Maryanne Baines is a distinguished authority in cloud technology, possessing over twenty years of deep-seated expertise across semiconductors, storage, and AI infrastructure. Throughout her career at leading firms like Western Digital and Druva, she has specialized in evaluating tech stacks and product applications to help industries navigate the complexities of digital transformation. In this discussion, we explore the evolving landscape of AI-driven storage, focusing on the strategic necessity of data mobility, the elimination of restrictive fee structures, and how organizations are leveraging specialized infrastructure to scale beyond the limitations of traditional hyperscalers.
Having spent over two decades working across semiconductors, storage, and AI infrastructure at prominent firms like Druva and Western Digital, how have you seen the fundamental storage requirements for AI labs shift away from traditional cloud models?
The shift we are seeing is primarily driven by a scale of demand that is simply unprecedented in the history of data management. AI workloads are no longer just about capacity; they are about the ability to support data across the entire AI lifecycle, from initial training to real-time inference. Organizations managing hundreds of petabytes of AI data need a framework that doesn’t just store information but allows it to move fluidly to wherever the compute resources are located. We are moving toward a model where the predictability of costs is just as important as the performance of the hardware, ensuring that innovation isn’t stifled by unforeseen financial hurdles.
The industry recently saw a massive migration where an image-generation lab moved 175 petabytes off a hyperscaler within just a few months; what does this tell us about the current frustrations regarding egress fees and architectural lock-in?
This level of migration is a clear indicator that enterprises are reaching a breaking point with the “walled garden” approach of traditional hyperscalers. When you have a robotics data consortium projecting annual savings of more than JPY 100 million just by switching providers, it becomes a matter of fiscal responsibility rather than just a technical choice. These organizations are demanding the freedom to move their data wherever their workloads take them without being penalized by egress or API fees that dictate their architecture. By removing these friction points, companies can finally treat their data as a strategic asset rather than a liability that is too expensive to move.
With 16 global storage regions and a network of 18,000 channel partners, how does a specialized storage strategy allow organizations to scale more effectively than they could within a single hyperscaler ecosystem?
A specialized strategy provides a level of agility that a single hyperscaler simply cannot match because it isn’t trying to be everything to everyone. By operating 16 global regions, a provider can offer localized performance and data sovereignty while the 18,000-strong partner network ensures that support is available across the entire AI ecosystem. This approach allows frontier model labs and generative AI startups to avoid being tied to a single cloud provider’s roadmap, giving them the freedom to evolve their workloads as the technology changes. It’s about building an ecosystem that prioritizes the customer’s architectural freedom, allowing them to scale their operations globally without the fear of hitting a ceiling.
Given your background in driving over $10 billion in strategic M&A and investment outcomes, how important are high-value partnerships in reshaping the market for AI infrastructure?
High-value partnerships are the actual engine that moves markets, especially when the goal is to challenge long-standing industry monopolies. When a strategy is backed by a track record of $2 billion in partnership-driven revenue, it demonstrates that the market is hungry for collaborative solutions rather than isolated platforms. These strategic alliances allow for a more integrated approach to AI infrastructure, where storage, compute, and networking work in harmony rather than in competition. In this high-stakes environment, the ability to build and scale an ecosystem of freedom is what will ultimately define the next generation of cloud leaders.
What is your forecast for AI storage?
From 2026 to 2028, I expect we will see a total transformation where storage is no longer viewed as a static repository but as a dynamic, high-performance engine for global inference. The industry will move toward a standard where egress fees are eliminated entirely, as customers will no longer tolerate paying a premium to access their own intellectual property. We will see the emergence of a truly “borderless” data environment where 18,000 partners and dozens of global regions work together to ensure data is always inches away from the compute power it needs. Ultimately, the successful AI strategies of the future will be those that prioritize data mobility and cost transparency, allowing researchers to focus on breakthroughs rather than their cloud bill.
