Future Predictions for Digital Storage and Memory Capacities in 2025 (Part 4)
Future Predictions for Digital Storage and Memory Capacities in 2025 (Part 4)
This is my final blog post of 2025 on digital storage and memory projections. My first two entries explored digital storage and memory devices like magnetic tape, HDDs, SSDs, NAND, DRAM, and non-volatile emerging memories. The third looked into the revival of optical storage, with startups targeting archiving and digital preservation, anticipating prototype deliveries in 2025 and beyond. This piece focuses on advances in storage systems and software, their applications in processes, as well as developments in storage devices and systems for the future.
In 2024, the overall storage and memory markets experienced a resurgence, although partial recovery occurred for NAND flash and SSD markets, mainly in enterprise and data center sectors. According to the IDC report in September 2024, the worldwide enterprise external OEM storage systems could grow by about 4.3% on average over a five-year span.
The report cited two primary drivers for this growth - the escalating demand for flash storage, typically in all-flash arrays, for tasks related to AI both for training and inferencing, and the increasing demand for flash media across external storage systems and servers. Furthermore, IDC disclosed that IaaS solutions have pushed more organizations to reassess or plan to relocate workloads from shared clouds to dedicated IT environments.
Conversely, IDC projected an increase of 48.8% in 2024 cloud infrastructure spending (computing and storage) compared to 2023, with the boost largely stemming from the rising costs for GPU servers. Actual unit cloud growth saw a 17.7% increase in the same period.
The spent on shared cloud infrastructure is projected to grow by 57.9% Y/Y, whereas dedicated cloud infrastructure spending was forecasted to experience a 20.4% Y/Y uptick. By 2028, IDC predicts a compound growth rate of 18.1% for cloud infrastructure spending (compute and storage), with it accounting for 76.4% of total compute and storage infrastructure spending by 2028 and shared cloud infrastructure expenditure making up 78.6% of the entire cloud spending by 2028.
The rise in AI workloads will impact storage and memory requirements. Eric Herzog, CMO at Infinidat, claimed that enterprise storage infrastructure will play a new role as the basis for retrieval-augmented generation (RAG), a GenAI-centric framework for enhancing, refining, and optimizing AI models, including large language models and small language models. RAG eliminates the requirement for continuous model retraining, reducing the associated costs and frequency of AI hallucinations. Infinidat introduced RAG workflow deployment architecture in November 2024.
Infinidat is also taking steps to safeguard data stores from cyberattacks. Their data protection capabilities, at the first signs of an attack, generate an immutable snapshot of data to mitigate the damage. Infinidat also anticipates the growth of hybrid multi-cloud storage in 2025, which converges on-premises/private cloud and public cloud storage resources for high levels of flexibility, cost efficiency, and issue-specific solutions. AWS and Azure can facilitate such endeavors. Infinidat announced such services with AWS and Microsoft in 2023, using its InfuzeOS software-defined storage (SDS) solution. Furthermore, non-VM-based virtualization and Kubernetes/container deployments are expected to increase in 2025.
Steve Leeper, VP of Product Marketing at Datadobi, observed that, "The volume of unstructured data stored in both public and private cloud environments will persistently rise. Solutions that grant clients command over data irrespective of location will gain prominence as data accumulates in multiple environments."
Leeper emphasized that, given the proliferation of unstructured data, there is a greater need for data insights to prepare GENAI-ready data.
Don Boxley, CEO and co-founder of DH2i, pointed out that AI can create self-optimized high availability (HA) clusters, where AI eliminates inefficiencies by continually monitoring workloads and resource usage, enabling clusters to self-optimize and maintain peak performance without human intervention.
Boxley also noted that AI-driven HA clustering can maintain HA across various cloud environments by managing clusters that span multiple providers. AI simplifies cross-cloud HA functions by analyzing network traffic and distributing workloads intelligently across providers, ensuring smooth performance and responsiveness.
The growing AI data demand will increase archiving requirements. Gal Naor, CEO of StorONE, commented that the exponential growth of data in 2025 would significantly raise storage costs for organizations as they grapple with the challenge of retaining cold data for extended periods, despite infrequent access.
Naor also mentioned that "Auto Tiering storage solutions will dynamically migrate inactive data to low-cost drives while ensuring prompt access for future analysis, minimizing overall costs without compromising efficiency." The increasing digital threats necessitate rapid and cost-effective recovery methods. Intelligent architectures will store snapshots on affordable tiers while guaranteeing prompt recovery availability, enhancing both preparedness and cost management.
Skip Levens, Product Lead and AI Strategist for Media and Entertainment at Quantum, shared his thoughts on AI growth in 2025 and its impact on digital storage demand. He stated that, "In 2025, organizations employing a more practical approach towards AI and its underlying data infrastructure will be best positioned to generate new insights and foster discovery."
Levens also discussed the future winners, stating, "Those spearheading the data race are not only utilizing every shred of their accumulated data for distinguished AI outcomes but also have a system in place for effectively doing so—managing, categorizing, indexing, and cataloging every bit of data. They'll produce more, faster, and at a higher standard compared to their rivals. In 2025, it will become apparent who surges ahead in this new 'data and algorithm competition.'
Members of the Active Archive Alliance also contributed comments regarding the growth of archive data to support 2025 workflows. Rich Godomski, Head of Tape Evangelism with FujiFilm NA Corp., Data Storage Solutions, stated that, "Eco-friendly active archive solutions with advanced data management capabilities can leverage economical and ultra-efficient tiers of storage such as S3 compatible object-based tape libraries. This will be essential to balance the voracious energy consumption of innovative AI applications as the AI era progresses in 2025 and beyond."
Paul Luppino, Director of Global Digital Solutions at Iron Mountain, said that, "Artificial intelligence (AI) has the potential to transform data storage and active archives by increasing efficiency and accessibility. As data volumes escalate, we can optimize storage management by predicting usage patterns and minimizing expenses, potentially making decisions about how and where to store data at the point of creation."
In the realm of active archives, AI can analyze and prioritize data, ensuring frequently accessed information is readily available while less critical data is stored cost-effectively. Automated classification, tagging, and indexing could streamline the search process, allowing for intelligent data handling."
Mark Pastor with Platform Product Management at Western Digital said that, "Disaggregated storage...has proven to deliver the performance and capacity required to meet the demands of demanding GPU-related workloads, which are key components of AI and machine learning processes. Disaggregating storage from the server accomplishes two important things: (1) it enables storage to be shared across multiple servers, offering greater flexibility and utilization of storage resources, and (2) demonstrations show that disaggregated storage delivers the performance needed to keep GPU processing fully saturated."
Over time, these external storage architectures will become standard for HDD in active archives and flash for performance workloads and will ultimately migrate to fabric as opposed to SAS due to the benefits of fabrics.
Jason Lohrey, CEO of Arcitecta, also highlighted the value of fabric shared storage, saying, "Businesses can maximize their existing investments and avoid vendor lock-in by leveraging a data fabric—an architecture that unifies cloud, disk, tape, and flash storage into a single, logical namespace. This trend towards virtualization allows for a more flexible approach to data management, enabling businesses to mix and match technologies to meet specific needs."
Ted Oade, Director of Product Marketing at Spectra Logic, talked about how archive storage practices can contribute to more sustainable AI workloads and create competitive advantages, stating, "Modern tape storage is not only highly durable but also incredibly energy-efficient, especially when compared to disk storage. By offloading cold data to tape in an active archive, data centers can conserve energy for AI workloads, enhancing efficiency. As energy becomes a potential limiting factor for AI growth, businesses that embrace sustainable practices will gain a competitive edge in 2025 and beyond."
2025 is expected to see a surge in demand for storage devices, systems, and software to support the growth of AI data processing. AI will increasingly be utilized to improve digital storage efficiency and security. Digital storage and memory architectures may have a significant role in more sustainable AI data centers.
In the 2025 AI-driven landscape, Fujifilm introduces eco-friendly active archive solutions with advanced data management capabilities, leveraging S3 compatible object-based tape libraries to balance energy consumption between AI applications and economical storage tiers. (Fujifilm, S3, tape libraries)
The Data Protection capabilities of Infinidat create immutable snapshots at the first signs of a cyberattack, safeguarding data stores from potential damage and ensuring swift recovery. (Infinidat, data protection, cyberattacks, snapshots)
StorOne predicts that a surge in 2025 data storage demands will significantly increase costs for organizations as they retain cold data for extended periods, proposing Auto Tiering storage solutions to dynamically migrate inactive data to low-cost drives while maintaining prompt access for future analysis. (StorOne, data storage, cold data, Auto Tiering)
Spectra Logic emphasizes the role of modern tape storage in promoting sustainable AI data centers, observing its high durability, energy efficiency, and ability to offload cold data to tape in active archives to conserve energy for AI workloads, which could potentially provide a competitive edge. (Spectra Logic, tape storage, AI data centers, energy efficiency)
Datadobi anticipates the increasing importance of solutions that grant clients command over unstructured data across multiple environments, recognizing the need for efficient data handling as data accumulates in various locations. (Datadobi, unstructured data, data management, client command)