI agents is starting to shift as well, possibly even to the agents themselves.

    Jeremiah Owyang: We are probably going to see AI agents create AI agents. There’s already no-code AI, so developers can be even more efficient and can create even faster. I anticipate that developers and software engineers, they’re still needed. We need them more so they can just produce more. I don’t see them going out of a job. That’s our take in this

    Jason Lopez: Space. Still, the development of AI agents in the enterprise lags behind the customer side. A lot of that has to do with security.

    Jeremiah Owyang: One rogue AI system within an enterprise, potentially from a nation that is not friendly with your nation, could upend your company. It could grab data and send it back to the home country. So I think there are appropriate concerns around ensuring that the AI is safe for the enterprise and the organization eventually, which would help with the customer relationships. So I think that’s why we see that lag.

    [Related: CIOs Sharpen Skills for the AI Era]

    Jason Lopez: Jeremiah points out that while AI agents can be like coworkers, they’re not human, but powerful tools, tools that aren’t neutral because they mirror the intentions of people.

    Jeremiah Owyang: The old adage in Silicon Valley that you and I know quite well is that if the product is free, then you are the product. If you’re paying a premium, whether it’s enterprise or consumer for the AI agent, then it is more likely to be aligned for your benefit. If not, then it is more likely to be towards the benefit of the builder. And I think the old business models in Silicon Valley apply here. They’re tools, and we have used Fire for cooking, which helped us to expand our brains with more nutrients by having clean food. We’ve used steel to build amazing vehicles and buildings, but we’ve also used fire and steel for weapons, and that choice is very much a human thing where we choose how to use these tools and technologies. The difference here with AI though is that it is a thinking machine and it’s trained off what humans will do, and it starts to think or simulate thinking on its own, and that’s something we haven’t seen before in every possible way you look at this, it’s an embodiment of the human condition, and that is a wild thing to think about, that we’re creating a new species.

    Jason Lopez: Jeremiah Al Yang is the founder of Lama Lounge AI events, and he’s a partner at Blitzscaling Ventures. You can find out more about his events at lu.ma/lama lounge. This is the Tech Barometer podcast. I’m Jason Lopez. Thanks for listening. Tech Barometer is a production of the forecast where you can find more stories about artificial intelligence as well as cloud and enterprise computing. Look for those at the forecast by nutanix.com, all one word, theforecastbynutanix.com.

    12 min

About Tech Barometer – From The Forecast by Nutanix

From the publisher's feed

Explore the cutting edge of enterprise cloud computing. Tech Barometer is the Podcast affiliate of The Forecast by Nutanix, which covers people and tech trends driving digital transformation. Business…