theCUBE + NYSE Wired: AI & Retail Trailblazers | Tina Tarquinio, IBM

Clips
More from theCUBE + NYSE Wired: AI & Retail Trailblazers

Tina Tarquinio

Chief Product Officer, IBM Z & LinuxOne

IBM

play_circle_outline Evolution and modernization of mainframe technology

play_circle_outline IBM Z's role in AI, hybrid cloud, and high performance computing

play_circle_outline Revolutionizing Retail: Mainframe, AI, Generative AI, Customer Feedback, IBM Design Thinking in Product Development

play_circle_outline Empowering Startups: Join IBM's ISV Accelerator Program for Collaboration and Growth

Info
Transcript

Tina Tarquinio, IBM

Tina Tarquinio

Chief Product Officer, IBM Z & LinuxOne IBM

John Furrier and Dave Vellante discuss AI and IoT in the retail industry. Tina Tarquinio, Chief Product Officer of IBM Z and LinuxONE, talks about IBM's innovative full-stack platforms. IBM Z, with 60 years of experience, is known for its computing capabilities. The mainframe remains relevant in today's AI infrastructure. IBM Z offers AI inferencing on chip and modernization strategies for evolving technology. The platform is secure, performant, and scalable for mission-critical workloads. IBM's hybrid cloud strategy combines mainframe qualities with cloud ap... Read more

explore Keep Exploring

What are IBM Z and LinuxONE and what makes them unique in terms of full-stack platforms? add

What are some advantages of mainframe technology in today's global banking and financial sector? add

What approach does IBM take in product development and how do they prioritize customer feedback and needs? add

How does IBM engage with the startup ecosystem in terms of partnering with them and aligning their engineering efforts to work with IBM in order to leverage their customer base and technology offerings? add

bolt Powered by CUBE AI

Tina Tarquinio, IBM

search

>> Hello everyone. Welcome back to theCUBE. I'm John Furrier, host of Cube here with Dave Vellante for the three days covering our Media Week program around NRF. It's the big retail show. It's basically retail week. And obviously it's a technology show. NRF has become, like all the events, really focused on AI. And it's been an IoT show, it's been a tech show with the retail retail industry vertical like healthcare and others. A lot of devices, a lot of things going on, a lot of data, a lot of back-end support. We've got a great guest here, Tina Tarquinio, Chief Product Officer of IBM Z and LinuxONE at IBM. Tina, thanks for coming in. Appreciate you.

Tina Tarquinio

>> Yeah, thanks for having me. I'm super excited to be here.

>> So IBM Z, which we've covered on theCUBE many times. I think the last event you had was a huge event at the Lincoln Center-

Tina Tarquinio

>> Yes....

>> many, many moons ago. But the role of the mainframe now is being discussed more and more, and certainly with theCUBE and SiliconANGLE and the industry, as people see supercomputing and go, "Wow, it's like the mainframe." Well, they're still around and they're doing well. You guys have a great business with Z. For the folks watching, explain what IBM Z is right now. What is the product portfolio? It's not just mainframes, there's a lot going on around it. Explain what IBM Z and LinuxONE is.

Tina Tarquinio

>> Sure. So first, thanks for having me here. And I could talk all day on mainframes. So thank you. This is a great question. So IBM Z and LinuxONE are really full-stack platforms. So I like to say we are from chip to ship, so we really own the full-stack and an open-stack, right? So we integrate with all of the open-source technology. So it starts with this really innovation-leading silicon, all the way through our operating systems into the middleware. And now we are a full participant in the IBM hybrid cloud and AI strategy along with the new announcements around Watsonx. And so we've been a platform around for 60 years. So we know a thing or two about enterprise-grade computing for sure.

>> And it's interesting because if you go back the history of computing, IBM, obviously going back to the roots, we all know the mainframes. And they run the biggest workloads certainly here in New York. The banks remember when I talk to, it's like they can squeeze an extra second or a minute out of it because they got to clear the trades. Whoever you're talking to, they're running big time, mission-critical workloads. If you look at right now, the biggest conversations is AI infrastructure. And all of those conversations around how do I run my mission-critical workloads? Because the whole wave around the big high performance computing and now supercomputing for the masses, whether it's you talk about NVIDIA and these super clusters, they're essentially large-scale computing platforms. As Dave and I joke big iron, which was the term that we used to call mainframes. So we're back full circle. Now it's distributed computing. The role of the mainframe is more relevant than ever before because you're starting to see other components like it in the system. Super clusters over here, you got this over here. And people are paying not for servers anymore, but systems. So the role of systems is huge.

Tina Tarquinio

>> They really want a platform approach. And you said everything spot on. So if you ever need a side job in my product management team, we'll take you right on. But I'm glad you bring that up. 77 of the world's top banks, eight of the top 10 payment providers, they're running mainframe. It'd be hard for anybody to really go throughout an entire day and not interact with the mainframe somewhere. And that is because we are highly performance scalable. So if you look right behind us, at stock open, there are seconds, milliseconds matter. And we are highly performant, highly secure, and very few platforms can behave the way the mainframe does in those kind of environments. And so we often sometimes say we were the original cloud. You had virtualization, and now you have this ability to really connect this mission-critical application and data with your public cloud or private cloud applications. And that's really the strategy IBM's put forward, the hybrid cloud. And now everybody wants to bring AI. And we're so well positioned for it with the data and the applications running on the platform.

>> Tina, the big story here at NRF, there's many story lines, but some of the top ones that I like that are relevant to this conversation, is one is you have a heterogeneous computing environment in retail, obviously. Edge computing is now intelligent. Industrial is powering businesses. There's business transformations. And the role of data, whether it's computer vision, or whatever data is happening at the edge, in real time. This is the hallmarks of what mainframe had to do. You make performance, real-time trading, clearing the end of the day, moving insurance claims, whatever is going on. It's always been about real-time accuracy performance. But now with retail, you have other actors involved, other ecosystem partners that are connected. You have this, I said on theCUBE last year, "Dave, we're going to see a connected ecosystem." I know you guys are talking a lot about this in your upcoming events, this connections. It's IBM connecting with other systems, other companies. Because it's a multi-vendor world we're living in. What is IBM's hybrid strategy? Because if that vectors into that, then that's now a winning hand.

Tina Tarquinio

>> Yeah, I think you're spot on. I think the other thing we should add in there is regulated industries, payments, which is very connected to retail. Their requirements are getting more and more stringent, instant payments, immediately reconciling the payment. And everybody wants it secure. We've all been on the other end of an email that says, "Oops, something's been breached." And the tolerance for that is really shrinking. And so IBM Z has the industry's only quantum safe server. And then we pair that with this ability to be connected and scale. And so IBM's hybrid cloud strategy is really around bringing all of that together. So leverage what you've had and invested. Fit for purpose. Run what needs to run on the mainframe there for the qualities of service, and leverage all the other fit for purpose applications.

>> And Z has evolved all over the years. Mainframes, it's not just the old school mainframe, there's a that's been added to it. Can you share what's been added to Z that modernizes the mainframe, keeps the best of the mainframe performance, and brings it into this connective tissue that's now multi-vendor, heterogeneous, heavy data AI? You got Watsonx out there. There's not yet a Y, I'm sure the IBM brand's working on something Y because you got Watsonx and you got Z. But take us through what is mainframes modernization story? How would you describe that?

Tina Tarquinio

>> Yeah, so we are a highly modern platform. If you were to think about the mainframe for it's really in a 19 by 19 inch square, I mentioned quantum safe. In z16 we have AI inferencing on chip, which means you could score every single transaction and never miss your SLA. You can't do that on any other platform. When you pair that with our modernization strategy, it's all about application modernization. So modernize your infrastructure, we got to plan for that. And then application modernization. How are you bringing your older applications current? So we've introduced things like Watsonx code assistant for Z, helping developers become more performant quickly. So using generative AI on code, using generative AI for an assistant, learning how to use the system, and then pairing with Red Hat OpenShift platform, and a host of other open source applications. And that's really the web we start to weave to be connected.

>> I think that's important to note that there's a lot there around Z. Where's LinuxONE fit in? How does that connect in with Z? How would you describe that method?

Tina Tarquinio

>> So LinuxONE is a Linux only platform. So it has all of the qualities of service that you would expect, the security, the scale, the resiliency, the reliability, and it's only Linux. So you can run your favorite distro. We support all, that's our commitment to an open platform. And it's Linux only. So we are a Red Hat OpenShift platform. And working through all of those, we just announced OpenShift virtualization on the platform. So that's super exciting. And both of these will have new generations coming at the first half of this year.

>> Yeah, I was joking about Watsonx and then saying Y, there's no Y yet, but it'd be nice to have an X, Y and Z. But let's talk about Watsonx because if you look at what mainframe connects to AI, that's a big story. Here, how do I get my high performance AI systems, super AI clusters, whatever clustered systems? In retail, you're seeing a lot of the companies evolve into very high scale platforms. So whether that's headless retail systems or other systems, how is mainframe and AI working? How would you talk about that? What's your perspective?

Tina Tarquinio

>> So there's really two ways. So first is traditional AI. Think about fraud detection. If you swipe your credit card, you want to get the score back right away. So it happens in transaction. Nobody wants to have it done overnight or a couple of days later and you get that email we don't like. So traditional AI. And now they want to start to use larger models on traditional AI because if you can even improve it, just one millisecond or another percent of accuracy, really major implications to your bottom line. The other is generative AI. So how can we bring that to the platform? So we introduced new technology that allows you to run these generative AI models through Watsonx. And IBM as a whole has introduced Granite, which is our family of multi-model, multi-size, open-sourced, generative AI models.

>> It's interesting, I was recently at the IBM event in Las Vegas at CES, the UFC relationship. And I was talking with Jonathan Adashek, who's the head marketing person and some of the folks there as well as the customer. And it dawned on me that as you start to see the formation of the AI landscape, you got producers of AI and consumers of AI. And a lot of the customers, some hybrid produce. Some of the big banks, they have their own internal machine learning and they're producing some AI. But for the most part they're consuming AI. And UFC is a consumer of IBM's AI. So you have a lot of enterprises that are looking to be consumers of the AI, and they have these demand for the systems. As a consumer of AI, what do you see as Z's contribution? To me as a consumer, I'm a customer like, "Hey, I need high performance. I either have a main thing. Or maybe I should consider because the price of these super clusters are really high too." You go look at, I'm going to get a bunch of H200s or whatever cluster, and when you get in the storage fabric and the networking, and it adds up.

Tina Tarquinio

>> Yeah, you really want to consider what are your requirements? Do you need real-time answers? We can provide a inference in less than a millisecond. So that's unique to us. And I think the other thing is the cost of those clusters, but the environmental use of those clusters, sustainability starts to be a real concern. And we can do all that within the envelope of the system we have. So that's something that the mainframe is bringing there. And for the consumers of AI, they're consuming the models, really. And then they're running the models. And so you really want to optimize for inferencing at scale, which is where we started with z16. And then in our next generation we'll bring a little bit more.

>> Yeah, it's to see customers about their outcome and working backwards from that. Is there any use cases that you can highlight on the retail front or any other industries you say, "Here's an example that's very unique where mainframe and Z shine." And how does that relate to the customer's benefit?

Tina Tarquinio

>> So we have a large US bank that had a goal to score every transaction real-time. And so they were really only doing about 20% real-time. And the rest were happening either overnight in a batch process or maybe a day later. And so we work with them to actually scale. And now they are scoring every transaction real-time. And they have saved over $20 million just in the fraud detection and avoidance on the platform. So that's one example I think hits close to home for retail because it's all about the payments.

>> One of the things as chief product officer, I know it's a big job, but you have the keys to the kingdom. You have to build the products, do the requirements, but also you got to look at the customers, you got to look at the engineering. You're on both sides. You straddle the fence. You got to talk to the customers, get those requirements in, build the product. What is the state of the art right now as a product lead? I know IBM's done very well this year with product-led growth. This is a big part of it. What's your focus in terms of how you look at the landscape? Because things are happening much faster, there is need for performance. What are some of the things you're looking at when you look at the customer needs? What are they saying they need? How are you bringing that to the market? What are some of the key things that you're focused on?

Tina Tarquinio

>> So we leverage what we call IBM design thinking. And so we have almost 1500 user hours of research for it towards our next system. So we are, I would say, a little obsessive about getting the right client feedback. I can't do everything I want. I don't have an unlimited budget. So that feedback from the clients is really critical. Where do I decide to invest? And so they want to use AI to get better business outcomes. They want to use AI to help their skills always. But they are not willing to compromise on security and performance. And so those are the edges we have designed within. And we work super closely with our clients. So it's a really great marriage back and forth of what if we did this? And okay, but I can't give up that. And we bring a new system out every three-ish years, but it takes probably six years to develop it. So we're always working on at least the next one and usually the next two. And so they see that flywheel with us working through it.

>> The product managers must have a real arm wrestling match when it's feature approval, getting into the next rev. I know you got a big rev coming up. Obviously AI is the center of it. I want to ask you about something that we've been hearing a lot in the marketplace around generative AI. Obviously machine learning's been going on for a while. That's most of the bank major fraud detection. All have been in place. But there's a real conversation around resilience. Not just storage, from a storage perspective and recovery and ransomware, resilience as an organization because there's a lot of new things going on with AI at scale that you can't just iterate on. Some stuff, you know, you're the mainframe, you know this can't fail. If bank doesn't clear those positions at end of the day, bad things happen. So you've been used to this. Performance, security, resilience. As AI enters the fold, how do you think about resilience? How should customers think about being resilient because there's a high bar in the world you're in?

Tina Tarquinio

>> Yeah, to me it goes back to the trust in the model, the governance. So I think the governance, which may not sound like the most exciting technology part, everybody wants to go to ChatGPT and get their cat poem or whatever, but the governance around AI is going to become absolutely critical. Doing all that in a trusted environment. And that's really our wheelhouse at IBM. But if you look at how we've rolled out AI, we introduced watsonx.data, .governance and .ai. And so that really says to me, we're leading with those as the three most important things. And governance is going to really start to come in because you're going to make real business decisions, real decisions that impact people based on these AI models and how can you be sure that they're trusted? And I think that'll be the next wave.

>> And you have such a history. If you could look back at some of the experiences you've had with the mainframe, because again, we've been seeing it, we've been documenting it a few 15 years, the mainframe has touched some of the most mission-critical, the mission-critical workloads in companies, big companies like large scale banks. I think JPMorgan Chase, Lori told me that they do 10 trillion a day in transactions as a group. They have a 17 billion IT budget. Now that's JPMorgan Chase again. But there are customers that you have that are operating at scale.

Tina Tarquinio

>> Yeah.

>> You have all that experience looking now at this AI world, how would you talk to those enterprises as they start to look at AI as just another app? To them they love AI, but still their is another app, I'm over simplifying it, but nothing gets in unless it's bulletproof.

Tina Tarquinio

>> So I would really think about outcome focus. What are your two SLA and requirements on what absolutely has to run on the mainframe? And really go through those requirements. And wherever possible, you want to shorten those distances. So if the data and the mission-critical application are on the mainframe, your AI should probably run on the mainframe too. Really think about the end-to-end life cycle of a transaction. And I think that will guide you. Because our clients can't do everything they want either. And just to build on your scale point, I was talking to one of our clients a couple of years ago. And as a point of pride, they shared that they had delivered, they're a shipping company, they had delivered over 1 billion of vaccines during the pandemic. And I just beamed with pride because I know behind the scenes what's really driving that technology that allows them to do that. And really was a proud moment.

>> Yeah, it's super cool. I want to ask you, I just came off a call before we came on with a bunch of startups, do some advising. We actually used a bunch of startup interviews. And there's a phenomenon, and I've never seen in my 30-year career in the enterprise and covering enterprise, and startups. We are seeing a generation of startups that are targeting the enterprise first. It's not the shiny new toy. Enterprise AI is so hot even here in New York under 30, 20 something young entrepreneurs, they've come from Facebook, they've come from Palantir, they've come from these companies. They're just coming out of school with computer science degrees or other degrees, technical degrees, and they're building companies. And it's not "I want to build the next smartphone." They want to go in and solve the big problems in the enterprise. So I'm blown away. They want to go into the enterprise. Now I've never seen that before. So I have to ask, they all ask me, "How do I get into these banks?" And the common thread is, "Yeah, I might build on AWS because they'd love startups, but I need to support IBM because they're in the bank I'm trying to sell to." So how would you talk to the startup ecosystem and say, "Hey, if you want to work with us, because we have the customers..." Because you do. I'm just saying, you have the customers. The startups want to get in there, so they need to run their AI on Z. They need to run their app in a cluster, in a mainframe. Is there thoughts around that? Do you guys talk about this? And give a position on how startups could align their engineering to work with IBM as a partner?

Tina Tarquinio

>> Yeah, so I have a peer, she's really wonderful woman, Meredith Stoll. She leads our IBM Z and LinuxONE ecosystem mission. So we have an entire group who's worked on this, and we actually have something called the ISV Accelerator, which is where these startups, first they meet each other and peers. And they get started and it's really an accelerator, a cohort. And they go through and they really learn. They probably aren't familiar with the architecture underneath. So that's really the first thing. And then really understand the outcomes these clients are trying to achieve. Because in the end, we're client-focused and we really want that. So we actually have something called the ISV Accelerator that they can participate in-

>> So they onboard, knowing the IBM stack. So they want to play with Watson and mainframe, this connective tissue there. As a startup, I can bring my-

Tina Tarquinio

>> So we have something called the ISV Accelerator program, and then that's where they join in. And then yeah, of course our strategy in Z is to be open and connected across all of IBM.

>> And not to put you in the spot, but I will. If you had to describe the white space opportunity, your product leads, you know all this answers anyway. So if I'm a startup and there's white space, I can go after, what would you recommend? Because they're hungry for the enterprise. Is there a white space in this ISV environment that I can go after that you would see as opportunities to grow with IBM? Expand with IBM? Is there any technical area product features that might not hit the roadmap or it's down the road or it's not something you want to do? Or you want to enable someone to do?

Tina Tarquinio

>> So again, I'm going to be client focused. Our clients are always talking about skills and how they can gain more skills for their team members and how they can build skills long-term. And so I would encourage any ISV to think with that in mind, how can I help... We all want to help our clients solve their problems. And so skills is definitely one of those. And then after the skills is always being the most resilient platform they can have within their engineering enterprise. So really think about resiliency, as you were saying, holistically and not just like I can fail over to another site. Truly thinking about it holistically.

>> Yeah, I think data might be a good angle. So I have to ask, you mentioned Code Assistant. I'm curious, Code Assistant has become a great asset opportunity to bring in more auto coding to create more time to do other things. Mainframe has a lot of legacy systems that have been running for years. Is that something that Code Assistant helps with? Is there tools there as companies look for talent or might not be able to find that COBOL programmer or CICS developer?

Tina Tarquinio

>> Yeah, so the Watsonx Code Assistant for Z was actually designed in mind for this. So a lot of this COBOL, the people who have written it are gone. And so they have this source code that's maybe not commented or not documented or it's hard to follow the breadcrumbs. And so the WatsonX Code Assistant is really great. It can go in and it can document your code. It can help everybody find the flow of the code. So that's really what our clients are finding. The biggest benefit is helping them get well commented, highly performing... COBOL is a highly performant language. So you don't just want to chuck it, but that's what it was designed for. And so we have a really great set of clients that are just getting started. An Australian client now is seeing about 20% productivity gains just in its short proof of concept phase.

>> All right, so I have to ask, what's next for Z? I know you've got a big launch. Can you share any details? I don't know what's going on. I know there's some big action happening this spring. What's new? What's next?

Tina Tarquinio

>> So in August, we previewed our technology. So we have a new Telum II processor. And we actually previewed an AI enterprise grade accelerator. So the Spire Accelerator will be a PCIe attached AI purpose-built accelerator, like I mentioned. And both of those will be featured in the next generation Z and LinuxONE that we'll announce the first half of this year. So stay tuned.

>> Well a lot of people here runs on IBM mainframes. Congratulations. Looking forward to a big year of IBM Z mainframe. It's back in vogue. Large-scale systems, big iron.

Tina Tarquinio

>> Yeah, we're back.

>> I want that word to come back. Dave's like, "I hate that word." Big iron, it's the big systems. It's fun. Everyone's buying large-scale distributed computing. It's back.

Tina Tarquinio

>> We're excited.

>> It never left.

Tina Tarquinio

>> It never left. But we're super excited for the next generation. Thank you.

>> Thank you so much for coming on theCUBE. Really appreciate it.

Tina Tarquinio

>> Great.

>> Okay. I'm John Furrier, host of theCUBE. Here at our NYSE CUBE e-studio, we have Palo Alto connecting Wall Street here. Stay tuned for more coverage after this break.