APAC enterprises face data sovereignty fragmentation, IT talent shortages, and sustainability mandates. Enterprise storage is now pivotal to AI success, reshaped by engines like IBM’s fifth-generation FlashCore module that autonomously handles deduplication, encryption, and compression.
Yet CIOs must secure data end-to-end, from ransomware recovery to quantum-safe archival, procuring storage that adapts, protects, and optimises relentlessly for the AI era.
Barry Whyte, principal storage specialist and master inventor at IBM, talks to us about the forgotten technology that is storage, which is core to the continuing development and use of AI in the enterprise.
Critical pain points
APAC enterprises must navigate pain points as AI transforms storage from passive repository to active computational engine.
According to Whyte, one of the most important ones involves cyber resiliency. He argued that while cyber resiliency and cybersecurity are the security team’s job, storage also plays a huge part in helping an organisation’s security posture by helping protect and recover critical data.
AI is also driving demand for computing resources, including RAM, CPUs, flash storage, and hard drives. This increased demand has pushed up hardware prices (“ramageddon”), making IT infrastructure more expensive for businesses.
“When I think the third thing here is around how AI is influencing the general operation within the businesses themselves? Where can I apply AI to make my teams more efficient? How can I free up more time for my teams to be doing something more interesting or looking at those next-generation technologies and so on?”, Whyte said.
Data sovereignty regulations, skills shortages, and sustainability mandates also compound these challenges across the data lifecycle.
Whyte said that organisations must consider not only where data is physically stored but also who owns the cloud or data centre, as ownership can determine which country’s laws apply.
He added that global regulations, such as GDPR, mean businesses may need to comply with foreign data protection laws, making data management more complex and costly.
As a result, companies must rethink their cloud strategies and data governance practices rather than assuming the cloud is the default solution.
“Something that people are really starting to realise is that cloud isn’t the answer to everything. Cloud is essentially a capability, not a destination,” he said.
“Cloud is essentially a capability, not a destination.” Barry Whyte.
AI reshapes storage
AI is also fundamentally changing what enterprise storage is capable of doing. Rather than acting as passive repositories, modern storage platforms are becoming intelligent systems that optimise performance, strengthen cyber resilience and even participate in AI workloads themselves.
Whyte believes storage should no longer be viewed as passive infrastructure. Instead, intelligence is moving directly into storage itself through technologies such as IBM’s FlashCore Modules.
Described by Whyte as “having a brain inside the drive”, it can detect ransomware in seconds, identify unusual data access patterns, and detect data exfiltration.
Whyte shared that the company also introduced a natural language processor built into the user interface, a solution that allows users to converse with it in their own language using WatsonX.AI natural language processing in the background.
“The idea is all about giving you back that time. It’s making those things that just take time and that you have to run through, trying to augment them and give you essentially another member of the team. And that’s just the start of it,” Whyte said.
For Whyte, autonomous AI agents deliver the most value by automating storage lifecycle management and performance optimisation.
AI can eliminate the disruption associated with hardware refreshes by enabling seamless migration of data and configurations between systems.
“How can we make this much simpler? You’ve got all the definitions, and you understand; you’ve got the data, and you’ve got all the logical objects that allow you to access that data on the existing system. What if we could clone all that straight away onto the new system, move the data across, and then there’s no disruption whatsoever,” he said.
It also continuously monitors workloads, detects performance anomalies, and automatically redistributes workloads to improve efficiency without manual intervention.
Adapting to unpredictable generative and agentic workloads
As storage becomes more intelligent, CIOs face another challenge: ensuring infrastructure can keep pace with unpredictable AI workloads without continually expanding physical data centres.
Whyte noted thatfootprint, space and capacity are going to be the main problems.
“Giventhat we can now store technically up to almost half a petabyte in a single drive, the latest generation of these flash core modules has that computational capability in there,” he explained.
He added: “They’re doing compression and deduplication on the drive itself. What a just-over-a-hundred-terabyte physical drive can address is almost six times the capability. If you’re getting good data reduction, now you can be fitting up to about half a petabyte in a single drive. That clearly is a differentiator and can help.”
By compressing and removing duplicate data directly on the storage drive, organisations can store significantly more data in the same physical space and better support growing AI workloads without constantly expanding their data centres.
APEX CIOs must also prioritise use cases that address region-specific constraints around storage, such as limited spaces for data centres and ESG carbon neutrality deadlines.
“Data classification is a critical piece to modern management of data,” Whyte emphasised.
He said that organisations must know what data they have to be able to make critical decisions.
“As data growth rates just continue to skyrocket, we have to get much better at classifying and saying what is the importance of this data and making sure it’s living in the right place,” he said.
“Data classification is a critical piece to modern management of data.” Barry Whyte.
Preparing for post-quantum cryptography
While AI is changing today’s storage requirements, Whyte believes CIOs must also prepare for tomorrow’s security threats. One of the most pressing is quantum computing, which could eventually render today’s encryption ineffective.
“Certainly you want to be using state-of-the-art now,” said Whyte. “Who knows in 20 years if quantum computers are going to become that commercially available, that everyone has them and everyone could do that.”
He emphasised the importance of understanding the access that people have in the environment and data security to stop people from harvesting data.
“Do the best you can as we can today with these algorithms that are being classed as post-quantum safe,” he said.
“Do the best you can as we can today with these algorithms that are being classed as post-quantum safe.” Barry Whyte
Optimising the AI data lifecycle
Technology alone, however, is only part of the equation. Whyte argues that organisations cannot optimise AI unless they first understand the value of the data they already own.
Once data is categorised by value and purpose, organisations can determine which workloads benefit from the cloud’s flexibility and which should remain on-premises for better performance, security, and cost efficiency.
“Without being able to classify what you have and being able to understand the importance of each of those different types of classified data that you have, it’s very difficult to be able to do that separation,” he said.
Whyte warns technology leaders against a “lift and shift type environment” where organisations just move everything into a cloud environment and then realise how expensive it is.
“Some of that data should be there probably, but a big chunk of it probably is going to be cheaper, and it’s going to be more secure if you hold onto it yourself,” he said.
Architecting storage to power AI agility
As AI becomes central to business operations, storage must also become intelligent. It must aid organisations in classifying data, automating management, strengthening cyber resilience and preparing for future threats.
CIOs must learn how to build storage architectures that can continuously adapt and evolve with technology.
Click on the PodChats player as Barry Whyte explains how storage remains a core component of developing and using AI in the enterprise:
- What are the three most critical pain points facing APAC enterprises as AI transforms storage from passive repository to active computational engine?
- How do you see data sovereignty regulations, skills shortages, and sustainability mandates compound these challenges across the data lifecycle?
- Apparently, AI is fundamentally reshaping storage technology development. We spoke about that earlier on. This includes computational offloading all the way to autonomous performance tuning.
- Where do autonomous AI agents deliver maximum value in the storage lifecycle? And how can APEX CIOs prioritise use cases that address region-specific constraints like space-limited data centres and carbon neutrality deadlines, for example?
- What does this mean for CIOs architecting infrastructure that must adapt to unpredictable generative and agentic workloads?
- We started to hear a lot more interest around use cases. How can APEX CIOs prioritise use cases that address region-specific constraints? In this case, the limited spaces for data centres. And the other thing is ESG carbon neutrality deadlines.
- As post-quantum cryptography transitions from decades away to plan now, how should APAC financial and government sectors rethink storage security architecture to protect AI training data and models across their entire lifecycle against things like harvest now, decrypt later threats?
- Given that AI is compressing hardware refresh cycles while demanding greater capital efficiency, how should APAC CIOs and storage architects balance cloud-adjacent consumption models with on-prem computational storage investments so they can optimise across the AI data lifecycle?
- What is your recommendation for CIOs architecting their storage to power AI agility?
