Exclusive Coverage: theCUBE at IBM Think | Sanjeev Mohan | SanjMo

theCUBE.net

Clips
More from Exclusive Coverage: theCUBE at IBM Think

Sanjeev Mohan

Principal

SanjMo

play_circle_outline Revolutionizing Data Architecture and Access: Exploring AI's Dependence on Data and Modern Compute Engines

play_circle_outline Unlocking the Power: IBM's Crucial Role as a Consolidating Force for Hyperscaler Consulting and Cloud Integration

play_circle_outline IBM's history in data and their potential in abstraction and agent technology

play_circle_outline Unlocking New Horizons: Harnessing AI Agents for Mainframe Modernization, Automation, and Orchestration

Info

Sanjeev Mohan | SanjMo

Sanjeev Mohan

Principal SanjMo

search

>> Hello, welcome back to theCUBE's coverage here in Boston, Massachusetts. I'm John Furrier, host of theCUBE. We have the pop-up CUBE. We'll do whatever it takes to get the stories. We're here, and Sanjeev Mohan with SanjMo's here, CUBE alumni, CUBE collective contributor, and also very distinguished and industry analyst that recognizes the best industry analyst, former Gartner, got your own practice. Sanjeev, a great friend of theCUBE, thanks for coming on.

>> Thank you so much for having me.

>> So all we talk about when we're on theCUBE between you, me, and Dave is data because that's your forte. But this data's now gotten so big from an aperture standpoint. There's data applications, there's data engineering, there's data analytics, there's data science, all kinds of new gyrations of data. AI right now, it's very clear that the role of data is the number one discussion. How do you think about it? What's the platform architecture look like? Is it a reset? Is it a redo? Is the script flipping? So it's going to be a constant discussion, because as we heard from the industry of the past year, there's a platform shift, tokens. Data's going to be available all the time. You got to move data around a lot, which shouldn't be that expensive, or it could be. There's a lot of open questions. What's your take right now? As we look at the future now, you're starting to see the clear line forming. Old way? New way?

>> So I strongly feel that we are going to rewrite the story of data in the next two years. Completely and utterly, it's going to transform. So moving the data, although technology wise, it's not a big deal, I think we have very fast networks and storage, but the other factors that stop us from moving data around, like security, compliance. So what I am seeing is that modern data stack has pretty much collapsed. And I see there the pendulum is once again shifting towards unification and a little bit of hybrid model with centralizing the infrastructure. So unifying the storage for structured, unstructured for batch, for streaming, and then putting a common layer of metadata and a semantic layer so that we can disaggregate compute. So we are going to multi engine compute for sure. So it could be Starburst, Trino, it could be Dremio, it could be Panda, Spark, Ray, and then LLMs. So the entire compute is disaggregated based on what you need. How you get to compute is also under massive change because it used to be sequel and Python rule the game, but now with natural language, even that is changing. So a lot of changes are coming.

>> So I see here, you're right. Obviously we know the history of hyperconvergence.

>> Correct.

>> Separating storage and compute,-

>> Right....

>> that was a generational shift of value and new architecture. You're saying compute and data will be decoupled.

>> Yeah. So when we disaggregated, compute, and store, what we actually forgot was, although we disaggregated it, they were still bundled together. You couldn't buy compute for one and store from another, but that unbundling is happening now, and some of it is thanks to standards like Iceberg.

>> And by the way, Databricks event's coming up, so Snowflake really recently.

>> Yeah, correct.

>> We will talk a little bit about that going in, but we're here at IBM's event, which they have database chops too, DB2 and a number of other things, and a lot of open source experience. So IBM, no stranger to database and data. We'll come back to that. But if you look at the success of say, data lakes and lake houses, that kind of validates your thesis because that beat federated data modeling. Okay, so that's an indicator that that was a good step. So the question is, what's next after the data lake? Are you going to have this series of streams going off the lake? Are you going to create pipelines of new waterways? Or what's the... How is that going to... Because I can see the data lake, but I don't see that as the final answer.

>> Yeah, things will change for sure. But for now, when you say streams, so you mean-

>> Rivers....

>> water metaphor, right?

>> Yes. A lake is water and sitting there. Is it moving? So I got to move water out, get a little out, move it out somewhere, or do I create another data lake and replicate the leg? So I use synthetic data.

>> Right. So synthetic data will happen, but that's for a different reason. That's orthogonal to our conversation. But when you said streams, my mind immediately went to the real-time-

>> Streaming....

>> streaming data because even that is now being written into the lake house format. So Confluent, for example, announced something called table flow. Every single vendor is now starting to support riding their JSON, Parquet, Avro, CSV, everything with a common table format. So there's a lake house wars going on right now, Delta from Databricks, Iceberg that pretty much everybody else supports. This week, there are actually four conferences, major conferences going on. We at IBM Think. There's Microsoft Build, there is a Dell Technology World,

>> Tech World....

>> and then there's Informatica world. So at Microsoft Fabric this week, Microsoft Fabric is a very big Delta user and they're going to announce support for two things, Iceberg and Fabric writing, reading data directly from Kafka, Kinesis and Google Pub/Sub.

>> Does that commoditize or is that unify?

>> I think it commoditizes compute completely. It doesn't matter what compute you bring. If my data is in a common format that any compute engine can access, then that compute becomes a commodity.

>> Okay, so who wins and who loses in that scenario?

>> So the way things are going, I think there are only one set of winners, and they're the hyperscalers. If you look at even the model, hardware will get commoditized. I think all the power is sort of concentrating into the hyperscalers.

>> The rich get richer. Here we go. How does IBM fit into that? They're not a hyperscaler. They missed that wave. They tried with SoftLayer and Bluemix, they scuttled that ship. Now they're in a whole other ball game. They're unifying through distributed computing. Red Hat was a great bye by Arvind. Loved that move. That was like his killer move when he became CEO because he's got that Satya Nadella vibe going on right now where he's reinventing IBM as a product company, not so much a services company. So I have to ask you, where do you see IBM playing it? Because they're not a hyperscaler. They're almost like a glue layer between all the hyperscalers. They're doing a deal with AWS, they're selling their software. watsonx in the marketplace. They got to deal with Azure. So why do they need to build a cloud? They just become a partner for the cloud providers.

>> Right. So when I said all the power is concentrating in hyperscalers, but then there's a cost to that, and the cost is that now you're locked into their own stack or their own ecosystem. IBM actually has a huge advantage because IBM can be that cross cloud, like what you call super cloud and hybrid because of Red Hat OpenShift, and all these other HashiCorp, Terraform now, and Ansible. So I think IBM has a very key role to play as a glue between all these different islands.

>> IBM could be the super cloud layer-

>> Yep, completely...

>> that connect, connects all the clouds together-

>> Right....

>> by bringing all their IP to the table-

>> Right....

>> and be that layer. Arvind said it today on the partner keynote, and I've never heard IBM speak like this before, "The pie is big enough for everybody." IBM was not a pie-share in their old days.

>> Right. And also, John-

>> That means they want to work together.

>> Yeah. Also, what's exciting is that all these hyperscalers we talked about, none of them have a very good consulting arm, and IBM does. So IBM can not only be that technology glue, but also have the services arm.

>> It's interesting because you know Azure, you know AWS, you know Google, they all have a great relationship with the Accentures, the PWCs, the Deloittes, the Persistent Systems of the world, massive integrators, WePro.

>> Yeah.

>> What's IBM's relationship to that? Do you think that's going to be threatening to some of those relationships or additive?

>> I don't know. Hopefully it'll be additive. I just came here today, although today, like you said, is a partner day. Our event starts tomorrow and I'm already blown away because there are a few thousand partners here. This place is packed. And in fact, I was surprised to see so many people. I asked somebody. They said, "Oh, our number of partners, we have are in tens of thousands."

>> Yeah. Yeah. And I tell you, the new IBM, they're back. You can see I've been saying it all on our intro package here, the test for Arvind and the team will be, can they get a partner ecosystem going? Because their old partner network tend to be true blue partners, blue washing. They buy a company, they call it blue wash and they become IBM blue. That's their color. So very IBM centric partners.

>> Correct. Yeah.

>> To be a true industry partner, like Amazon has a great ecosystem, so does Google now developing, you got to be independent. You got to offer them that value.

>> Yeah.

>> So the question is can these partners, can IBM sell more through the partners, and can the partners sell more through IBM?

>> Correct.

>> So IBM looks to me, at least from what I'm seeing here at the event and briefing with other customers on the side, is they're lining up properly. They're saying, "Hey, look, we'll give you distribution with our salesforce, go to market. So sell through us, you can sell through you, and yeah, you can have some services." The pie, the sharing the pie.

>> Correct.

>> Arvind is saying directly, "We want to do business with you."

>> Right.

>> Now the question is, will they follow through? We'll see, but I don't see any reason that they're not.

>> When I talk to the product heads in IBM, I am surprised they're asking me who should we partner with. And when I tell tell small companies they should partner with IBM, they're shocked. They're like, "Wait, IBM and us? We are too small." I'm like, "No, no, IBM is asking to partner with small companies." So the equation has shifted.

>> If they could replicate what Amazon did and create a true enablement model with partners-

>> Right, Correct....

>> where there's a good distribution and good monetization, actually they could pull it off. They got the big brand, but they haven't done it before, so we'll see. I'm very skeptical on that. I'm not down on it. Show me. That's where I'm at.

>> Correct.

>> Well, let's get into the data side, back to the database because we got Snowflake and Databricks coming up.

>> Correct.

>> TheCUBE will be at, it looks like definitely Snowflake four days, and I'm waiting to hear from Databricks since we're going to get spotted Databricks as well, their event. So we'll have theCUBE at both events this year like we did last year. IBM is no stranger to data. How do you see their data play here? Because they have a rich history, DB2 and a variety of lots of data products, data analytics. I've been to the past decade many times. In fact, he quoted the famous line that I still use today. "You can't have AI without IA," information architecture. I don't know why they don't bring that back, because that's a great slogan. So they have history, history spark.

>> Yeah.

>> What's your assessment? Have they been playing off defense, not enough offense. What's your analysis of IBM's data?

>> So they have history, but sometimes that history is not that rosy. I remember the days when you talk to one team at IBM, they say, "Oh, we only do final net," one team said, "We only do cast iron," one team that only did web sphere. I'm glad that some of these things islands have gone away. The conversations I'm now having with IBM, surprisingly are now on how do we bring DB2 and Informix into the mix? How do we showcase Cognos TM1? So these are very mature products that are no longer kind of sexy anymore, but IBM is... If Microsoft could bring back the mojo, IBM definitely can.

>> If you look at what Dario presented today... Dario Gill is on stage. You probably may or not have seen the keynote, but he was doing a demo with COBOL, training an agent. What's going on right now? We are living in an abstraction layer revolution as well, systems revolution, I've been saying on theCUBE. But you can look at Cognos and Informix and DB2 and look at all that pre-existing brownfield or existing legacy. You could put an abstraction layer around that. You could build an agent technology and low code or no code automatic coding-

>> Correct....

>> to connect this and integrate it. So we're seeing the iPass market, integrated platform as a service, connect with the new Gen ai. So I see an opportunity big time.

>> Actually, you don't even need to put abstraction layer anymore because now with LLMs, the game has again transformed. For example, I know a bunch of banks, over my years of consulting that have been on this mainframe modernization program and it has stalled. How do you modernize a joint mainframe? But now with an LLM, that's trained on thousands and thousands of COBOL programs. I can actually use an LLM to do 80% of my COBOL migration to Java and modernize my entire mainframe ecosystem.

>> I think your point about that is so big, and this is why I'm bullish on IBM. I was just talking with another analyst on earlier today, that as these agents become more sophisticated, not chat bots, we're talking about real agents,-

>> Correct....

>> real AI agents that are working on behalf of something, a user, a program, a machine, can do a lot of work on well bounded, well-formed environments. So things like mainframe migration integration, API management, observability.

>> Testing.

>> Testing.

>> Yeah. Deployment. Yeah. Yeah. Yeah.

>> Just working down the manual automation tasks and with Ansible.

>> Right.

>> IBM and Red Hat, and now HashiCorp... I think HashiCorp is going to be much more impacted by agents than say, Ansible and Red Hat because there's not a lot of GenAI in there. But other than doing some small things, but with Terraform and DevOps, that can be completely run by agents.

>> That's a future. I think next year, it'll take a couple of years, a year maybe or two years, but I think having agents that can understand the context and the environment, and then reason, and then do orchestration and use HashiCorp to products to do that, I think that'll be an amazing future.

>> Sanjeev, great to have you on theCUBE and great to have you as a Collector, a CUBE collective member.

>> Thank you so much.

>> And I want to ask you one final question. We're here at IBM Think in Boston. If we had the executive team in front of us, Arvind and his staff, what would you ask them and what would you advise them? First of all, we'll do things. Give them a direct order, mandate. And then what question would you ask?

>> So the mandate would be make things as simple as possible and not any simpler for the consumers. Consumers do not have the time of the day to be researching like we do as analysts. They have a day job. So my question to him would be, how can you have a cohesive strategy that solves the business problem, not technology problems.

>> Sajeev Mohan here on theCUBE with me, John Furrier, your host. We are in the Pop-up CUBE, small little one-camera shot. We'll do whatever it takes to get the content, get the best analysis we can, share that with you. That's our mission on theCUBE. And behind me, tomorrow we will be in that convention center in Boston, getting here in Arvind to talk about the vision for IBM. And this next year, as they go to the next level, IBM well positioned to make it to the winner's circle in this next generation AI rates. I'm John Furrier for theCUBE. Thanks for watching.

Sanjeev Mohan | SanjMo

search

>> Hello, welcome back to theCUBE's coverage here in Boston, Massachusetts. I'm John Furrier, host of theCUBE. We have the pop-up CUBE. We'll do whatever it takes to get the stories. We're here, and Sanjeev Mohan with SanjMo's here, CUBE alumni, CUBE collective contributor, and also very distinguished and industry analyst that recognizes the best industry analyst, former Gartner, got your own practice. Sanjeev, a great friend of theCUBE, thanks for coming on.

>> Thank you so much for having me.

>> So all we talk about when we're on theCUBE between you, me, and Dave is data because that's your forte. But this data's now gotten so big from an aperture standpoint. There's data applications, there's data engineering, there's data analytics, there's data science, all kinds of new gyrations of data. AI right now, it's very clear that the role of data is the number one discussion. How do you think about it? What's the platform architecture look like? Is it a reset? Is it a redo? Is the script flipping? So it's going to be a constant discussion, because as we heard from the industry of the past year, there's a platform shift, tokens. Data's going to be available all the time. You got to move data around a lot, which shouldn't be that expensive, or it could be. There's a lot of open questions. What's your take right now? As we look at the future now, you're starting to see the clear line forming. Old way? New way?

>> So I strongly feel that we are going to rewrite the story of data in the next two years. Completely and utterly, it's going to transform. So moving the data, although technology wise, it's not a big deal, I think we have very fast networks and storage, but the other factors that stop us from moving data around, like security, compliance. So what I am seeing is that modern data stack has pretty much collapsed. And I see there the pendulum is once again shifting towards unification and a little bit of hybrid model with centralizing the infrastructure. So unifying the storage for structured, unstructured for batch, for streaming, and then putting a common layer of metadata and a semantic layer so that we can disaggregate compute. So we are going to multi engine compute for sure. So it could be Starburst, Trino, it could be Dremio, it could be Panda, Spark, Ray, and then LLMs. So the entire compute is disaggregated based on what you need. How you get to compute is also under massive change because it used to be sequel and Python rule the game, but now with natural language, even that is changing. So a lot of changes are coming.

>> So I see here, you're right. Obviously we know the history of hyperconvergence.

>> Correct.

>> Separating storage and compute,-

>> Right....

>> that was a generational shift of value and new architecture. You're saying compute and data will be decoupled.

>> Yeah. So when we disaggregated, compute, and store, what we actually forgot was, although we disaggregated it, they were still bundled together. You couldn't buy compute for one and store from another, but that unbundling is happening now, and some of it is thanks to standards like Iceberg.

>> And by the way, Databricks event's coming up, so Snowflake really recently.

>> Yeah, correct.

>> We will talk a little bit about that going in, but we're here at IBM's event, which they have database chops too, DB2 and a number of other things, and a lot of open source experience. So IBM, no stranger to database and data. We'll come back to that. But if you look at the success of say, data lakes and lake houses, that kind of validates your thesis because that beat federated data modeling. Okay, so that's an indicator that that was a good step. So the question is, what's next after the data lake? Are you going to have this series of streams going off the lake? Are you going to create pipelines of new waterways? Or what's the... How is that going to... Because I can see the data lake, but I don't see that as the final answer.

>> Yeah, things will change for sure. But for now, when you say streams, so you mean-

>> Rivers....

>> water metaphor, right?

>> Yes. A lake is water and sitting there. Is it moving? So I got to move water out, get a little out, move it out somewhere, or do I create another data lake and replicate the leg? So I use synthetic data.

>> Right. So synthetic data will happen, but that's for a different reason. That's orthogonal to our conversation. But when you said streams, my mind immediately went to the real-time-

>> Streaming....

>> streaming data because even that is now being written into the lake house format. So Confluent, for example, announced something called table flow. Every single vendor is now starting to support riding their JSON, Parquet, Avro, CSV, everything with a common table format. So there's a lake house wars going on right now, Delta from Databricks, Iceberg that pretty much everybody else supports. This week, there are actually four conferences, major conferences going on. We at IBM Think. There's Microsoft Build, there is a Dell Technology World,

>> Tech World....

>> and then there's Informatica world. So at Microsoft Fabric this week, Microsoft Fabric is a very big Delta user and they're going to announce support for two things, Iceberg and Fabric writing, reading data directly from Kafka, Kinesis and Google Pub/Sub.

>> Does that commoditize or is that unify?

>> I think it commoditizes compute completely. It doesn't matter what compute you bring. If my data is in a common format that any compute engine can access, then that compute becomes a commodity.

>> Okay, so who wins and who loses in that scenario?

>> So the way things are going, I think there are only one set of winners, and they're the hyperscalers. If you look at even the model, hardware will get commoditized. I think all the power is sort of concentrating into the hyperscalers.

>> The rich get richer. Here we go. How does IBM fit into that? They're not a hyperscaler. They missed that wave. They tried with SoftLayer and Bluemix, they scuttled that ship. Now they're in a whole other ball game. They're unifying through distributed computing. Red Hat was a great bye by Arvind. Loved that move. That was like his killer move when he became CEO because he's got that Satya Nadella vibe going on right now where he's reinventing IBM as a product company, not so much a services company. So I have to ask you, where do you see IBM playing it? Because they're not a hyperscaler. They're almost like a glue layer between all the hyperscalers. They're doing a deal with AWS, they're selling their software. watsonx in the marketplace. They got to deal with Azure. So why do they need to build a cloud? They just become a partner for the cloud providers.

>> Right. So when I said all the power is concentrating in hyperscalers, but then there's a cost to that, and the cost is that now you're locked into their own stack or their own ecosystem. IBM actually has a huge advantage because IBM can be that cross cloud, like what you call super cloud and hybrid because of Red Hat OpenShift, and all these other HashiCorp, Terraform now, and Ansible. So I think IBM has a very key role to play as a glue between all these different islands.

>> IBM could be the super cloud layer-

>> Yep, completely...

>> that connect, connects all the clouds together-

>> Right....

>> by bringing all their IP to the table-

>> Right....

>> and be that layer. Arvind said it today on the partner keynote, and I've never heard IBM speak like this before, "The pie is big enough for everybody." IBM was not a pie-share in their old days.

>> Right. And also, John-

>> That means they want to work together.

>> Yeah. Also, what's exciting is that all these hyperscalers we talked about, none of them have a very good consulting arm, and IBM does. So IBM can not only be that technology glue, but also have the services arm.

>> It's interesting because you know Azure, you know AWS, you know Google, they all have a great relationship with the Accentures, the PWCs, the Deloittes, the Persistent Systems of the world, massive integrators, WePro.

>> Yeah.

>> What's IBM's relationship to that? Do you think that's going to be threatening to some of those relationships or additive?

>> I don't know. Hopefully it'll be additive. I just came here today, although today, like you said, is a partner day. Our event starts tomorrow and I'm already blown away because there are a few thousand partners here. This place is packed. And in fact, I was surprised to see so many people. I asked somebody. They said, "Oh, our number of partners, we have are in tens of thousands."

>> Yeah. Yeah. And I tell you, the new IBM, they're back. You can see I've been saying it all on our intro package here, the test for Arvind and the team will be, can they get a partner ecosystem going? Because their old partner network tend to be true blue partners, blue washing. They buy a company, they call it blue wash and they become IBM blue. That's their color. So very IBM centric partners.

>> Correct. Yeah.

>> To be a true industry partner, like Amazon has a great ecosystem, so does Google now developing, you got to be independent. You got to offer them that value.

>> Yeah.

>> So the question is can these partners, can IBM sell more through the partners, and can the partners sell more through IBM?

>> Correct.

>> So IBM looks to me, at least from what I'm seeing here at the event and briefing with other customers on the side, is they're lining up properly. They're saying, "Hey, look, we'll give you distribution with our salesforce, go to market. So sell through us, you can sell through you, and yeah, you can have some services." The pie, the sharing the pie.

>> Correct.

>> Arvind is saying directly, "We want to do business with you."

>> Right.

>> Now the question is, will they follow through? We'll see, but I don't see any reason that they're not.

>> When I talk to the product heads in IBM, I am surprised they're asking me who should we partner with. And when I tell tell small companies they should partner with IBM, they're shocked. They're like, "Wait, IBM and us? We are too small." I'm like, "No, no, IBM is asking to partner with small companies." So the equation has shifted.

>> If they could replicate what Amazon did and create a true enablement model with partners-

>> Right, Correct....

>> where there's a good distribution and good monetization, actually they could pull it off. They got the big brand, but they haven't done it before, so we'll see. I'm very skeptical on that. I'm not down on it. Show me. That's where I'm at.

>> Correct.

>> Well, let's get into the data side, back to the database because we got Snowflake and Databricks coming up.

>> Correct.

>> TheCUBE will be at, it looks like definitely Snowflake four days, and I'm waiting to hear from Databricks since we're going to get spotted Databricks as well, their event. So we'll have theCUBE at both events this year like we did last year. IBM is no stranger to data. How do you see their data play here? Because they have a rich history, DB2 and a variety of lots of data products, data analytics. I've been to the past decade many times. In fact, he quoted the famous line that I still use today. "You can't have AI without IA," information architecture. I don't know why they don't bring that back, because that's a great slogan. So they have history, history spark.

>> Yeah.

>> What's your assessment? Have they been playing off defense, not enough offense. What's your analysis of IBM's data?

>> So they have history, but sometimes that history is not that rosy. I remember the days when you talk to one team at IBM, they say, "Oh, we only do final net," one team said, "We only do cast iron," one team that only did web sphere. I'm glad that some of these things islands have gone away. The conversations I'm now having with IBM, surprisingly are now on how do we bring DB2 and Informix into the mix? How do we showcase Cognos TM1? So these are very mature products that are no longer kind of sexy anymore, but IBM is... If Microsoft could bring back the mojo, IBM definitely can.

>> If you look at what Dario presented today... Dario Gill is on stage. You probably may or not have seen the keynote, but he was doing a demo with COBOL, training an agent. What's going on right now? We are living in an abstraction layer revolution as well, systems revolution, I've been saying on theCUBE. But you can look at Cognos and Informix and DB2 and look at all that pre-existing brownfield or existing legacy. You could put an abstraction layer around that. You could build an agent technology and low code or no code automatic coding-

>> Correct....

>> to connect this and integrate it. So we're seeing the iPass market, integrated platform as a service, connect with the new Gen ai. So I see an opportunity big time.

>> Actually, you don't even need to put abstraction layer anymore because now with LLMs, the game has again transformed. For example, I know a bunch of banks, over my years of consulting that have been on this mainframe modernization program and it has stalled. How do you modernize a joint mainframe? But now with an LLM, that's trained on thousands and thousands of COBOL programs. I can actually use an LLM to do 80% of my COBOL migration to Java and modernize my entire mainframe ecosystem.

>> I think your point about that is so big, and this is why I'm bullish on IBM. I was just talking with another analyst on earlier today, that as these agents become more sophisticated, not chat bots, we're talking about real agents,-

>> Correct....

>> real AI agents that are working on behalf of something, a user, a program, a machine, can do a lot of work on well bounded, well-formed environments. So things like mainframe migration integration, API management, observability.

>> Testing.

>> Testing.

>> Yeah. Deployment. Yeah. Yeah. Yeah.

>> Just working down the manual automation tasks and with Ansible.

>> Right.

>> IBM and Red Hat, and now HashiCorp... I think HashiCorp is going to be much more impacted by agents than say, Ansible and Red Hat because there's not a lot of GenAI in there. But other than doing some small things, but with Terraform and DevOps, that can be completely run by agents.

>> That's a future. I think next year, it'll take a couple of years, a year maybe or two years, but I think having agents that can understand the context and the environment, and then reason, and then do orchestration and use HashiCorp to products to do that, I think that'll be an amazing future.

>> Sanjeev, great to have you on theCUBE and great to have you as a Collector, a CUBE collective member.

>> Thank you so much.

>> And I want to ask you one final question. We're here at IBM Think in Boston. If we had the executive team in front of us, Arvind and his staff, what would you ask them and what would you advise them? First of all, we'll do things. Give them a direct order, mandate. And then what question would you ask?

>> So the mandate would be make things as simple as possible and not any simpler for the consumers. Consumers do not have the time of the day to be researching like we do as analysts. They have a day job. So my question to him would be, how can you have a cohesive strategy that solves the business problem, not technology problems.

>> Sajeev Mohan here on theCUBE with me, John Furrier, your host. We are in the Pop-up CUBE, small little one-camera shot. We'll do whatever it takes to get the content, get the best analysis we can, share that with you. That's our mission on theCUBE. And behind me, tomorrow we will be in that convention center in Boston, getting here in Arvind to talk about the vision for IBM. And this next year, as they go to the next level, IBM well positioned to make it to the winner's circle in this next generation AI rates. I'm John Furrier for theCUBE. Thanks for watching.