Making data fast, secure, and permission-aware for AI is one of the biggest challenges: CEO, Nasuni

Enterprise IT teams racing to adopt gen AI are running into a familiar bottleneck: unstructured data spread across on-premises systems and multiple clouds, with inconsistent access controls and governance. In a conversation with TechCircle, Nasuni CEO Sam King discussed why she sees data fragmentation, not model access, as a core constraint on enterprise AI, and how buyers are prioritising cost, security, and recovery outcomes over storage architecture labels. 

Edited Excerpts: 

In practical terms, what does a unified file data platform category mean today—and where do other approaches fall short?

I don’t think traditional vendors or cloud-native file services have defined the problem too narrowly. I think the issue is that the solutions enterprises end up with are very fragmented and very expensive.

Right now, enterprises are trying to use AI more, and the constraint usually isn’t access to frontier models. Organisations can get that. One of the biggest challenges is that their data—the historical information that acts as a source of intelligence for the organisation—has to be made accessible to AI in a way that is fast, secure, and permission-aware, while still respecting the permissions and data governance decisions they’ve made elsewhere. That’s very difficult.

What I think a lot of folks are missing is that many storage solutions have produced even more fragmentation: something on a box here, something on a box there, some of it in the cloud. You don’t have a single pane of glass (a unified management view) across all of your unstructured data. I think you have to solve that first. It’s foundational before you can make use of AI, or even provide organisational insights to enterprise teams in general.

A lot of infrastructure vendors say they are AI-ready. What does AI readiness mean on your platform?

We think we help customers become AI-ready by virtue of what we do and what we’ve done for a long time. When customers are trying to deploy agentic solutions (AI systems that can take actions) or other AI tools, they want fast, secure, permission-aware access to datasets. They often struggle because the data is fragmented across the organisation.

By unifying it on the Nasuni platform, they’re able to provide access to their AI solutions. I don’t think about it as, “What is the story about Nasuni?” I think about what we enable for customers. We help them become AI-ready by getting their unstructured data foundation in place.

In India, the conversation has shifted from scarcity of data to data exists, but it isn’t clean enough. How do you see that debate?

I would say the challenge people describe with respect to data in India applies globally. We have 800-plus customers across the globe, and the issue—making sense of all the data that defines an enterprise’s intellectual property and the work products of engineers, physicians, and researchers—is not unique to India.

It’s a shared global problem that people have to get their arms around before they can truly get organisational and domain-specific insights from AI. At the same time, I think the focus in India on governed use of AI, secure use of AI, and equitable use of AI could put the Indian ecosystem in a position to solve these problems in a way others can emulate. But I don’t think the problem is singular to India.

There is a long road ahead, but it helps us get our arms around something we should have been getting our arms around to begin with. If AI is the motivation that helps people get to a better place with their unstructured data platforms, so be it.

Buyers hear file storage, object storage, backup, and data services converging into one-platform conversations. Are enterprises ready, or do they still want silos?

I’ll tell you what customers are ready for: they’re ready for their storage solution to not be as expensive as it is. They’re ready for storage to scale in a cost-effective way as their data grows. And with AI, data is growing even more exponentially than it was before. Customers are also ready to have an easy, reliable, secure solution.

I think our industry uses a lot of technical jargon and acronyms. Customers care about business outcomes: storing data cost-effectively; ensuring that as data needs grow, costs don’t grow exponentially; being able to recover from a data loss event—maybe ransomware; and providing fast access to teams and now to AI across the globe.

So rather than being hyper-focused on individual technology architectures, customers are focusing on the outcomes they’re trying to drive and evaluating which platform, versus any one technology, solves the business problem best.

Your company has Nasuni Data Service, described as a cloud-native API with read-only programmatic access to file data. How does that change your role in enterprise architecture, and what guardrails matter most?

In order to be a platform—which is how we think of ourselves, and how we think about our product evolution journey—we have to provide insights to customers about what’s happening with access to their files and what’s happening with the data they’re storing in the Nasuni platform. We also have to give them the ability to access that data for various purposes they might have.

What becomes really important is making sure that the governance structures and permission structures customers have adopted for their datasets in general are aligned with how we make Nasuni Data Service and other capabilities available. We build on top of that.

Many vendors market multi-cloud, but in practice, some customers standardise on one cloud and keep a second for leverage or compliance. What does real multi-cloud look like for file data infrastructure?

One attractive part of Nasuni’s value proposition is that we don’t care whether you want to use object storage with AWS, run on top of Azure Files, or use us inside Google Cloud. We also have a strong partnership with Microsoft.

Increasingly, customers are trying to diversify across cloud platforms. One reason is what you described: they don’t want to give too much leverage to any single cloud vendor. Another is that different workloads and applications may be better suited to one platform versus another.

From our perspective, we’re offering an infrastructure-level solution. Our platform becomes the unstructured data foundation for a company, and we want companies to be able to use us regardless of what cloud decisions they make for what’s best for their organisation. I do see companies wanting true multi-cloud and also hybrid cloud (mixing on-premises and cloud), including with the recent push around rebuilding data centres. For us, we don’t necessarily care where it’s happening—we want to run on top of all those pieces of infrastructure.

Across sectors like manufacturing, healthcare, and financial services, where are file data priorities diverging—and which areas are moving fastest?

We see a lot of uptake of AI-level solutions in pharmaceutical research and other heavy research-based organisations where there’s a lot of intellectual property collected over the years. There are huge datasets you can mine to inform future work.

More broadly, any organisation with a lot of work product that can form the basis of new projects is leaning into this. For example, some customers in manufacturing and in architecture, engineering, and construction have done hundreds or thousands of projects to build buildings, bridges, and so on. When they get a new project, they don’t want to redo all that work. We’re seeing them lean into AI and ask how they can direct AI agents and AI solutions—using the unstructured data we can make available—to speed up decision-making on new projects.

Looking ahead, do you expect unstructured data infrastructure to consolidate around a few platforms or fragment further into specialised tools—and where does defensibility come from?

Customers need and want standard platforms that are trusted, reliable, and offer multiple capabilities in one place. I don’t think customers want fragmentation to go further than it already has.

Our goal is to be an unstructured data platform that provides cost-effective, scalable storage, built-in ransomware protection and recoverability, and then gets your data AI-ready. Customers want to focus on the business problems they’re solving rather than stitching together lots of small, bit-by-bit technology components. What’s best for customers is that consolidation occurs, and you have platforms like Nasuni looking to keep expanding capabilities.