How Should IT Manage Data in the AI Age?

The recent news cycles around DeepSeek exploded the conversations—and stock market gyrations—of AI again. Yet the reality is far from this hype. Over and again in surveys and polls, enterprises are not jumping into AI yet. However, we do know that workers in all industries are using GenAI to boost their productivity. So the question of how to manage data in the AI age could not be more relevant in 2025. Whether executives and IT directors like it or not, their company is already doing AI.

So, for IT leaders, CISOs, risk and compliance directors, there is ample concern about what risks employees are undertaking with corporate data. Preparing for AI is the top business challenge for unstructured data management, according to the Komprise 2024 State of Unstructured Data Management. Yet it is important to think about the opportunity. Nobody wants to be left behind when the time is right to invest significantly in AI. It begins with the data infrastructure, and the requisite technologies, policies and training to keep data safe and ensure data quality and performance when feeding corporate data to AI.

In this latest edition of the Komprise Intelligent Data Management newsletter, we share some of the recent analysis and coverage about data infrastructure trends as relates to AI. Learn more about Komprise, a SaaS solution for unstructured data management and mobility here and follow us on LinkedIn.??

Storage IT in 2025

First though, let’s look at the industry at large. In this Blocks & Files article, Chris Mellor gives an updated overview of the storage technology and data management industry. “Right now, the two dominant storage demand and development drivers are cyber-resilience and security on the one hand, and AI training and inference on the other. The IT storage picture at the start of 2025 is vibrant and healthy with developments ongoing at all levels of the storage stack, driven by demands for more memory, more and larger SSDs, higher-capacity disk drives, better data serving to GPUs, and preparation for AI-driven analysis. We also see improved data protection and security and the ongoing promotion of object storage to higher-performing access for AI training and inference.”

?

The data center boondoggle

How the global economy is going to fuel AI workloads has been a topic of continuing debate, given the dearth in data center space as well as the difficulty in finding adequate natural resources to power these massive data centers. NAND Research founder and Steve McDowell writes: ?“OpenAI,?SoftBank, and?Oracle?are?teaming up?on a massive?$50 billion data center project?(called Stargate) to create the infrastructure needed for the next wave of AI. With Oracle’s cloud expertise and SoftBank’s investment muscle, this project could redefine the AI compute landscape. Or it could just be another big CapEx boondoggle (in which both Oracle and?Nvidia?will be the big winners). Time will tell.” ?Another variable here is the longer-term impact of DeepSeek’s compute-efficient AI model, which may lend secrets for the industry at large.

?

Breaking down the IT infrastructure requirements for AI

In this InfoWeek article, freelance writer Richard Pallady covers the storage, networking, chips and compute requirements for enterprise AI. ?“Organizations looking to optimize their operations through AI need to figure out where to store that data securely while still allowing machine learning algorithms to access and utilize it.?Hard disk drives or flash-based solid-state drive arrays may be sufficient for some projects.?Organizations that rely on larger amounts of data may need non-volatile memory express (NVMe)-based storage arrays. These systems are primed to communicate with CPUs and channel the data into the AI program where it can be analyzed and deployed.”

?

The data management connection

But wait: only the very largest, well-heeled organizations will create and host a massive, optimized infrastructure for AI. The reason being: it’s astronomically expensive to do so. Instead, most organizations will use AI services in the cloud and will not be creating their own AI solutions or models but using and modifying the AI models already in market.

Krishna Subramanian , COO of Komprise, explains in Techopedia:

“The key challenge isn’t just where data is stored but how it’s orchestrated across distributed environments. Ninety percent of data is unstructured. So, tying unstructured data to AI with proper data governance and security is key for a business. Ninety-five percent of the market uses a trained model, and the cost of model training will keep dropping over time. So yes, performance is important, but even more important is how your corporate data connects to AI because AI has already been trained on all the data in the public domain.”

Optimizing unstructured data for AI

When it comes to preparing data—especially unstructured data—for AI, there are several considerations, as Subramanian shares in StorageNewsletter. “First, AI data governance is aided by automated capabilities to protect, segment and audit sensitive and internal data use in AI with the goal of protecting data from breaches or misuse, maintaining compliance with industry regulations, managing biases in data, and ensuring that AI does not lead to false, misleading or libelous results. Once this is in place, IT will need to create systematic, automated ways for users to search across corporate data stores, curate the right data, check for sensitive data and move data to AI with audit reporting.

Finally, most IT organizations will need to do all of the above and more without dedicated AI budgets, another finding from the Komprise survey. Cloud-based, turnkey AI services will grow in popularity along with no-code or low-code AI platforms which don’t require extensive coding knowledge. As well, continually analyzing and right-placing unstructured data into the most cost-effective storage will free up funds for AI.”

Cleaning up and protecting data for AI

“Before leveraging AI for decision-making, you need a clear understanding of the data available across your organization,” Subramanian writes in BusinessCloud. “This includes identifying the types of unstructured data – such as text files, images, videos, and more – and ensuring they are easily accessible and well-organized across a hybrid cloud environment. Data cleanup is the first priority: remove irrelevant or redundant information to reduce storage costs and security risks, particularly exposure to ransomware. Implementing a data indexing system will make it easier to search and apply AI effectively.”

Komprise and data management for AI?

VMBlog’s David Marshall wrote about Komprise and our vision for unstructured data management and AI: “They're pushing for a fundamental shift - moving from basic storage administration to strategic, analytics-driven data management, and from data being "locked away" to making it "securely available" whenever and wherever it's needed. The real game-changer is what Goswami calls their "crown jewel" - the Global File Index. This isn't just another metadata database. Instead, it creates a comprehensive view of your data ecosystem, storing detailed information about every file: who created it, when it was last accessed, what type of data it contains, and more.”

Last Words?

What your organization needs to implement safe, effective and business-aligned AI programs goes beyond the IT infrastructure required to secure and process data and host the software. An unstructured data management infrastructure that allows users to quickly search, classify, organize and prepare data workflows for different use cases without exposing sensitive data is table stakes. You can subscribe to our blog to receive new posts in your inbox and check out what's new by visiting our Resource Center.??

What will it take for enterprises to place their bets on AI in production? We'd love to hear from you.

Comment on the post or send a note to:??

[email protected]??

?

?

?

?

?

?

?