Why 2025 will redefine data infrastructure: 11 expert predictions

If 2023 is a world of generative chatbots and search driven by artificial intelligence, Introducing agent AI in 2024 — Tools that enable planning and execution of multi-step operations across digital environments. from german Engineering breakthrough from Microsoft’s early trials Co-pilot’s viewthe innovations are many and varied, but one constant remains: the need to keep data infrastructure organized and reliable.

As businesses lean toward advanced AI initiatives, several trends are reshaping how data is managed, protected, and used. More and more companies are adopting partly cloudy, open dataand an open governance strategy to avoid vendor lock-in and gain greater flexibility. They also focus on unstructured datatransforming the data marketplace into a hub for providing pre-trained AI models with proprietary datasets and applications. At the same time, advances in vector and graphics libraries have added new possibilities and laid the foundation for the next step.

Now, as the AI ​​story continues to unfold, industry leaders share their predictions for how the data infrastructure that supports AI will evolve in 2025.

1. Instant multimodal data will power the smart data flywheel

“By 2025, enterprises will fully embrace multimodal data and artificial intelligence to transform the way they operate and deliver[ing] value. At the heart of this transformation is the “intelligent data flywheel”—a dynamic cycle in which real-time data powers AI-driven insights, driving continuous innovation and improvement. Today’s dark data (imagery, video, audio and sensor output) will be at the core of unlocking clearer predictions, smarter automation and instant adaptability, ultimately leading to a richer, more nuanced understanding of business reality.

“With an instant data flywheel, AI will automatically diagnose problems, optimize processes, and generate innovative solutions. Enterprises will rely on AI agents to ensure Data qualityuncover insights and develop strategies that allow talent to focus on higher-level tasks. This will redefine efficiency, accelerate innovation, and transform businesses into more dynamic and intelligent organizations.

– Yasmeen Ahmad, General Manager Data, Analytics & AI Strategy and Outbound Product Management, Google Cloud

2. Cooling factors: Liquid-cooled data centers

“As AI workloads continue to drive growth, pioneering organizations will turn to liquid cooling to maximize performance and energy efficiency. Hyperscale cloud providers and large enterprises will lead the way, accommodating hundreds of thousands of AI accelerators, Liquid cooling is used in new artificial intelligence data centers for networking and software.

“Companies will increasingly choose to deploy AI infrastructure in colocation facilities rather than building it themselves – in part to reduce the financial burden of designing, deploying and operating smart manufacturing at scale. Alternatively, they will rent as needed Capacity. These deployments will help enterprises take advantage of the latest infrastructure without having to install and operate it themselves. This shift will accelerate the industry’s wider adoption of liquid cooling as a mainstream solution for AI data centers.

– Charlie Boyle, Vice President, Nvidia DGX Platform

3. Global data explosion creates storage shortage

“The world is creating data in unprecedented quantities. In 2028, Up to 400 zettabytes It is expected to generate a compound annual growth rate (CAGR) of 24%. However, the storage installed base is expected to grow at a CAGR of 17% — so [growing] Its growth rate is significantly slower than the growth rate of the data generated. It takes a full year to build a hard drive. This difference in growth rates will disrupt the balance of global storage supply and demand. As organizations become less experimental and more strategic in their use of AI, they will need to establish larger physical data center space and capacity plans to ensure storage provision and fully realize AI and data Monetizing infrastructure investments while balancing financial, regulatory and environmental concerns”.

– BS Teh, Executive Vice President and Commercial Officer, Seagate Technology

4. AI factory will evolve to PaaS

“By 2025, AI factories will evolve beyond the initial stage of providing infrastructure as a service, providing computing, network and storage services, to providing platform as a service capabilities. While basic services are critical to driving the adoption of artificial intelligence, The next wave of AI factories will prioritize platforms that drive data affinity and deliver lasting value. This shift is critical to making AI factories sustainable and competitive in the long term.

– Rajan Goyal, Co-Founder and CEO, DataPelago

5. Companies will use their massive data sets but require reliability

“For the most part, early applications of artificial intelligence simply used basic models trained on large amounts of public data. As complex RAG applications become mainstream and products that generate structured data rapidly mature, applications that leverage large amounts of private enterprise data Programs will start to create real value. But the bar for these applications will be high: Businesses will demand reliability from AI applications, not just great presentations.

“In addition, AI companies that provide these models must work well with publishers and content providers to secure the future of AI development. They need to enter into licensing agreements with content providers to ensure that they are recognized for the extremely valuable content they provide. be compensated for the data. This must happen quickly to avoid triggering a series of lawsuits and blocking artificial intelligence crawlers.

–Sridhar Ramaswamy, CEO, Snowflake

6. Corporate proxies will eat up communications data

“By 2025, enterprises will mine terabytes of communication data, such as emails, Slack messages, and Zoom recordings, using agents that provide analytical insights, dashboards, and actionable decision support tools.

“This will drive significant productivity improvements across industries.”

– Nikolaos Vasiloglou, Vice President of Research and Machine Learning at RelationalAI

7. Data governance and quality will be the biggest barriers to successful and ethical adoption of artificial intelligence

“By 2025, data governance, accuracy, and privacy will be the biggest barriers to effective AI adoption. As organizations look to scale AI, they will realize that successful AI outcomes rely entirely on trusted data. Management and preparation Large amounts of data, ensuring compliance and maintaining accuracy will create complex challenges, and companies will need to invest to overcome these obstacles. Basic data platform Achieve unified management across different data sources.

“As a result, we will see an increased emphasis on data management roles and governance frameworks aligned with AI initiatives, as businesses recognize that unreliable data directly impacts the effectiveness of AI.”

– Jeremy Kelway, Vice President of Analytics, Data and AI Engineering at EDB

“By 2025, a unified data observability platform will become an essential tool for large enterprises to provide comprehensive visibility into data infrastructure performance, quality, pipeline health, cost management and user behavior to address complex governance and integration challenges.” Through automated anomaly detection and real-time insights, these platforms will support data reliability and simplify compliance efforts across industries.

——Ashwin Rajeeva, co-founder and chief technology officer of Acceldata

9. All hail the sovereign cloud

“By 2025, we’re going to see a real push for sovereign cloud and private cloud. We’re already seeing the largest hyperscalers investing billions of dollars building data centers around the world to provide these capabilities. This… capacity is going to take a while can come online; at the same time, driven by a wave of legislation, mainly from the EU, demand will surge. Those with flexible, scalable and elastic cloud infrastructure will be able to quickly adopt sovereign or private approaches. of people will fall behind the curve.

– Kevin Cochrane, Chief Marketing Officer, Vultr

10. Rise Edge data processing

“I’m looking closely at the potential expansion of edge computing driven by 5G adoption, which brings data processing closer to the source and reduces latency. This could help democratize artificial intelligence. The question is, can we build systems that run on mobile devices? Efficient AI applications that may not rely on cloud resources?

“If technicians in the field have access to 5G, they can leverage artificial intelligence to assist in their work—whether it’s medical professionals providing diagnosis and treatment in a disaster area where there is 5G but no Wi-Fi, or engineers and scientists conducting diagnostics in the field and treatment.

– Jerod Johnson, senior technical evangelist at CData

11. The protection of unstructured data will become more urgent

“Traditionally, data protection has focused primarily on mission-critical data because these data require faster recovery. However, the situation has changed, and unstructured data continues to grow, accounting for 90% of all data generated in the past 10 years. The large surface area of ​​unstructured data, as well as its widespread use and rapid growth, make it highly vulnerable to ransomware attacks. Cybercriminals can use unstructured data as Trojans to infect enterprises cost-effectively. Ransomware compromise will become a key defense strategy, starting with moving cold, inactive data to immutable object storage that cannot be modified.

“To this end, IT and storage leaders will look to unstructured data management solutions that provide automated capabilities to protect, segment and audit the use of sensitive data and internal data in artificial intelligence – a use case that will become increasingly important as artificial intelligence matures. will inevitably expand. Additionally, they need to create systematic ways for users to search enterprise data stores, manage the correct data, inspect sensitive data, and transfer data to artificial intelligence through audit reports.

——Krishna Subramanian, co-founder of Komprise

All in all, by 2025 there will be significant advancements in enterprise data infrastructure, from multimodal data flywheels to sovereign clouds. However, challenges such as data governance and storage shortages will continue to exist. Success in this dynamic space will depend on balancing innovation with trust and sustainability, turning data into a lasting competitive advantage.