Following NetApp’s announcements during INSIGHT 2024 in Las Vegas, it’s clear that enterprises want to maximise the value of their data as quickly as possible. However, many customers are still facing challenges with their siloed and unstructured data, particularly when it comes to AI.
According to George Kurian, CEO of NetApp, there is a huge gap between AI systems and data systems for many of their customers.
“AI systems are being built, often in silos, with specialised chips and specialised networking. What we have seen from many clients we’ve met over the last year is that those AI systems don’t have access to data. They are sitting in a silo with no data, and without data, the systems are useless,” he said during a media conference.
Kurian compared this situation to the early days of cloud computing, where cloud and on-premises systems were separate silos that needed to be bridged.
AI acceleration
From recognising handwriting, speech, image, and objects, AI has made significant advances in recent years where it is nearly capable of understanding the domain in which it operates without human input, Kurian noted in a separate keynote address.
“The pace of improvement is absolutely profound. Today, you can build rich, immersive experiences by combining multimodal capabilities to understand all of the data within your enterprise; not just the transactional data, but all of the unstructured data, which typically comprises 85% to 90% of your enterprise. The conversations within your teams, discussions with your employees, design blueprints locked in various documents, underwriting and risk policies — all of this data can now be structured and analysed using available tools. We are now in the era of data and intelligence,” he remarked.
Despite these developments, Kurian emphasised that the importance of data has not diminished. Data has become even more critical because it serves as the foundation for intelligence.
Currently, the enterprise world is in the third stage of modern data collection and analysis. The first stage involved digitising data production, the second involved maintaining a historical record for a single business process, and the third stage focuses on unifying all data to derive deeper insights.
To succeed in this era of data intelligence, Kurian outlined four key strategies:
- Having a solid data strategy and organisation.
- Having deep domain knowledge and insight into the data necessary for the business.
- Having the ability to test, learn, and adapt quickly to move technologies and experiments from lab to production.
- Having a data ecosystem alongside a business ecosystem, offering a richer perspective on the data.
As enterprises move from siloed to unified data, data management has often emerged as a major hurdle to seamless migration.
“If you look at the applications that clients are trying to drive, whether it’s advanced analytics or AI applications of various types, the organisations that are further ahead are those that have cleaned their data, catalogued it well, maintained a good understanding of security and controls for sensitive data, and have a strong handle on how data changes throughout its lifecycle,” Kurian said.
Kurian also observed that regulated industries tend to adopt AI techniques faster than unregulated ones. For example, in life sciences, there is a responsibility to maintain good, clean, and high-quality data that’s catalogued with the right clinical data codes or treatment codes.
Data innovation
To address its customers’ data management challenges, NetApp developed a data fabric to unify on-prem and cloud environments, so that AI is more accessible across various data locations. According to Kurian, the company is tackling this in three key ways:
- First, by bringing intelligence to both the infrastructure and the data stored within it. With a set of tools, customers can identify and explore their data assets, quickly selecting the data they need for AI experiments.
- Second, by innovating in infrastructure to support AI use cases, particularly in areas such as model traceability, data versioning, and the ability to move data from infrastructure to applications.
- Third, by allowing customers to maintain security and privacy controls over the lifecycle of their data.
Acknowledging the risks associated with AI, Kurian announced several measures the company is implementing to address issues like bias and data breaches.
“We are focused on ensuring the explainability of the data side, while others in the industry are working on model explainability. For example, we give customers the ability to clearly trace which data set was used with which model, so they can explain how certain conclusions were reached. If the data set and outcomes change, customers can easily understand what exactly in the data set changed,” he explained.
Regarding AI-enabled cyberthreats, Kurian said NetApp is leveraging machine learning techniques to protect customers’ data.
“We are using AI to enhance our own cyber resilience, and in areas like privacy and governance, so that when a consumer doesn’t want their data included in an AI model, it can be expunged,” he added.
APAC focus
In Asia-Pacific, data and AI use cases vary widely across industries, with business demands and regulatory landscapes shaping how companies manage their data.
For example, the needs of the financial sector in the region differ significantly from those of the telecommunications sector, noted Andrew Sotiropoulos, Senior Vice President and General Manager for Asia-Pacific, NetApp.
“The financial services sector is a great innovator of technology because of competitive pressure. You have global banks and financial services institutions entering APAC, so the local banking sector has to be innovative and dynamic to compete effectively. Then you have the telco sector, which faces different dynamics, focusing more on automation and efficiency due to cost pressures within their industry,” he said.
Meanwhile, with many countries and governments focusing on developing cybersecurity skills, NetApp sees an opportunity to assist in building cyber resilience.
“We have technology that can help in that environment. At the end of the day, customers themselves realise they must build cyber walls around their data and environments to protect themselves, and they are working with campuses and technology companies to develop those skills,” he concluded.