IBM CEO Arvind Krishna: ‘Win-Win-Win’ When Partners, Customers, Big Blue Work Together

‘[Partners] all make a lot of money by helping our clients deploy these technologies—and by these technologies, it is Red Hat, it is automation, it is AI, it is cybersecurity, it is even mainframe and power and storage in some cases,’ IBM CEO Arvind Krishna tells CRN ahead of the company’s Think conference.

Ahead of IBM’s annual Think conference this week, IBM Chairman and CEO Arvind Krishna took questions from a select group of journalists on his company’s investments in artificial intelligence, hybrid cloud, quantum computing and its partner ecosystem.

In response to CRN’s questions about IBM’s partners, Krishna said that MSPs, resellers and other partners have multiple opportunities to make money with the Armonk, N.Y.-based tech giant.

“They all make a lot of money by helping our clients deploy these technologies—and by these technologies, it is Red Hat, it is automation, it is AI, it is cybersecurity, it is even mainframe and power and storage in some cases,” Krishna told CRN. “And going into clients and giving our clients value with our technologies is something they should all know we want them to do. And we are happy to work with them. We are happy to train them. We are happy to co-invest and co-market with them.”

Krishna continued: “I call it a ‘win-win-win.’ It should be a win for our partner. And there is a win for the client. And consequently there’s a win for IBM.”

IBM Think 2022 marked the return of an in-person element to the annual conference, held this year with a limited attendance in Boston. IBM also published conference content online as the world continues to grapple with the COVID-19 pandemic.

IBM subsidiary Red Hat also held its annual Summit in Boston at a separate venue from IBM.

Along with touting IBM’s opportunity for partners, Krishna said that partners shouldn’t worry about IBM growing its capabilities to service clients—such as with the acquisition of Microsoft Azure partner Neudesic—and its still-strong consulting arm even after the spin-off of managed infrastructure business Kyndryl.

While there are a few hundred large-scale clients IBM will service, that leaves hundreds of thousands of businesses for MSPs and resellers.

“We, best case. are going to have a market share of a few small single digits in this market,” Krishna told CRN. “So we should think that there’s a few hundred clients that IBM may go to. And even within those clients, we have a small footprint. We are aimed at those massive SAP implementations, at the large maybe Azure implementations.”

He continued: “But when we get to our MSP colleagues, that’s completely complementary. We don’t really do much in that space. When we think about most of the VARs and VADs, we don’t do anything in those spaces.”

Here is more of what Krishna had to say ahead of the Think conference.

On IBM’s AI Plans

So as an example, here on the floor [what] we’ll be showing in Boston is Watson Orders as a partnership at McDonald’s.

When you look at that, it’s all about how in an environment you have to match what a person orders against the menu—it’s not general question and answer.

You have to do that and you have to understand how they might modify an order. As an example, we were just this morning trying it out just as a pretend customer.

You order a Quarter Pounder, then you take onions off. You might add extra tomato. You order a drink. It asks you ‘medium or large or small?’ When you get to combo meals, it asks you all the things.

So this is, I think, a great example of enterprise AI. It has to align to McDonald’s back-end process off the menu, which changes weekly. Or even more often sometimes.

It has a constrained set of answers and has to understand the workflow of what the restaurant wants.

A second great example is AI being applied to information technology. How could you make somebody in IT eight to 10 times more productive? When we think about running applications and understanding what can happen and understanding that a problem may be showing some signs of coming on—so being predictive, not just reactive—is how you get an uptime increase, as well as much better productivity. And that’s the second example we have around a set of products that I’ll put under the category of Watson AIOps.

So these are examples of how we’re solving problems. People are willing to put them into production. People are willing to try them out. With the current demographics of labor and skill shortages, that becomes even more relevant, as opposed to—by the way, I do believe that that’s why … Watson is alive and well.

I do believe that some of the health-care examples will happen, but they might take a half-decade or a decade more to come to fruition just given how hard those problems are. And the implications are life and death. So that’s how I categorize what we should be doing now. As well as keep working on the others, but they’ll take a lot more time.

On Red Hat And Hybrid Cloud

Red Hat is absolutely essential and core to our strategy. Not just the hybrid cloud, I say to our strategy.

Red Hat gives us a platform on which we do two things. One, it is the base of the hybrid cloud because Red Hat runs on the public clouds and it runs on-premises and on private clouds.

Second, we have taken all the IBM software and optimized it for the Red Hat OpenShift platform. So now it becomes—we have written things once and we can deploy them anywhere that a client wants.

But more than that, I mean, when we think about developer productivity, and we think about security, and we think about how to keep an operating system and a platform secure, as opposed to a client having to download 60,000 pieces of open source and then worry about is it all patched, is it all correctit—it gives us a great entree into the whole world as people are looking for productivity and they’re looking for enterprise strength. So those three dimensions.

Iit is the hybrid cloud platform that goes across multiple public and private. It is the base on which all IBM software is delivered. And three, it is a basis on which our clients are doing their own digital transformations, given all the capabilities I talked about. … Red Hat is a company that helped bring—how do you take the innovation of 60,000 or 600,000 people, but put it into a way that the enterprise can digest through the products that they bring to market?

And I’ll mention just three—Red Hat Linux, Red Hat OpenShift and Red Hat Ansible. I think those are three great examples, but among many more, that Red Hat has used to bring great innovation and capability to enterprise clients.

Message To MSPs, VARs And Systems Integrators

Some of you who know me well know that I’m passionate about our ecosystem. And I’ll use the word ‘ecosystem’ because it shouldn’t be just ISVs. ISV is one category … And what I’d like them to take away from the themes of Think this week is, one, there is opportunity for them as they look upon our capabilities and how we go to different clients with these capabilities. Especially when I think about all of these.

They all make a lot of money by helping our clients deploy these technologies—and by these technologies, it is Red Hat, it is automation, it is AI, it is cybersecurity, it is even mainframe and power and storage in some cases.

And going into clients and giving our clients value with our technologies is something they should all know we want them to do. And we are happy to work with them. We are happy to train them. We are happy to co-invest and co-market with them.

So there is—I call it a ‘win-win-win.’ It should be a win for our partner. And there is a win for the client. And consequently there’s a win for IBM.

On Quantum

I’ll begin with the headline that we have now put out a road map that goes out to 2025 and beyond. And we have now stated we’ll put a 4,000-qubit quantum computer [out] in 2025.

A couple of years ago, we had put out a road map that said 1,000 qubits by 2023. We are on 127 qubits in 2021. We have said about 400 cubits this year, ‘22. And then up to the 1,000 qubits in 2023.

Now to get to 4,000, there’s quite a few problems we have to solve. How do you begin to scale these systems? How do you begin to communicate amongst them? How do you get the software to scale and work from cloud into these computers? Those are all the problems we believe we have line of sight to.

And so we have high confidence in our 2025 road map of 4,000 qubits. And that’s on the physical quantum computer. Not a simulator. Not the software.

In addition, we’re talking about also improvements to Qiskit, which is our open-source software, which is how, I’ll call it, ‘the user’ will interact with a quantum computer.

I find it hard to use the word ‘program’ because if we ever look at the screens, it’s not quite like programming. It’s somewhere between mathematics and composing music because you’re having all these qubits, you’re interconnecting them. And it’s a lot of fun to go play around those circuits and begin to see real problems getting solved.

I mean materials are getting solved. You can begin to think about lithium hydride. You say, ‘Why do I care about that?’ Well, if you care about EVs [electric vehicles] and car batteries, you care a lot about those.

We care about alloys that could be used to make lighter-weight—but strong—vehicles. How you can begin to worry about financial risk, something that dynamic markets today have brought back to the fore?

You can worry about optimization like in artificial intelligence and search. So a lot of excitement. And as we begin to get to 4,000 qubits, a lot of these problems become within reach of quantum computers.

On The Osprey Processor

So if I think about quantum computers, one is just a pure size—I’m going from 100 to 400 toward 1,000. The second dimension is, ‘How long?’ Because you’re talking about quantum states. So you measure these in, let’s call it milliseconds.

I may be stretching it a little bit, but let’s say they can work for a millisecond. So the question becomes how much competition can you get done? So, Osprey is going to have more coherence time.

And the third dimension is around errors because errors creep in as these machines keep up. So you can say, ‘What is a single qubit error? What is the coherence time at which point you collapse into pure noise?’ And those are all examples of improvement.

But I would also like to call out my software colleagues with the Qiskit because they’re putting more and more prebuilt circuits, which means it’s more and more usable even for people who don’t want to get into, ‘How do you build circuits that can solve a particular chemistry or optimization problem?’

On IBM Investments In Services And Consulting

If I remember right, the market for consulting is $450 billion. And I think that’s actually an underestimate. I think the real market is even bigger.

We, best case. are going to have a market share of a few small single digits in this market. So we should think that there’s a few hundred clients that IBM may go to. And even within those clients, we have a small footprint.

We are aimed at those massive SAP implementations, at the large maybe Azure implementations.

But when we get to our MSP colleagues, that’s completely complementary. We don’t really do much in that space. When we think about most of the VARs and VADs, we don’t do anything in those spaces.

We’re also pretty clear on client segmentation. There are about 300 clients where we are going to be much more direct, but there is a lot of space for services by our partners.

And then there is the remaining 100,000 or 200,000 where we really are wanting our partners to focus on even for the VAR piece. Everything else applies across the board. But even for the VAR piece, everything but those few 100, we can work on.

By the way, we know many VARs have services so they can even work in the top 300, by the way, with the services and expertise that they bring.

On Quantum Moving Beyond Experimentation And In What Format

I’d like to be a little humble and say we’ve got to see how the market goes.

I want to begin by saying, ‘Do we see it as a service, or do we see it selling as a machine?’ And the answer is both.

I can go to many other models. Maybe it’s outcome based, based on what an algorithm achieves. Maybe it’s also through our own cloud. Maybe it’s through other clouds. Maybe a client wants us to run the machine for them. That’s as-a-service, I would believe.

Some may want to have it on their own premises, so that’s a machine being sold. So I want to say that it’s too early to predict. But from my answer, you can see us being open to all of the approaches.

Now you asked the question, ‘When do you see quantum moving beyond experimentation?’ I think that that’s going to be in the 2023 to 2025 time frame. That’s not all use cases. So I want to be careful. I think some simple materials use cases, some optimization use cases, will be somewhere in that time period. But if you think about pharmaceutical drugs—an area we’re quite excited about, quantum—I think that’s probably going to be a little bit later.

Now, the interesting point is, when you talk to some of those companies—and we are in deep discussion with a few of those biotech companies, but they’re not public, so I’m not going to name them but you can all imagine who they are and how—let’s put it this way.

COVID vaccines have taught many of them that computation as applied to medicine can make things happen a lot quicker than it did only with a wet lab. And so consequently, they’ve all woken up to what computation can achieve.

And you could imagine that some of them might be thinking a bit further to say, ‘What could we do with quantum?’ So they’re working with us, but I would call this ‘experimentation’ today. But I predict somewhere between 2025 and 2030, they will see advantages also. So two different answers depending upon the complexity of the use cases.

On IBM’s Role In AI Regulation

Regulation may be the easiest but also the most complicated question. We as a company should not lead on regulation. What we should do is work with regulators to give them a way to think about AI. But they will be the ones who will make regulations, to be straightforward.

A great example is the work we did in the EU [European Union]. Because we do believe there should be precision regulation. I think a blunt instrument like ‘don’t do AI’ or ‘do do AI’ is too blunt.

But by precision we mean these use cases are allowed, in these use cases maybe a bit more carefulness. And by the way, in these use cases you’ve got to have a lot more carefulness and be able to show the regulator the data, the creating of the model, make sure the model doesn’t drift.

But confine that to the few things where there are a lot of implications. And the EU came up with—I think they call it level one, level two, three and four—and they divvied up the enterprise use cases into those four buckets. I think that’s a great example. But a lot more work is needed.

On AI bias and explainability, we have done a lot of work. We have put out our toolkits called the fairness toolkits. They’re open source. Anybody can use them. And we work with our clients as well as with other people in the industry about trying to advance those topics.

Look, I’ve been on the record as saying the following. ‘I fundamentally believe that AI is going to be a massive productivity booster for the globe.’ It’s not our estimate only, I think people estimate $16 trillion of productivity by the end of the decade. So that’s by 2030.

Now to get that for productivity—because we don’t have more than a few percentage of that today—you are going to need to work on both bias and ethics.

Now, in order to then achieve that full productivity, you’re going to say, ‘Well, how do I make sure that all the bias that may be in historical data is not there?’ And hence the push from us. Our researchers work on these topics and publish freely. We allow our open source toolkits to be out there that can be used by others and our clients. And so those are a few areas where we go forward. And we also encourage our researchers to participate in some of the conversations happening externally as well as regulatory bodies on these topics.

On Buying Envizi And Sustainability

Our internal goal … is straightforward. We have said that we are going to get to net zero for what we use and consume by 2030. Let me note: Not 2040. Not 2050. 2030.

We also believe that a simple goal at the end of a decade is insufficient. So we also set a goal of 65 percent by 2025. Because we believe the last third is harder than the first two-thirds.

So those are expressed goals. By the way, this is without purchase of offsets. We believe we should be able to get to well north of 90 percent, the last few percent we might have to do carbon sequestration and such techniques. So that is our internal goal.

And I hope you’ll all agree that that’s a reasonably aggressive goal compared to most others.

The reason I’m saying that I think Paris [Agreement, a global climate change limiting framework] is 2050, and that is for half. We’re at 400 by 2030.

Now, is it a signal that buying Envizi—I guess when you spend money and capital … I’d say it is a signal, so thank you for that question.

Look, when we do our surveys—I think it’s up to now 48 percent, maybe 51 percent, somewhere in that range—of CEOs and corporate executives agree that sustainability is now a serious business topic. Not just a checkmark, not just a little bit of a paragraph at the end of a long ESG [environmental, social and governance] report.

So given that sustainability is important, Envizi then begins to offer what I think is the first thing people need, which is reporting, data collection, clean data, analytics and ways to begin to make progress on the topic.

So the short answer is, yes, we do believe that sustainability is a big business opportunity. We try to be attentive to it, which is why our own commitments are getting there quickly.

We are also users of it. And so we believe that this is something that is both good for the business and good for the environment.

On AI Adoption

The reason that AI is getting and advancing more and more rapidly—we are at 2.5 quintillion bytes of data being produced each day. So that’s ‘2.5’ followed by 18 zeros.

There is no way that any amount of humans are ever going to process that. All analytic database techniques are insufficient. So AI is the only tool able to harness and harvest that data for insights and able to help the business.

Also, as the demographics imply that skills become more and more scarce, we are going to use more and more techniques like AI and the automation that it enables to go do that.

So the data shows the following—that now 35 percent of enterprises that we surveyed acknowledge that they are using AI in their internal processes and for how they interact with clients, and that is 4 percent up from last year.

Now, I personally believe that these numbers typically go 20, 25, 30, 35, 40. You reach a tipping point somewhere around 50 percent, and then it tips over to 90 very quickly.

So this means we’re just before that tipping point. And that is what unlocks all of the productivity I spoke about. By the way, the same survey also showed that people are worried about bias and ethics and explainability and all those topics. And it shows that the biggest inhibitor to AI adoption is actually clean data and data management and data governance within the enterprise.

On AI And Sustainability

I believe that the answer is an absolute resounding yes. When we look through some of the examples, without AI this is an incredibly manual task.

Let me take somebody who might have—I’ll be a little bit random, but [say] somebody who has 1,000 branch offices. Are the lights on at the branch offices all night when nobody’s there? So how do you reconcile the motion sensors? What may be the temperature in the building? Why are you running your air conditioning? Why can’t you turn it up to 82 degrees, maybe not so random a number, as opposed to turning it down to 70? What happens if a unit is completely malfunctioning and is running cold air continuously where it’s not even needed?

Or the other way—heating in all places, you can turn down the heating … I personally believe that there is about 30 percent efficiency to be garnered on these optimizations.

I mean we can turn the oil and gas consumption down by 30 percent. Imagine the benefits to the planet.

I’ll take another example that may not be so random again. Upstream, before we can even get consumable oil and gas, in oil fields there are a lot of greenhouse gasses that are in the production before it even gets to a tanker or a pipeline.

How can we begin to use, now, science techniques using AI to reduce that greenhouse gas production upstream? Something people don’t realize. As you begin to put all these together, the positive impact on the environment can be tremendous. But the sheer amount of data and the sheer amount of analysis needed means AI is the only answer. Otherwise, it’s impossible to be able to solve these problems.

On Former IBM CEO Lou Gerstner Jr. And Balancing IBM’s Role In Technology Of Today And Technology Of Tomorrow

[Former IBM CEO] Lou Gerstner [Jr.] is going to go down as one of the iconic business leaders of the past century. No question about it.

But when you think about Lou—and I do talk to Lou quite a bit, and he gives me a lot of very, very solid advice—No. 1, you’ve got to get your customer delighted.

If I have to delight customers, that’s going to be the basis of what we do today. So that’s going to be mainframe, AI, hybrid cloud.

As we delight them on this, some on to deploy technologies. Some want an end-to-end solution, which is where consulting plays a big role.

Then as we go beyond these, you also have to be future-proof. That is just the nature of technology. And channeling Lou—I mean, remember Lou invested in the internet starting in 1993. I don’t think people and businesses were really big on the internet until 1998. Lou asked us all to invest in Java in 1995. By the way, the inventors of Java thought it was only good for handheld devices and networked computers. Java turned out to be the basis of enterprise applications and still is.

So, the same way, it’s an add. To me, there is no conflict. You’ve got to invest in what can be used and deployed today. And you have to invest in a few things that are future.

But the lesson in there is also not everything. We at IBM should invest in those things that play to our strengths, and that our clients would like to get from us. So that’s kind of where I think the magic lies as opposed to everything. So our focus is critical.

On IBM Diversity Efforts

So we have said that we are going to scale 30 million people by the end of the decade.

Within that there are a number of diversity efforts and targets we have put out. We are members of OneTen, whose goal is to get a million Black Americans hired into well-paying jobs.

We also work with HBCUs [historically black colleges and universities], and we bring them into our AI, into our internship, into our apprenticeship programs. And so that’s not in conflict, that is simply reinforcing each of these.

When we look at veterans, this is great for business. I mean, veterans are highly, highly disciplined. And we have had great luck and success in giving them the training that lets them become great employees down the road.

But I would actually add to that, in addition to all those, there is also gender. We find that helping women re-enter the workforce sometimes after they take a five- or 10-year break can be very useful.

So there’s a whole series of these that are reinforcing to each other. And … maybe it’s a personal passion a little bit. I think [supporting] girls in STEM [science, technology, engineering and math], not just women in STEM, is also important. There’s a certain dropout rate that is happening as women leave high school that trying to reinforce that and participating in programs like Girls Who Code is important in the U.S. and in other countries. This is a global issue.

On IBM Researching Artificial General Intelligence

That artificial general intelligence, the ability to understand and learn an intelligent task humans can understand and learn should be research that you work with in the lab. … I don’t want us to work only on moonshots. We have to work on those things that are near end, that provide value this year and next year, in the year after, for our clients.

In addition, we might work on a couple of what we are labeling as ‘moonshots.’ Look, generalized artificial intelligence, personally, I think is still a long time away.

I think the last time we polled scientists—not just IBMers but other scientists—you get a spread. Some people believe it could be as early as 2030. But the vast majority put it out in the 2050 to 2075 range.

Personally, having grown up as a scientist, I react with anything that is 25 years or further away—my conclusion is that means we have no idea how this is going to happen.

So is it worth working on? Sure. Is it worth making that the majority of the effort? I think it is too stretched.