International Business Machines (IBM) Management Presents at UBS 2016 Global Technology Conference (Transcript)

| About: International Business (IBM)

International Business Machines Corp. (NYSE:IBM)

UBS 2016 Global Technology Conference

November 15, 2016 03:45 PM ET

Executives

Tom Rosamilia - SVP, IBM Systems

Analysts

Steve Milunovich - UBS

Steve Milunovich

Good afternoon. We have second key-note now with IBM. And we’re very excited to have with us Tom Rosamilia, who is the Senior Vice President of IBM Systems. Tom thanks so much for being here.

Tom Rosamilia

Thank you, Steve.

Steve Milunovich

Perhaps, you can just remind us what your responsibilities are. And then I know you’ve been with the Company a long time, maybe construct by describing some of the changes that you’ve seen in the systems’ marketplace.

Tom Rosamilia

Very good. So as you heard, I’m Tom Rosamilia. I’m please to be with you today. I’m the Senior Vice President of IBM Systems. And when you said a long time it’s 33.5 years. So it truly is a long-time and I’ve seen a lot. Systems, for those of you who don’t know, is a combination of servers and storage, some responsible for our power business, mainframe business, and our storage business, hardware business software.

Transformation, boy I’ll tell you this is an industry that’s gone through a number of transformations. And as a Company that’s 105 years old, I am not quite that old. But we’ve gone through a number of these transformations and I’ve seen them first-hand. I was a part of the bipolar to CMOS transformation at the time, and we went through transitional of client server. And seeing what’s happening in the marketplace today with the transformation into cognitive, build on cloud in industry context that’s profound.

And I would say that what’s most profound to me and the impact is I’ve got to make sure I am designing systems for that cognitive cloud era. They’re not the same as systems that have been built for prior eras like what we continue to do with transaction processing. So I’ve got to build design for cognitive, design for machine learning, design for artificially intelligence, design for systemic performance not just what I get from the processor. So I think the biggest change I am seeing and in addition to the speed is the fact that the infrastructure has to deliver for the workload.

Steve Milunovich

So what are some of the biggest issues that you see customers graphically work. I know you spend a lot of time with customers. And maybe you can lead into that, the discussion of the cloud and then how much you’ve seen in terms of private versus public cloud?

Tom Rosamilia

Good question. And listing a combination, and I think what we’ve seen from the industry data is that, and from UBS data, in fact from your Evidence Lab survey is that people are rushing to do things with hybrid clouds. Now, we have to define what that means, because I think everybody use it as a term use when they use it loosely. But we’re doing combinations. We have to make sure we’re a preeminent and first-class deliverer of public cloud, and we do that with the IBM cloud known as SoftLayer.

We have to do that from a platform-as-a-service layer, and the Bluemix services that are used lots of developers, more than million developers around the world, we’re adding many tens of thousands per week developers onto Bluemix in the environment, making lots and services available to be called. And so people can do think that hybrid fashion, so they can invoke a sentiment analysis at the time and incorporate that into their analytics. So whatever else they’re doing on-prem, so they’re using a hybrid technology.

So grate and you have to be first-class and public. You got to manage a lot of people’s cloud environments for them and first-class and hybrid to do that integration. Now for me in the systems business, it means that I’ve got to make sure that I am delivering hardware and software, management software, that allows me to go into the hybrid scale data centers that allows me to provide for shared HPC or high performance compute environment that allows me to sell into enterprise as more traditional as we’ve done for years and years.

And so I need to make sure that I’m continuing to deliver value to my high value Linux customers, while I move into the Linux environment on the POWER platform, as an example. I need to make sure I continue to deliver the most robust service in the world of hardware and software and the mainframe, while I make sure that I’m embracing things like blockchain on the cloud on the mainframe and that Linux want environment. So it’s a combination of all of those things, and that’s why it’s a hybrid world for me.

Steve Milunovich

So when people hear IBM Systems, probably the first thing they think is the mainframe. And Tony mentioned the bipolar, I remember the analyst announcement when you went from bipolar to CMOS, which only people at the time what a tremendous architecture change and how many years that probably took. But where is the mainframe going at this point? What that’s we’re already see going into future, both from a business usage as well as technology standpoint?

Tom Rosamilia

So among the many transformations and the reinvigorations, I would describe them as, one of the things we did and I made the decision back in 2001 to turn Linux on for the mainframe. And so, we made Linux available in addition to z/OS, or what you might know as MBS at the time to the data operating system that was running on the mainframe. And by broadening our appeal back to starting 15 years ago to Linux we were injecting new workload capabilities into the mainframe. We were so much better at running analytics on the mainframe having Linux as an environment to support. It was much easier to do reporting.

We’ve done other things like making job available on the mainframe; so all the new languages are available, so reinvigorating that enterprise. And now blockchain, blockchain to me is a great technology to embrace, because it in of itself is transformational. And so when the research team came to me a little over year ago and said we’re embracing this thing called Hyperledger, and we’d like to work with you on the mainframe environment, I was anxious to do so. And we sit-up in the cloud on the mainframe on LinuxONE back in the summer time a offering for blockchain in cloud. And we’ve got over 40 claims now using it in proof-of-concept mode, transitioning into production mode now to begin to use what they see as a secure container for Hyperledger.

Now, what that means is they’re running on a partition of this LinuxONE environment and they’re running fully encrypted with secure boot technology that doesn’t allow somebody to get in the way of when the systems coming up. So it’s secured against outside, secure against inside threat. And it’s a great secure environment to run blockchain. So blockchain itself is transformational. But we’re providing -- I’m providing a great environment for clients to test and use Hyperledger on LinuxONE.

So in each one of these steps, whether it’s Linux, whether it’s blockchain, whether it’s Java, last year or year-end half ago when we delivered, almost two years ago, when we delivered the Z30, it was all about analytics collocated with transaction processing, so being able to do real-time storage checks while you do other transaction. The ability to do mobile transactions, $30 billion a day, so new for the mobile environment; so we’re continually injecting and reinvigorating what we’re doing in the mainframe environment.

Steve Milunovich

You mentioned blockchain, is that just for financial institutions at this point, or do you see it broadly now over-time?

Tom Rosamilia

I see it dramatically broadening. But specific use-case I’ll give you is one around supply chain. And I see a lot of the logistics in the supply chain capabilities being transformed by blockchain. Anything that takes a long time to do settlements today and takes multiple players and where I can benefit from having this audit trail and being able to do faster hand-offs, I think is a great example of what we can do with blockchain.

I’ll give you a couple of examples. We’re working with the Wal-Mart, they happen to be using LinuxONE on the mainframe for their food supply in China. They want to know where this product started from and where it ended up. If they get a problem in their supply chain somewhere, they want to be able to trace it back to exactly what batch that came from. And if there is a problem with that, I only have to take that batch out of production or that batch off the market. I don’t have to take all the spillage off the shelves because I’ve got one place where it contaminated.

So by being able to check the providence of that food supply, they can ensure a much safer food supply in China. But the same thing is true in other companies, every ledger a company we’re working with who does diamonds, they want to know exactly what hand-offs occur along the way. And they want to know that the same diamond was delivered for me to you, you to somebody else, and then to the end user. And that each step along the way has to check of carrot, color, clarity, cut. It’s the same diamond. And they also want to know where it started from. Did it start from a conflict mine? I don’t want diamond starting from conflict mines in Africa. I want to know where it started. And then you get into financial services.

So, to me, it’s shipping ports in Singapore and around the world where you’re waiting for things to clear import-exports. You’re waiting for money to change hands. You’re waiting for somebody to verify they can take the containers and deliver. All of those things can be expedited, and we can take that inefficiency out of the system with blockchain.

Steve Milunovich

Other major platform of course is POWER, and it's seen its ups and downs. So, can you just talk about where we are today in terms of both of the POWER processor and the POWER system?

Tom Rosamilia

So today we deliver POWER8 processor. But I think the key transformation around POWER is about not just doing it myself but doing it with an ecosystem or partners. So couple of years ago, we introduced the OpenPOWER Foundation. We’re up to over 270 members. We just had Super Micro join this week. What this is really about is embracing this concept of systematic computing, that’s because I want a great micro processor, but I also want these accelerators, like GPUs from companies like Nvidia. I want great FPGAs like those that come from Xilinx, both numbers in the OpenPOWER Foundation. And then I want this outstanding bus architecture between the two, so that I’m not waiting for data to come-in. And the more you’re waiting the slower you’re going.

So we announced a couple of weeks ago something called OpenCAPI, which allows us to have this very high-speed bus between processors and GPUs, or processors and FPGAs. And it was endorsed not only by the OpenPOWER Foundation members but the addition of AMD, HPE, Dell EMC, all got on-board on this OpenCAPI high-speed bus technology to be delivered next year, but available in CAPI form today in the POWER8 processor.

We also have a big presence at Supercomputer 16 in Utah this week in Salt Lake City. Where we announced an incredible partnership we have with Nvidia about making something called POWER AI available, Artificial Intelligent. So the three premier machine learning libraries available optimized for POWER8 plus Nvidia Tesla GPUs, and using their NVLink technology to bridge the two, making available Caffe, Torch, and TensorFlow on the power platform supported by IBM.

We made an announcement last week with Tencent in China, one other hyper-scale data centers, that they set a record in a sort benchmark that had never been seen before. They were able to do a terabyte in a second on a POWER8 processor, and they did it spy back what has been done last year; so huge performance benefits versus competition. They won the benchmark. They did it with POWER8. We’ve got Nimbix servers available with HPC environments so people can run simulations in a cloud environment on the Nimbix servers, running POWER8 and Nvidia.

So lot of examples where we’re embracing hyperscales and embracing high performance computing, clearly embracing a Linux marketplace, which for us was about 15% of revenue in the third quarter, but I expect that to continue to grow as we continue to embrace our high-value UNIX customers, but that’s a declining marketplace.

Steve Milunovich

It’s been a long time since we’ve had that CISC versus RICS war. But maybe just for folks if you could explain one of the architectural advantages of RISC, why would a Tencent or Google, which I think is using POWER have an interest in doing that versus Intel?

Tom Rosamilia

So, they’re not going to be all that interested in the actual instructions set, which is the CISC versus RISC discussion. They’re going to be interested in the outcome. And what they’re seeing as a combination of share performance plus our willingness that have in open environment where they can do plug and play of some number of processors, plus some number of GPUs, plus a very high speed bus, it’s really the system architecture that they’re embracing. And yes the processor running Linux on POWER is a terrific one for them to run with high speed and the highest speed threads in the marketplace, the most threads in the marketplace.

So they are getting the advantages of that. But I think they’re getting even more advantage from the OpenPOWER consortium that’s delivering multiple pieces that they can plug together depending on what they want to run. In the Tencent case, they did it without GPUs. In Google’s case, they’re going to want to do it with some, with their own, called TPUs. And others, people will use FPGA. So it’s the plug and play of OpenPOWER that I think gives us the real advantage and the high speed connection between the two.

We’ll deliver a POWER9, we’re working on it and we’ll deliver it commensurate at the same time as the ecosystem delivers their products, which is a little bit different from what we did in POWER8 where I rolled out POWER8 and then the other partners were a little bit further behind that was a brand new idea. But now they’ve had time to bake that, they understand what they can do. And so when I deliver, they can deliver at the same time.

Steve Milunovich

And so we’ve got these areas, the POWER system, and third leg of the stool is storage. Storage has been a difficult market for everyone. IBM, I think was pretty early in flash and protections memory acquisition. Could you kind of bring us up to speed in terms of what you’re doing?

Tom Rosamilia

So storage is absolutely in transition and transformation, and the two key things I think in the storage, probably three. Three key things in the storage marketplace around flash, as we mentioned, we did Texas Memory Systems acquisition, which was the premier flash provider. And we’ve incorporated that into our storage system. And I will tell you from a flash perspective, we were here together three years ago just off the road in Sausalito, and we were talking about convincing people that flash would replace spinning disk. I don’t have that discussion anymore, it’s very clear that that’s happening. So I don’t have to have that price performance discussion.

Now, I mostly have the discussion about why I am better than the other guys. But I don’t have to talk them into the transition into flash. We’ve moved flash across our entire product line. You can buy all flash arrays for apron systems, for DSAK model for flash. You can buy store-wise, at models, all-flash. You can buy DeepFlash 150 for scale-out environments for each PC computing. So we put flash in every step along the way in our storage portfolio from the high-end to the low-end. And so that is clearly a transformation that’s happening in the industry and for me.

The second one drilling around software defined storage, where we have number one market share about, according to IDC, 31.5% market share. And that’s where the special sauce is really moving to software. In fact even in Texas Memory, it was about the controller not about the actual parts in that new software. So it’s all about the software, not just the virtualization but the tearing, the de-duplication, the compression we can do. All of this in making different storage capabilities available to seamlessly to systems so that they don’t have to care about where it is at any point in time.

Things that automatically optimize where it would be depending on the performance characteristics even cognitive systems that learn from behaviors of that system and where I should put that data. So software defined becomes very large. And that’s double-digit growing revenue in the industry, and one where we’re number one.

So I think this -- and the third one I should say is around object store, which is a part of software defined. But we went out last year and bought the number one object store provider in the world, we have Cleversafe. We’ve incorporated it into my portfolio, so for the on-prem deployments, we do them on-prem out of my sales team. And we’ve also recently, just a month ago, made it available on SoftLayer on the IBM cloud, and so it’s available on-prem or as a service. So three big trends in storage are around flash, around software defined, and a part of that is around object store.

Steve Milunovich

Do you know roughly what percentage of your capacity shipments are flash today?

Tom Rosamilia

I don’t and it’s because some of it, and in fact even the way the industry counts is, if I do a hybrid array where some of it in spinning disk and some flash, that doesn’t count as flash. It’s only if it’s in all-flash model that it actually counts. So, where you mentioned Texas Memory Systems, which is only flash, but it’s also parts of each of the devices we delivered and in some cases, it’s a 100% of the devices we delivered. So, it really varies across all the different segments. But it’s happening. The transformation is happening.

Steve Milunovich

And the storage software of your sales is not actually counted in storage revenue as investor see, is it elsewhere?

Tom Rosamilia

It’s in the cognitive segment. So, yes, the software that we sell that takes advantage of that storage and then that software defined piece gets reported in the cognitive segment.

Steve Milunovich

You managed your claims and some credit for that?

Tom Rosamilia

I do, absolutely. It’s an important part of our portfolio. So, I’m happy to pay a role in the strategic imperatives within IBM.

Steve Milunovich

So you talked about hybrid cloud earlier. There is obviously a lot of competition for hybrid cloud. I think all the vendors kind of agree that what users are shooting for. So you’ve got HP, which is now very focused on that with some of the divestitures, just heard from VMware, Dell EMC obviously are very focused on as well, Cisco is a player. How do you think about how IBM is competitively positioned?

Tom Rosamilia

In for hybrid?

Steve Milunovich

Yes.

Tom Rosamilia

So I think, first of all, it’s interesting on playing right here. I was reading some of the research material from UBS and from Evidence Labs. And there is difference of opinion on what is hybrid, is it real. And I think you guys are spot on in calling this out that it really needs to be defined for people. So they understand what do we mean by hybrid environment. First thing I’ll say is we plan and intend to be a first-class provider in public, in public cloud. We do with the IBM cloud offering. And that is really important for people to understand. We’re not just giving that to others, we’re embracing it. We’re going after it. And making our platform-as-a-service layer available on top of that, it is essential for our strategy going forward.

We also do a lot of cloud managed services for customers today, whether it’s on-prem or whether we do in our data centers. And my services colleagues banish that. And then there is this world in between of hybrid. And as I said, really need to define that. Ask vendors to define what that means. What some people think it means is the ability to use capacity on-prem and eventually use it off-prem when I need it.

And that’s a very difficult thing for workloads to do to burst out, because of physics, because of gravity of data, and because have to have an exactly mirrored environment, somewhere else like I have on-prem in order to make that work. But as VMware folks have said and they’ve partnered with us on this. Portability is a different topic if I can run this workload here, and then I can select when I want to run the workload in a cloud. And so the VMware software partnership is critical to us to be able to duplicate all the VMs that they have, millions of VMs running on-prem today and make that available on the IBM cloud, and so they can run it either place when I choose but not in a nano-second when I have capacity need.

So I think portability and making sure I have the same kind of environments available in both places is critical. I would also say what we could do with cloud object store to be able to store locally, or chose to store that data in a IBM cloud, running our cloud object store. So I can store it in either place. I am not talking about storing in both. I am saying this data goes here and that data goes there. But it's exactly the same interface for me to use one place to the other. That gives me the hybrid capability. I can do back-up and recovery. And say, I want to store my data in case I need it somewhere in a public cloud, or a dedicated cloud environment.

And then this to me a really interesting case, which is one we would have called integration before. Let's take an example of an on-prime application that’s running and it wants to do sentiment analysis. And it wants to use a Watson service and Bluemix in order to do that. Well, I could augment in existing application with that cognitive computing running on Watson, running on IBM to cloud and say let me find out what Twitter, what social media is saying about my product that day as I decide what to make available or how to price something.

Now I’ve done a hybrid environment where some of it's running on-prem and I am calling a set of services that could be in a public cloud somewhere looking at public data somewhere. So to me it's about the combination of all of these things. And therefore I think that’s why a lot of clients say, while I want embrace things in the public cloud, and they should, they also have to reflect that it also have a tremendous investment in what they have already done, and they want to be able to augment that.

Steve Milunovich

So, many of your are like competitors [multiple speakers] they initially, I think, were trying to compute AWS, and they’ve one-by-one pretty much given up on that. But you made the investment in SoftLayer early-on. Is that turning out to be a real advantage at least to be able to offer that public cloud capability to customers?

Tom Rosamilia

I believe, as I said, that it is really important for us to have a first-class public cloud offering. And not just say, I am going to give that over to AWS or I am going to give that over to Microsoft. I think it's essential for us to be a first-class public cloud provider. And then we can have the discussion about how we can help you integrate what you are doing today on-prem with that public cloud capability.

Steve Milunovich

Let's turn to Watson a bit. By the way, great ads. I think you could watch the U.S. open tennis without seeing the Watson ads. I think investors are extremely intrigued by what you’re doing clearly, you guys have been very vocal about $2 trillion opportunity potentially. But investors also have trouble getting their arms around exactly what is Watson. How do you monetize this? Can you talk us through the position in the platform today?

Tom Rosamilia

Yes, it's understandable that people having trouble with it, because we started-off having this monolithic thing is called Watson. It played a game-show and it did well. But going forward it wasn’t going to be about duplicating this specific use-case around Q&A or around call center or around oncology, or around whatever it might be. We had to take that code-base and re-factor into set of services. So these are cognitive services that are available in the Bluemix environment. And so people can make that sentiment analysis call. They can make that voice recognition call. They can make whatever call they want to make and learn something about using that service and learning some information back.

So it shows up in a lot of services. It's not one thing. It’s a set of things and set of capabilities that now we can apply that to a call center example, or an oncology example, or a financial services example. In the oncology example, for example, we’ve partnered with companies like firms like MSK, Memorial Sloan Kettering, or Quest Diagnostics, or the Broad Institute. We’ve made the oncological services available from Watson to 70% -- or the oncologist treating 70% of the patients in the U.S. that happened to have cancer. So they’ve got this access to this incredible database of information stored from images, in some cases for companies we’ve acquired that learned about what’s benign and what’s not, and how can I reflect that back to the oncologist. Or learning about genomics, so that I can match specifically whatever cancer this patient has, or what treatment I should use. And so that’s the set of services.

The translation of all that is its cognitive everywhere, it’s not about a product called cognitive. Cognitive is an aspect. Cognitive is like resiliency, or like high availability, it's can I do that in a cognitive way or am I doing that in a traditional way. Am I doing that in a learning way or am I doing that in a programming way. And so, that’s why it’s difficult for people to get their arms around well how big is Watson. Because my boss Ginni Rometty said two years ago, you’re going to see cognitive everywhere. And I think he was right. We’re seeing it all over the place.

I design for cognitive, because I want to design for the AI world where we’re doing a lot of machine learning. It turns out accelerated computing is exactly the right model for cognitive computing. I’ve got to be able to provide that to IBM and to others who are doing cognitive computing. So that I can be that arm supplier to the world that’s embraced in cognitive, whether they do it from us or they do it from others.

Steve Milunovich

How you monetizing that, I think, at the Analyst Day, there were three, four, five different ways you were looking at monetizing. And I’m sure you kind of figuring it out as you go up bit. And where do investors see that in terms of where IBM reports revenue?

Tom Rosamilia

We mostly see it in the cognitive solutions segment. So that’s obviously where it would be given this is cognitive of computing. We’ve gone into new industries. It’s about Watson Healthcare, where we did a number of acquisitions but we’re combining that with the force of Watson. We do it and Watson IoT, where we can learn things from all the connected devices, or we can learn things from information, we can learn about weather, and that weather has a profound effect on lots of businesses.

And so the Internet of Things using a weather channel transport is a very effective way of tying all that data that’s available. And in this case from the weather sensors, but in other case from all kinds of sensors around the world. And so you’re seeing it in -- it will be in financials, it will be in call center applications that we’re doing with many of our partners, and our customers, it’s in healthcare. So it’s in all of those difference places. But I think, you as investors, you should look for in the cognitive segment and see how that’s growing.

Steve Milunovich

We recently did Evidence Lab survey on AI talking to both users and ISVs, and we’re surprised to find 90% of the companies say they’re doing something in the area. IBM came out very well. And I guess one of conclusions was, in terms of what’s going to differentiate folks only the data probably matters a lot. And I think that’s something where you’ve kind of apart from history in terms of making acquisitions in recent years to own the data?

Tom Rosamilia

I think that’s true. But I think it’s also a way we differentiate ourselves from competition. So, there are certain domains where it really helps us to learn and inform a Watson Analytics capability around oncology. So, we make acquisitions like we’ve done with Merge and others where we bought lots of either records for images, and we can learn from those. We’ve trained Watson to recognize what’s the benign and what carcinogenic. And so if it’s cancer or not cancer, I want to know with 99% assurance and we can train it to know that owning those images in that case makes a lot of sense.

So as we went into the healthcare industry, and going to transform the way people do health, provide healthcare, I think it was really important for us to have a chunk of that data that we can learn from. In other places, that’s not true. We do a lot of Watson engagements with people where your data is your data, you can keep that. The competition for IBM doesn’t always do that.

But I think a difference that we make is that your data, if you want to keep that private that’s fine. If you want to combine that with publicly available data, if you want to look at social media, you want to use Watson to say tell me what Twitter is saying about me, my brand today, you can do that and combine it with your own data. But keep that to yourself. So, yes, for healthcare it was important for us to learn from medical records, from images. But in other industries, not so much.

Steve Milunovich

We’ve also hearing a lot about multisided platforms, and how many new tech companies are formed these platforms and the benefits you potentially get in terms of network effects. It appears that you’re trying to create Watson as a platform, opening up APIs, allowing third parties to add value. Could you talk about that?

Tom Rosamilia

I think we’ve learned this lesson as a valuable lesson for the industry. I would say that our mainframe as a platform that we built upon. I would say that what we’ve done around OpenPOWER as a platform to say there is a set APIs that people can use and a set of standard interfaces so that they know with assurance that they can combine forces and use OpenCAPI as hopefully a platform that we’re delivering together.

And clearly, Watson is one where we believe that we’re not going to be the only ones contributing for this. We want to marketplace where others can write things and use maybe recombine those APIs into different use cases, and take advantage of the ones that are already there. So, it’s a big world. And I think if you can win at the platform level then it's a win for everybody. It’s a win for us, but it's also a win for the ecosystem of partners. And therefore a lot more people delivering value than just IBM.

Steve Milunovich

I’ve got a few questions from investors here. One is about, what is IBM doing in China? And I know that you had a lot of success in your area in China, and mainframes are the backbone of a lot of China transaction processing systems in airlines. But obviously given the political changes here, people also interested.

Tom Rosamilia

Yes. So there is a lot happening in China. I’ve been to China five-times this year, I'll be back again for the sixth time this year. So I spent a fair amount of time there. And it's an exciting place. It's a very important marketplace to us. And it's going through its own transformation for sure. But I think it is essential that you noted that there are places within China that are critically depended on our mainframe technology and we’re happy to keep those folks as customers or high-volume transaction processing customers. And I hope they continue to be forever.

I think that as we look at what we’re doing with POWER, one of the reasons I called the OpenPOWER play was to be able to do indigenous development in China, not only in China but in China as well. And so I’ve done licensing models for companies in China that are building their own OpenPOWER systems. They’re delivering their own servers. Inspur is delivering a server today based POWER8 technology. But I’ve licensed the capability for them to produce their own chips and manufacture their own chips. So as they get better at semiconductor development, they can use power as reference architecture or with a license from me, sort of the arm model, in this case for servers, so that they can do indigenous development. So I go over visit with them and see how they are crossing along on their roadmap to delivering POWER8 and eventually further follow-on products and technologies.

And so we’re embracing a lot of capabilities in China with the different government agencies, with the different commercial entities, in addition to the customers that I’ve traditionally called on along the way. But I think this win that we just had with Tencent is another breakthrough, because Tencent being one of the hyperscale data centers, them using POWER8 to win the sort benchmark, 5x performance of what was been available prior. They’re one of the three big hyperscale providers in China. So wining with them and hopefully winning with some of the others is another good way for us to be a part of what's happening in China.

My team in China is designed with China not just manufacturer in China. But I am doing that as a co-development effort with Chinese firms that have lots of good skills and they can bring their capital intellectual one other way to the table. I can do the same and we can build for the China market together.

Steve Milunovich

Another question here, how will IBM Systems address the parallel computing opportunity for high-performance computing?

Tom Rosamilia

I think that the high-performance model is significantly changed. We used to be, several years ago, a big player in high-performance computing. And I did it with very unique and custom hardware that was only available for high-performance computing models. That turned out to be non-economically a good idea. Even though there were some great wins and some really good examples of systems. It wasn’t a good economic model for me. And so I stopped doing that for a while. And OpenPOWER has really opened up this model to say, it’s not just about the processor, it’s about the storage class memory, what I can do with accelerators, what I can do with flash, and delivering on this promise of the OpenPOWER and OpenCAPI ecosystem.

And so, the wins that we’ve had with the big U.S. government wins, the wins in France, the wins in the UK at Hatchery, have all been wins at University of Michigan, have all been embracing this OpenPOWER model, it's systemic performance delivered by CPU and GPU, or CPU and FPGA. And so, what we’ve announced with POWER AI, in VDS, saying that their compilers are available for POWER8 and GPU, the wins we’ve had, the fact that it’s available on the HPC cloud and Nimbix. All of these I think are good proof points that this model of accelerated computing is the answer for HPC going forward.

Steve Milunovich

How does your new hybrid cloud off flash storage solution differ from your competitors?

Tom Rosamilia

So our off flash storage solution, the biggest difference that we can provide is that as you’ve sold this thing up, it continues to perform. So unlike spinning disc where tend to undersell, they sell it to like 30% utilization. At a certain point in utilization on a spinning disc, you’ve used to begin to fall over and that is that you created I/O bottleneck. And so everybody is waiting for data. And so, people spend a fair amount of money spreading their data out on spinning disc. In flash, you can go a little further. And so I can go to 40%-50% utilization and not create a bottleneck, because it’s a solid state and I can just -- I don’t have to move heads and drives and the like.

And our environment, we can go up to 80%, almost 90%, utilization and not have a performance set. So, in our case, it's about filling this thing up, about providing compression, so that you use less of the data. If you think about how much disc people buy for how much production data they actually have, it’s a multiple of it. And so for us the better we can get to one-to-one, the better off customers are, because it’s less money for them and less floor space for them and less energy for them. And so our solution in the flash area delivers higher utilization, greater capacity for the smallest footprint possible, and the best performance in the marketplace.

Steve Milunovich

I guess selling hard drive business years ago was a pretty good idea?

Tom Rosamilia

It was a good idea. Although, we continue to package those hard drives into storage devices that we continue to sell today, and we will for years and years to come. But being in that component business, you’ve got to really be able to do at scale, same thing around the moves to get out of the semiconductor manufacturing business. I was ways up scale in that.

Steve Milunovich

So the industry is obviously going through a significant change to cloud, and sometimes we say, we haven’t seen anything like this since the shift to client server in the 90s, which almost put IBM out of the business at the time. But IBM did recover. How would you compare the challenges that IBM is facing today to the 90s? When I think you were around, Ginni was around, there has to in fact the thinking of the management team here as you’ve seen other one of these disruptive periods.

Tom Rosamilia

So I would actually recommend a near-death experience to anyone in the room. I wouldn’t recommend a death experience. But near-death is a good learning experience. And we certainly went through that. And I think it helps motivate us through transformation and saying, I don’t want to get to the point where I am struggling to keep my head above water. The thing that’s so different for me this time is this move cognitive is one that we’ve initiated, not the one that we’ve reacted to. And so, to me, we’re on the forefront of this. We have the leading capability in cognitive computing. A lot of it invented in IBM research. Some of it augmented by things we’ve done, but most of it really home-grown for our R&D dollar.

What I am delivering in systems for cognitive computing is all things that we’ve either done or we’ve partnered with others to do. So, I feel good about it being on the leading edge of this, not being on the bleeding edge of this and thinking about what's going to happen if we don’t embrace something. We’re driving this. We’re not reacting to this. So that’s a fundamentals difference.

The other difference I would think is the speed. This is all happening very quickly. And so possibility is aligned like never before, things that were not possible before are possible now. And I would say the last difference to me is around business model. So I think business models are changing now more than they ever have before. And it's not just about technology, it's about the way people go to market with that technology. The way people tie supply and demand.

And the fact that we’re doing it this case back in the '90s, we were doing it pretty much on our own. In this case we’re really doing it through partnerships. Whether its partnerships apple, or the fact that I am doing things at SAP or on POWER. I’ve got a partnership with Cisco. You mentioned them as a competitor. They’re also a great partner to me. I sell storage with their servers. We’ve done things around collaboration together. We’ve done things around IoT together. So to me this is more about an industry rather than just a company reacting to the opportunities that’s out there, and that’s very different than it was for me in the 90s.

Steve Milunovich

That’s fair. I think IBM is always been as its best when it’s a thought leader, and not reacting. And a number of years ago, Ginni, at Analyst Day, talked about the cognitive computing wave, but it’s a 50-year wave. So what do you tell investors in terms of the timing of when you will see the benefits of the investments that you’ve been making the last couple of years?

Tom Rosamilia

I think it's happening, and I say that because it's cognitive everywhere as recording my boss, she said cognitive everywhere. And you see it happening. I see it happening. Even in the operating system function I deliver, they become smarter every year. Every delivery we have we can learn things about what's normal for customer A versus customer B by watching and observing the behavior and doing better as a result. We call that B aware, as an example. So it’s a ware, it has awareness of am I operating within normal parameters or outside, is there something really bad about to happen.

So it shown up in the operating system, it shows up in our storage. It shows up in all of the different Watson set of services that are available. So, the comment about this transformation and how long it may take, that’s probably an answer to complete transformation, but I see it happen every day in what we’re doing with many companies around oncology, around call centers, around financial markets, it's happening every day.

Steve Milunovich

And the question is then, all the future talent acquisition or do you believe it will impact the bottom line?

Tom Rosamilia

In this case, I think, when you do things that coming on the bottom line is we always do acquisitions with the business case and the business case is usually around what we think has happened to the greater good of IBM. And in cases like where we’re doing labor, it's usually an acceleration of how to get talent. So, labor-based acquisitions are really for us about an injection of talent.

Steve Milunovich

And then maybe to finish up, you’ve divested your fabs but I believe you’ve test a lot of your semiconductor designers. And I remember John Kelly at one of the analyst days a few years ago talking about substantial changes in hardware, cloud and computing. One sort of wonders if the value ever shifts back to hardware in a way, so why did you keep those designers and what sort of there is the future of Systems business?

Tom Rosamilia

Good question. And I think for foreseeable we’re talking about what I am able to do with processors that I continue to innovate in more traditional ways in silicon and the combination of accelerators. And that’s true for the next four years. But what we’re beginning to see deliver is things like the next generation. And by that I mean the generation beyond silicon. So, it’s about quantum computing. We sit-up and research a quantum computer last year, made it available on the cloud this year for experimentation.

So people can get on to our cloud and take advantage of quantum computing and see what it can do. We have this awesome fundamental quantum issue. This has been promised for years. But the breakthrough is really where last year and this year are being able to correct for some of the errors that occur. And make this available at scale. And so quantum computing at scale, we can now see it, we can touch it, we can feel it. It’s not commercially available, but it’s available for people to do experiments with.

If I think about silicon photonics, and the comment I made about OpenCAPI is the same thing for silicon photonics. I look for the place where I am limited on the system and eventually it will be having to move information around. If I can use silicon photonics and we’re on the leading edge of doing design and our research team to do that, to me it’s about taking that bottleneck out of the system. If I think about what we’re doing with carbon nanotubes and the little layering of carbon, we can do this for orders of magnitude faster, and using a whole lot less energy than the systems today use. So between neuromorphic brain inspired systems, like TrueNorth that John has described, or quantum computing silicon photonics, what we’re doing the carbon nanotubes, all of this is about leading edge research. Some of its driving near deliveries like what we’re doing with POWER9 and beyond, and some of it will be about the next wave of computing.

I am glad you noted that while I divested of the manufacturing business, I’m still doing the design. I’m doing with the partners, so they have to stab what I designed. I have to do within their parameters. And then I package it up into our servers and I deliver it in mainframes and in power systems for our clients of today, and hopefully for Linux hyperscale computing centers and HPC environments tomorrow.

Steve Milunovich

Great. Well, Tom, thank you so much for being with us and providing flash lights and idea.

Tom Rosamilia

Yes. Thank you, Steve. I appreciate it.

Steve Milunovich

Thank you.

Question-and-Answer Session

End of Q&A

Copyright policy: All transcripts on this site are the copyright of Seeking Alpha. However, we view them as an important resource for bloggers and journalists, and are excited to contribute to the democratization of financial information on the Internet. (Until now investors have had to pay thousands of dollars in subscription fees for transcripts.) So our reproduction policy is as follows: You may quote up to 400 words of any transcript on the condition that you attribute the transcript to Seeking Alpha and either link to the original transcript or to www.SeekingAlpha.com. All other use is prohibited.

THE INFORMATION CONTAINED HERE IS A TEXTUAL REPRESENTATION OF THE APPLICABLE COMPANY'S CONFERENCE CALL, CONFERENCE PRESENTATION OR OTHER AUDIO PRESENTATION, AND WHILE EFFORTS ARE MADE TO PROVIDE AN ACCURATE TRANSCRIPTION, THERE MAY BE MATERIAL ERRORS, OMISSIONS, OR INACCURACIES IN THE REPORTING OF THE SUBSTANCE OF THE AUDIO PRESENTATIONS. IN NO WAY DOES SEEKING ALPHA ASSUME ANY RESPONSIBILITY FOR ANY INVESTMENT OR OTHER DECISIONS MADE BASED UPON THE INFORMATION PROVIDED ON THIS WEB SITE OR IN ANY TRANSCRIPT. USERS ARE ADVISED TO REVIEW THE APPLICABLE COMPANY'S AUDIO PRESENTATION ITSELF AND THE APPLICABLE COMPANY'S SEC FILINGS BEFORE MAKING ANY INVESTMENT OR OTHER DECISIONS.

If you have any additional questions about our online transcripts, please contact us at: transcripts@seekingalpha.com. Thank you!