IBM Think 2025 | Sanjeev Mohan, Sanjmo

Clips
More from IBM Think 2025

Sanjeev Mohan

Principal

SanjMo

play_circle_outline IBM's differentiated strategy with focus on enterprise maturity and data integration

play_circle_outline Mainframes' role in IBM's differentiation, transaction processing, and reliability

play_circle_outline IBM Enhances Unstructured Data Strategy with DataStax Acquisition for AI, OpenSearch, and JSON Capabilities

play_circle_outline Rapid consolidation in the data market with companies offering integrated stacks

play_circle_outline Empowering Developers with Microsoft Build: Exploring the Excitement Surrounding Purview in Data Fabric

Info
Transcript

Sanjeev Mohan, Sanjmo

Sanjeev Mohan

Principal SanjMo

Sanjeev Mohan, principal at SanjMo, joins theCUBE’s Dave Vellante at IBM Think 2025 for a deep dive into IBM’s strategy in a fast-moving tech landscape. The conversation explores how IBM distinguishes itself among competitors through its enterprise depth and data innovation.

Mohan, a former Gartner analyst and expert in data management and analytics, shares insights into IBM’s evolving approach to hybrid cloud and mainframe integration. He highlights how IBM’s growing focus on unstructured data and its acquisition of DataStax align with the demands of ... Read more

explore Keep Exploring

What are some key differences in strategy between IBM and Oracle in the current market landscape? add

What is one reason that IBM mainframes are still relevant in today's technology landscape? add

What are the three main reasons why IBM bought DataStax? add

What impact is the rapid consolidation of technology companies having on the concept of defending a moat in the industry? add

What products or technologies are included under the umbrella term "data fabric" and how do they differ in their approach to handling data? add

bolt Powered by CUBE AI

Sanjeev Mohan, Sanjmo

search

Dave Vellante

>> Hi everybody. Welcome back to Boston. My name is Dave Vellante. We're here at theCUBE. We have a special guest coming in. We are almost wrapped up with our day here at IBM Think 2025. Sanjeev Mohan is in the house. He is the principal at Sanjmo and a member of theCUBE Collective, friend of theCUBE. Sanjeev, great to see you.

>> Likewise, always.

Dave Vellante

>> Thanks for making some time.

>> Thank you.

Dave Vellante

>> We always love to sit down with you. You're deep into data. You have a great background, former Gartner analyst. You have your own firm now. You're well-loved, super respected, so thank you for spending some time with us. I want to start with IBM's strategy. Now, one of the things that I often say about Oracle is they've got a highly differentiated strategy. Love them or hate them, but Oracle with its Exadata and its red stack, it's highly differentiated. IBM seems to be pursuing a different path in terms of its differentiation. Very open, heavy focus on integration and data. Obviously it's got its legacy businesses and it's what they call foundational or enduring businesses. What's your take? Is IBM in your mind as an analyst executing on a differentiated strategy?

>> I think IBM has a uniquely differentiated strategy. Oracle is amazing. We love Oracle and what it's doing now that Oracle is in every hyperscaler, Azure, AWS, and GCP now, but they are still a database company, so they're differentiated at that level. In fact, the reason they are in other hyperscalers is because other hyperscalers have better BI story, better ML story, to be honest. IBM is differentiated in the fact that it has enterprise maturity that very few other vendors have. And so with its mainframe... Mainframe, by the way, people don't even realize that even IMS and KICKS from way back are still alive and kicking.

Dave Vellante

>> So is COBOL.

>> And COBOL. 70% of world's transactions actually go through an IBM mainframe. So if you look at mainframe and everything that they have in the rest of the stack, it's differentiated. If you look at Google, for instance, Google Cloud is also differentiated because they have TPUs all the way up. So they're vertically integrated. IBM is not that vertically integrated. They rely on chips from other companies, although they have Telum too and Spyre, we learned about.

Dave Vellante

>> Yeah, they design some of their own silicon, yeah.

>> They make some, yes. But their differentiation comes from the fact that with Red Hat, they're deep into hybrid on-prem and cloud. Most hyperscalers, their revenue is compute. They're driving compute. Snowflake is encouraged to run on AWS at the cost of Redshift because it drives compute. IBM is not in the same game because IBM has IBM Cloud, but it's not really their center of focus. They can run very well on AWS. They are driving this application story now with AI agents, with IBM Consulting. So they're coming from top down and data being spread across hybrid and cloud and on mainframe gives them the differentiation.

Dave Vellante

>> It's interesting. I just wrote a report on the... I did my cloud summary. Amazon, as you saw, AWS specifically has 39% operating margins. Not talking about gross margin, talking about operating margin.

>> Oh wow.

Dave Vellante

>> That's tremendous. I mean, that's software company level. I mean, even the Oracles and the Microsofts of the world, mayve they're in the low 40s percent range. Broadcom and video have higher ones, but that's because they got crazy pricing these days. Broadcom has just an amazing business, but 39% operating margin for what is essentially a hardware company, to your point, that sells compute. Now, I will say this, that we published a while back that less than 50% of AWS's revenue is coming from compute. I don't know if you know that or believe that, but it's less than half, but it's still almost half. So that's a huge chunk. And so storage is also a decent size piece. Let's call it 10 to 15%. And so I think from Amazon's perspective, they would like a higher contribution from data and the data tooling, the AI work that they're doing-

>> Correct.

Dave Vellante

>> and further up the stack with Q and things like Connect. But they have a long way to go. I mean, probably what, 2% at best, maybe 1% of their business comes from SaaS, whereas Microsoft, for its cloud, it's probably 70% of its business. And I think Google's about 50/50. Google's operating margins of Google Cloud, we don't know about GCP, about 17, almost 18%. So big divergence there. Microsoft is probably in the 40% range. But to your point, IBM doesn't care about any of that infrastructure. I mean, they'll sell Z, they'll sell some Power, LGCY Power, but it's really, okay, so where is their value add? Where are they going to make money? Obviously in software.

>> Yeah, exactly. And their software, now that Linux is a first-class citizen on mainframe, they can take MongoDB for instance, and they can just take that MongoDB, no change, and run that software business on mainframe and then through their data gate, they can go populate Watsonx.data, which is a lakehouse, and do analytics on it. I mean, the potential is immense for mainframe and the rest of IBM software to grow the business.

Dave Vellante

>> Our colleagues are probably like, "Oh, Dave and Sanjeev talking about mainframe at IBM Think," but it's true.

>> It's true. Yeah.

Dave Vellante

>> They've got an integrated mainframe is the platform for vertical integrated hardware and software.

>> Correct.

Dave Vellante

>> No question about it. It never goes down. I mean, it really doesn't. I was talking to one of your former colleagues, a former Gartner analyst, database guy. I'm picturing him. Anyway, he said to me, we were at an event recently and he said, "You know, even Exadata goes down sometimes. Mainframes never go down." I was kind of curious because I always think of Oracle as very highly mission-critical, but the mainframe is the most mission-critical. I know banks that will buy the next Z sight unseen, "Just ship it. We don't need to test it. We just go." So it's legit. It's a highly differentiated business. And one point I wanted to make and see if this resonates is we always talk about bringing AI to the data on-prem, but there's a lot of do-it-yourself cobbling together and necessary doing your own integration that's not necessary with the mainframe. They're doing it for you. Don Feinstein is the name I couldn't remember.

>> I see.

Dave Vellante

>> Database guy.

>> Ah, okay. Yes.

Dave Vellante

>> So you buy that? I mean, does that make sense to you? Here we are talking about mainframe, but-

>> Yeah, so there is a program in AWS that's called Mainframe Organization Program. I'm sure you've heard of it.

Dave Vellante

>> Yeah, mainframe, the migration off of... yeah, yeah.

>> It is actually a migration. It's a euphemism for migration. So although IBM and AWS are friends, but given the reality, AWS is actually trying to get customers away from mainframe, but that business is actually growing, so I think mainframes will stay.

Dave Vellante

>> I don't think Amazon's going to have much... They've never had much success with that, and I've never... The interesting thing is you can convert COBOL to Java, okay, but the mainframe experience, banks aren't going to just chuck their mainframe. Here's why. They make money. If they can do transactions faster, they make more money.

>> Even if companies move from COBOL to Java, it still plays in IBM's camp because IBM is heavily invested in COBOL and Java. The funny thing is IBM doesn't want to compete in Python because they know that Databricks and all these other companies are going to .

Dave Vellante

>> It's a race to the bottom.

>> Yes. So they want to stick to where the enterprises are, and that's COBOL and Java.

Dave Vellante

>> All right, let's get off the mainframe. Let's get to data.

>> Yes.

Dave Vellante

>> The DataStax acquisition, they don't talk about it much here because they can't, the acquisition's not done. I mean, I remember the early days of DataStax and Billy Bosworth and Cassandra, and I always liked that I felt like DataStax never had the marketing engine and the go-to-market engine, but I always thought they had a great product. What's your take on DataStax? Why is IBM buying DataStax? Where does it fit into the portfolio and what's the future?

>> So I was following Cassandra from its early days when it came out of Facebook. So I think the reason why it never really caught on was a self-inflicted wound by DataStax. DataStax had this idea that, "We are special, we don't want to play in the open source stuff," so they divorced themselves from Cassandra, and that was not a good move. It was never really accepted by the public. In fact, every time you buy an iPhone, that iPhone transaction in Apple is actually going on Cassandra.

Dave Vellante

>> Is running on... yeah.

>> Not DataStax. Uber is Cassandra, not DataStax.

Dave Vellante

>> They should have done a managed service the way that Databricks did with Spark. That was the smart move.

>> Correct. So then DataStax realized the folly and they tried to come back, but it was already too late.

Dave Vellante

>> Too late, so the market had moved.

>> Correct.

Dave Vellante

>> Okay. Yeah. So IBM, first of all, they picked it up probably for a good price. I don't think the price was announced, but where does it fit into the IBM portfolio and what's the future of that technology?

>> At this Think 25, we are learning that the big focus for IBM is unstructured data because they feel that structured data stories pretty much .

Dave Vellante

>> They got that nailed.

>> Yeah, we nailed it. Let's now focus on unstructured data. There is a vector database that's embedded Milvus into their lakehouse. By the way, AWS SageMaker does not have a embedded vector database. It's at OpenSearch.

Dave Vellante

>> We use Milvus. theCube AI is Milvus, which is an open source.

>> Correct.

Dave Vellante

>> Chinese, isn't it?

>> Yes. Zilliz is a company, and yes, that is correct, based in Bay Area, though. A lot of Chinese companies, by the way, are rebranding to becoming American companies, so yes.

Dave Vellante

>> Of course. It's like Israeli companies going, "No, no, no, we're based in the Bay Area," but please, carry on.

>> There is a company that I know very well called PingCAP. TiDB's database is becoming an American company. Even the name is going to change to TiDB. Maybe I shouldn't have said that.

Dave Vellante

>> Good. Bring that technology here. I love it.

>> Yes, correct. Okay, so my point is that the reason for buying DataStax is not Cassandra. Cassandra's open source. If they bought it for Cassandra, that would not be a very smart move because everyone can get to Cassandra. They bought it for Langflow. Langflow is the business developer friendly way to build your generative AI applications.

Dave Vellante

>> There's an AI play here is why they bought it.

>> It is mainly for AI. So there are three reasons why they bought it. One is AI, which is Langflow. Second is OpenSearch, which I just talked about, which is an amazing search engine. And then the third thing is for the JSON capabilities that Cassandra provides, in that order.

Dave Vellante

>> And the fourth is it made financial sense. Jim Kavanaugh, I'm sure, did the IRR on it and said, "Oh wow, we can pick this up for a relatively low denominator." It wasn't like DataStax was a $60 billion market cap company. So they're probably not paying a lot. And when you put it through the IBM go-to-market machine, they're going to get an IRR. I mean, IBM seems to have its M&A mojo back. You heard me yesterday in the Rob Thomas AMA. I really do think that IBM used to be great at M&A and it kind of lost its way and now it's very good at M&A. It appears to be, anyway.

>> So talk about M&A, by the way, DataStax acquisition came as a surprise to me, and the reason is because I had felt that there is a gap in IBM's portfolio, which is Postgres database, because AWS has Aurora, Google Cloud has Alloy and Cloud SQL and Spanner. So they have three different Postgres and Azure has its own, but IBM does not have a Postgres database. So I was expecting that they'll buy a Postgres database and not DataStax. So that was a surprise.

Dave Vellante

>> Explain that. I thought, so Aurora is MySQL, no?

>> No, no. Aurora is MySQL and Postgres.

Dave Vellante

>> Oh, okay.

>> Yes. So it is managed MySQL, managed Postgres.

Dave Vellante

>> Okay.

>> And AlloyDB is the same idea, but for Google Cloud, and then Microsoft Azure has its own.

Dave Vellante

>> What are you... Changing topics, IBM's making a play in the data stack, no pun intended, to really be the integration engine.

>> Correct.

Dave Vellante

>> They're playing in Switzerland and not trying to take Databricks head-on.

>> Correct.

Dave Vellante

>> I mean, they've got database, so in that sense, they're competing against Oracle and Snowflake, but they're really trying to go after that white space across the stack from the harmonization layer, the governance layer up to the point where they can support intelligent data apps. What do you make of that play? How do you think about that market? I think of it sometimes as a modern day integration like a TIBCO or a Boomi, but maybe that's not the right way to think of it. What do you think about that?

>> So one of the problems IBM has had historically is that there are too many products, way too many products. During iPaaS days, they bought a company called Cast Iron. They had FileNet, they had WebSphere, WebMethods.

Dave Vellante

>> Mind-boggling set of companies.

>> So this is the first time in the last couple of years I've seen there's a much more focus on telling a more cohesive story. So although there are a lot of new products, so I think with Watsonx.data, Watsonx.governance, Watsonx.ai, they've got the three pillars. They've got data, AI, and governance. Watsonx Orchestrate is the layer they're putting on top to orchestrate different pieces and help business users develop AI pieces. So I think IBM is now doing a much better job in unifying the story, even though it's still a lot of pieces.

Dave Vellante

>> Can they build a moat around that story? I mean, maybe moat is an outdated concept, but I don't think it is. I think it's still... If you can maintain a competitive advantage and be open and you can dominate that business, you're going to attract customers, you're going to get good margin. But do you think that's a big enough business? I mean, the TAM is probably pretty big and the competition seems to be in pockets and there seems to be... Nobody's trying to swallow the whole thing in the middle, not the compute necessarily and the storage, even though IBM has compute and storage. Let those do whatever they do and stand on their own. But that whole sort of new middleware piece, what do you think about that? Is that a big enough market for them to achieve?

>> I don't look at it as middleware, because to me, middleware reminds me of just the integration iPaaS layer and WebMethods -

Dave Vellante

>> I know, but what do you call it? I mean, it's-

>> It's an integrated stack and I think this is where we are going to see some massive changes happening. I'll give you some examples of what I have observed in literally the last few days. So today we are talking about IBM buying DataStax. It's not a new story, but relatively new. Today I also found out ServiceNow, so part of your crew is at ServiceNow, right?

Dave Vellante

>> Yeah, yeah.

>> So they announced they're buying a data catalog company, Data.world.

Dave Vellante

>> Oh really?

>> Yes.

Dave Vellante

>> Well, that's an interesting move for them. Smart.

>> Yes, correct.

Dave Vellante

>> Smart move. I like that move. That's good for them.

>> Yesterday there was a rumor, a public rumor that DataStax may buy a serverless Postgres company called Neon.

Dave Vellante

>> Wait, wait, who? DataStax?

>> Sorry, Databricks.

Dave Vellante

>> Databricks.

>> Yes. So Databricks may buy Neon. It was public.

Dave Vellante

>> A serverless company?

>> Serverless Postgres, which will put them into operational-

Dave Vellante

>> Which is, of course they announced their serverless play I think a year ago, right?

>> Correct.

Dave Vellante

>> Which was kind of a game changer.

>> So serverless on the analytical side. This would be serverless on the operational side. You know what Snowflake has been trying to do with Unistore and-

Dave Vellante

>> What happened to Unistore?

>> It's called Hybrid Tables.

Dave Vellante

>> Is it alive?

>> I don't know. We'll find out. We'll be at Snowflake's summit in three weeks.

Dave Vellante

>> Sounds like you have some information that you can't disclose.

>> Yes, yes. So we'll find out. Okay. So last week, Datadog, they bought a data observability company called Metaplane from here, Boston. So it came out of MIT. So my point is that what you're seeing now is that there is a massive and a rapid consolidation that's already kicked off. And I think it's going to gather steam and companies like ServiceNow, Palantir, Salesforce, they're all going to offer the same services like Databricks, Snowflake, hyperscalers, IBM, Cloudera, Oracle, they'll all be lakehouses. So to your question, how do you defend the moat? I don't know. Because this story is being written and everybody wants to now offer an integrated stack. So those days of modern data stack where we had best of breed specialized software has rapidly come to a screeching halt. Customers are like, "Well, if I'm a ServiceNow customer and I get a data catalog, why would I go with a standalone data catalog?"

Dave Vellante

>> Right. It's interesting because what we've talked about is the point of control moving from the database to the governance catalog, but the governance catalog is opening up and it's wild, wild west. Open table formats bring complexity in terms of governance and then the values floating up the stack. So when you think about the companies like Databricks and Snowflake, they're now competing in a way with the ServiceNows and Salesforces. You look at Salesforce with its data cloud and its federated data strategy.

>> Yes, correct.

Dave Vellante

>> We asked Benioff and it resonated with him, "Could you be another hyperscaler, a software-only hyperscaler?" We'd love that, right? Come on into our data cloud. We'll harmonize the data. We'll control those agents for you. We have the best agent platform. So it's a whole new competitive dynamic for those companies.

>> Yep. And add to that Nvidia, because you know Nvidia has already moved up.

Dave Vellante

>> Yes, with , and yeah, right.

>> If they buy a database company-

Dave Vellante

>> company. Right.

>> Actually, OpenAI is another one. OpenAI did buy Rocket Software, but they weren't into... but I wouldn't be surprised.

Dave Vellante

>> So how does IBM fit into all that?

>> So I think, again, we started a conversation with mainframe. To find a moat, you have to find something that nobody else is doing or even can do. And nobody can do mainframe, nobody can do what Red Hat OpenShift does. So with HashiCorp, with Turbonomic, Aptio and Instana and all of those things, I think IBM is in a very, very unique position that none of these other players we talked about can match.

Dave Vellante

>> Let me push on that because one of the things, you just said something and the previous IBM guests were talking about it as well, Red Hat has a strategic advantage, but Red Hat's open. Anybody can work with Red Hat, right? And so is that moat defensible and why is it? Is it because they can integrate and do things like LinuxONE that are totally differentiable and-

Dave Vellante

>> InstructLab.

Dave Vellante

>> InstructLab, so they can do things that are unique to IBM with integration.

>> Correct.

Dave Vellante

>> So it's open. Anybody can work with Red Hat, but at the same time, when you own the stack, you can do some optimizations that nobody else can do or at least not do as well.

>> Correct. Yep, that is true. I think that's IBM's moat.

Dave Vellante

>> What do you think about what Zhamak is doing with Nextdata?

>> So I love Zhamak.

Dave Vellante

>> First of all, how do you think about it? I know basically, that's a but, so how do you describe what she's doing? Because it's somewhat complicated.

>> It is complicated. So her take is that, "Data products are here to stay. I'm going to deliver them in containers." That's the Nextdata OS, Data Mesh OS or whatever she's calling it. So I think there's a problem. We made a leap of faith that data products are here, and I don't think data products are here. Although I've spent better part of my last few years talking about data products, I don't think businesses are clamoring to buy data products. They're clamoring to solve their business problem, whatever label you may put on it. So I think Zhamak has gone to the next level of the deployment story of containerizing and deploying data product, but the first half is not really established and I think that's going to be a big change.

Dave Vellante

>> So it's a necessary condition for her.

>> Yeah.

Dave Vellante

>> Okay. Final thoughts on what you learned this week and what's next for Sanjeev Mohan?

>> Microsoft Build is in two weeks, the week after next. So I'm very excited to go there because we haven't talked about data fabric, but whatever... Again, data fabric is a label, but if you look at AWS SageMaker, lakehouse. BigQuery and a bunch of other things, they're all kind of data fabrics. So I'm very interested in understanding how these different offerings stack up against each other. On surface, they look very similar, but when you start peeling the onion, you start finding out that there are actually some pretty critical differences in how they approach data.

Dave Vellante

>> Data fabric. I'm more excited about Purview than I am data fabric.

>> So Purview is part of the data fabric story, although it's sold as a separate product, but it is integrated into this.

Dave Vellante

>> Cool. Sanjeev, thank you.

>> Thank you so much. It was a pleasure.

Dave Vellante

>> Appreciate your time. We'll see you around the block.

>> Yes, absolutely.

Dave Vellante

>> Okay, thank you for watching. This wraps up the SiliconANGLE and theCUBE's coverage of IBM Think 2025. Again, thecube.net, all the on-demand videos, SiliconANGLE.com for all the news, thecuberesearch.com for all the deep dive research. Check out thecubeai.com. Have some fun with it, ask it questions and see what you get. Thanks for watching everybody. This is Dave Vellante and we'll see you next time.