iPods and XBOXes and early PCs, and we had data center customers and everything else, but we really grew up in that PC revolution, started building very low-cost boxes for everybody to get on the internet for the first time.
Right.
Now all the data, the mass capacity data we call it, is up in data centers. The boxes that have really morphed, the whole supply chain has had to morph over the last 10 years to the point where 80% of the world's data is on a hard drive in the data center right now, and it's continuing to grow at leaps and bounds. Largely because as the technology has moved, these application spaces have moved, and if I think anything about AI right now, it's really the different applications have just exploded. It wasn't happening 10 years ago, but there's so many different fronts where you've got applications pulling data, creating data, sending it back down, and then wanting to do it as efficiently as you can. That's what's exciting.
Yeah. I'm excited about the opportunity, as you are. I'm also amazed at how far the technology has come. You have photonics engineers, lasers. It's not your father's spinning disc anymore. I know we're going to get into that a little later, but it does just show how far it has moved. Some people might take for granted how much technology it takes to get on one of these single platters, and I'm amazed every time that I come back. So you and I have both been through some of all the major transitions out there in techno I guess, when we started. But, I was part of, I would say mini to mainframe was a little bit after me, but mainframe to client server, e-commerce, social media, the web, mobile. You talked a little bit about the iPod there. And there was a first wave of AI.
More of an edge play, in a way, object recognition. But here we are in this age of agentic AI. What is fundamentally different about this era that we're in compared to maybe all the other ones that I had discussed?
Yeah. I think software is being disrupted. A lot of people can speak more eloquently than I can about that. Not just software itself, but also the entire software ecosystem, the stack that is outside of most of the drives in the application spaces we see. We talk to a lot of cloud service providers. They have different applications. They are not all created equal, nor are their applications, but more and more what we're seeing is software that it's not lossy. There's not a whole lot of handoffs as you get up to the application layer. People don't want to pay for that. They want to pay for actually data in the application. So I think you're seeing a stronger tie between the ultimate application, call it agentic application if you want, down through the software stack and into the hardware now.
It's actually more integrated.
That's right.
Per se, less layers and less handoffs.
I think the cloud service providers have done a miraculous job at some of this architectural stuff, whether it's hardware architecture or software architecture. I think people are watching that and trying to duplicate that. It will ultimately come out to the edge as well.
Yeah. Most interesting one, I think, even related to data and storage is the compressing of the database layer, and even the embeddings layer, that fit inside of storage. We've seen a lot of startup companies and a lot of the incumbent storage companies pull that together. One of the biggest shifts we're hearing about, it was one thing to take advantage of structured data inside of an Oracle database or any type of relational database. But unstructured is an area of data that we really weren't able to exploit this year. Organizations really haven't taken advantage of that, and they've been rethinking their strategies to do this. What does the disconnect between how much data is actually available and I think just to be honest, the lack of use of it, how do you explain this disconnect between the two?
It's actually interesting because, like you said before, the applications are much faster now, enabled by Silicon Technologies that are making them really quick. Then you also have much more efficient algorithms being written to be able to extract data or create data or save data. From my perspective, down inside the storage layer, we've always been kind of subservient to the file systems and things above us. Now we're starting to see more what I'll call object stores.
Yeah.
Where what's really happening at layers above us, in memory layers, say, is that you are having very fast transactions on metadata, then when you need that specific data underneath, you pull it up. That architecture was never really honed before, but it is now. What that means is all of this, the data can be stored cheaper, relatively cheaper, and you can store a lot more of it on the hard drive tiers. Then you can use it much more effectively than you have before. To your point, we have had machine learning in our factories for a long time. A lot of these AI applications have been around for a while, but they are really coming to maturity given what is happening in the compute layers.
Yeah. Let us talk about a little, let me up level it. I got a little deep there on my question. Why is this unstructured data becoming more strategic? Was it just, it always could have been strategic, we did not have enough compute power, we did not have these new models to take advantage of it? What has changed? Why are companies finally viewing this as strategic? Because it used to be, "Hey, I am just going to delete this," or, "I am going to archive this. There is no value here.
I think you hit on it already, which was you could have already saved data before, but was it accessible to you? Where was it? Its geographic locality matters. Your ability to actually use the data matters. Then, can you get machines to actually pull the data in, do the right amount of scrubbing of the data, and send it back down, or do you have to have human intervention? All that has changed in the last 10 years dramatically. I think that is why the unstructured data is becoming much more useful. In some cases, the unstructured data is not even used, it is just the metadata around the unstructured data. You only know which data you need when the algorithm tells you you need it, you pull it up.
Yeah. You did a research study that talked a little bit about this. People are still deleting it, people are not deleting it. They see it as strategic. Can you walk through some of the biggest finding out of that?
We've tried to ask people doing some research on CIOs type people, what's your organization struggling with today, and where has it come from the last, say, 10 years? What we find is not surprising. Data's more important than it ever was. People are identifying that the fact that they need to store more data. There's a lot of, I'll say 55% of the people answering the survey said, "I've lost a lot of money because I deleted something that I shouldn't have deleted. That's not a penalty, that's not because of compliance issues. That's because of opportunity lost, because they deleted something. Running factories ourselves, we see this every day.
If we could only store information about what's going on in the factories for a little bit longer, we believe you can get utility out of that. Other companies will have the same kind of things on customer data. Whether it's customer acquisition or customer service or something. I think companies are really coming to the realization that I need to save this stuff, and if I lose it, then I might have lost opportunity.
Is the whole notion of cold storage just going to be a non-discussion anymore, where almost all of your data has to be hot or lukewarm or at least relatively easily accessible?
We call our products nearline products. These monikers, I don't think we've all come up with the right language yet. But, to me, cold storage means it's so far offline that it's virtually unusable unless a human brings it back.
That's right.
That probably is not going to be very efficient in the future. I think if we can continue to really drive our technology hard and grow exabytes in our industry.
Right.
Then that nearline will become the preferential storage mechanism.
Looking at the business leaders out there, the C-suite out there, what's the big mind shift that they're going to have to change to essentially get into a place where they can best leverage their data to help them in the age of AI?
It is interesting because everybody has their own skill sets, but not very many people have broad skill sets across all of these different vectors. What I would say is that you better be thinking about your data infrastructure. What are you going to keep? What are you going to throw away? You won't get it right all the time, but if you're thinking about it and you're trying to cost reduce it as much as you can, then you're preparing yourself for what's coming in the future. The software stacks are getting a lot more efficient, and they're getting cheaper. They're getting more accessible to everyone.
That's great.
They're getting faster because of all the compute stuff that we talked about before. If you ever run out of data that's new data that's serving that, you're going to be in trouble. Some of the data that happens out there in the world that occurs out in the world needs to be ingested more efficiently than it has before. Sometimes it needs to be stored for a day, a week, a month, sometimes longer. You better be thinking through all those vectors in your strategy.
At what point in the organization, so you're chairman of your board, and I'm curious, what type of level of discussion is this? Who needs to be accountable for that, lack of a better term, data architecture?
This is actually interesting as well because I think a lot of times the storage aspects of the company would be deferred to the CIO. And t he IT person is not necessarily the right person to be thinking through this. I think if you're a big, diverse company like we are, you have factories, you have customer service, you have all the normal G&A activities, you probably need different strategies for each. I think there's a lot of opportunity for people to come in and help with that. Also, you need the business leaders in each of the different functions thinking about this.
Is this on the Chief Data Officer? So there's Chief AI Officer in some of these companies, Chief Transformation Officer. I've seen a lot of Chief Transformation Officers turn into the Chief AI Officer.
Yeah.
There's Chief Data Officers. Is this where you're seeing this sit, or is it a BU led, or both?
There are different titles emerging, for sure. What I have always said in my career is we have to work cross-functionally. So I would expect teams of teams. Not necessarily one person, but how do you get all your.
Yeah.
Organizations to work together on a common strategy? Because to your point.
Yeah.
Usually, when people say Chief Data Officer to me 10 years ago, it was all about security and data leakage. It wasn't necessarily about what to store and how to use it. I think people have got to be adding these things to their repertoire, and then they've got to be working together cross-functionally.
Dave, if you had one thing to leave with the audience on where to start on this journey, what would you recommend?
I would say there's a bottoms up and a tops down way to think about data. From a bottoms up perspective, you have to make your entire organization aware of what the techniques are in AI today. Because mandating from the top down is not going to work. You want the people that are closest to the business problems to be working with the tools that are going to make them most efficient. From the tops-down perspective, you have to be willing to change very quickly because things are moving so fast on a lot of these different fronts. Not all of them. Grew up in a world where we learned a new application, then we didn't want it to change for 10 or 15 years, and you still see that in organizations. That's not what we're dealing with here.
I think we have to be a lot more nimble.
Yeah. If you look at your strategic long range plan, obviously not the financial part of it, but from a technology and visionary standpoint, what are some things that you're looking at right now that could maybe three years from now might even shift the conversation? Any research that you see being done, any major shifts, even, I want to say politically, but more culturally, that people need to be looking at?
First, I always answer with our technology, which we're really proud of. We're going to keep driving density, and we think we can, which is great, makes the data centers more efficient. We can also drive the technology that actually makes the data more accessible on each individual hard drive or at the system level. We're going to get pushed very hard on that because it's not only more data, it's also data faster, which are the vectors that our customers are really driving. Then as you step outside of the storage piece itself, data protection, making sure that you have the right data in the right location, that it's actually my data and not shared with everybody in the world. These are techniques that are going to get a lot of discussion topics. Sometimes it's called sovereign data.
Sometimes that still falls under the banner of security, but I think it's much more interesting than just those traditional discussion points.
Yeah, I am really interested to see heterogeneous computing, heterogeneous places to manage and deploy and activate agents seems to be a big conversation. Heck, in five months, we went from token maxing to token efficiency. I think in the end, the compute and everything follows the data. Where the data is typically created, all things equal, or where it makes more sense in the cloud, where the benefit of having it distributed is outweighed by the simplicity of the cloud. Typically, that is where it ends up. Compute always follows the data. Dave, it was great to have you on the show. I really appreciate your time. It is incredible the impact that Seagate is making to the AI deployment, and the entire benefit they can provide.
If there is nothing else that my firm has learned from talking to a multitude of CIOs, CEOs, and even all the hyperscalers, the best results come from having the right data in the right place, secured in the right place, and taking unstructured data, which quite frankly was ignored, deleted, archived, until you never look at it or you have a lawsuit against you, is finally adding value to that result. It is great to see.
Yeah, we are very excited. These trends are happening in our favor, and we know there is a lot riding on our shoulders. We have to keep our technology going forward so that we can keep gaining efficiency so that everyone can take advantage of it.
I would like to thank everybody for joining us for this AI Infrastructure portion of The Six Five Summit 2026. Hit that Subscribe button, be part of our community, and check out all of the other data related videos and conversations that we are having here at The Six Five. Take care.