The Gap Between AI Ambition and AI Readiness

There is no shortage of ambition when it comes to AI. It shows up in every boardroom conversation, every strategy document, every budget cycle where AI is no longer a novelty project but a line item with real expectations attached to it. Yet, very few organizations actually execute AI in a consistent, repeatable way that’s tied to reliable business outcomes.

The problem with readiness is that we tend to treat it like a milestone: something you reach and then move on from. In reality, it only proves itself when systems are asked to operate continuously, when data has to move, when models have to be fed, when governance has to hold, and when everything that looked clean and organized in a pilot suddenly has to function at scale.

As an industry, we keep saying organizations are “getting ready for AI,” but a closer look suggests that the same data preparation conversations are happening now that we had a decade ago — just with better tooling, more compute, and significantly higher expectations.

AI Readiness is the Elephant in the Room

When we talk about “AI readiness,” we’re not all talking about the same thing. In most cases, it’s become a catch-all phrase that gets thrown around in marketing decks and strategy sessions — something that sounds like a milestone you can reach if you just buy the right platform or move enough data into the right place. But that’s not how it plays out in real environments. What gets labeled as readiness is often just preparation. And even that’s being generous.

Remember “Big Data?” Variety, velocity, and volume were the big three V’s we talked about, and the industry spent years building systems to handle all three: more data, faster data, more types of data. We’ve carried some of those same behaviors forward, because organizations still point to the volume of data they have as proof that they’re ready.

As we’ve learned, though, none of that matters if the data itself can’t be trusted or used. Data is fragmented, inconsistent, and in many cases not structured or governed in a way that makes it usable — with 40% of organizations lacking trust in their own AI model inputs and outputs, according to Forbes. So instead of enabling AI, fragmented data introduces risk: if the data is wrong, AI doesn’t fix it, it just scales it. Financial services organizations know this problem better than most other industries, with data quality and fragmentation cited as a top barrier toward AI deployment.

Organizations equate buying a platform with being ready to use it because procurement is easy. Changing how people work, how decisions are made, and how data is governed is where things actually get hard, and that’s where most of the work still needs to happen. If the organization isn’t aligned around what it’s trying to do and why, the technology doesn’t matter.

So when we say “AI readiness,” what are we actually talking about?

If readiness just means you’ve ingested data, deployed a platform, and stood up a pilot, then the bar is far too low. And, it explains why so many initiatives stall the moment they try to move beyond that phase.

Pipelines, Data, and Platforms, Oh My!

Pipelines are built, data is ingested, and platforms are deployed. On paper, everything looks right. It always does when it’s sitting on a whiteboard. But when it’s time to move beyond the pilot and execute, things begin to break or stall out in ways that are hard to diagnose because nothing is technically “failing.”

That’s what a lot of AI efforts look like today, and the research continues to back this up in ways that can no longer be ignored. Across financial services, object storage is already embedded in AI and analytics strategies, with 43% of organizations placing it at the center of what they’re doing, according to research conducted by Hitachi Vantara and FSTech. On the surface this sounds like progress, but under the hood more than half of these organizations are still applying strategies selectively, supporting individual workloads rather than operating as a unified data foundation. The pieces are there, but they’re not working together in a way that supports real execution.

At the same time, 65% of organizations still cite cost of storage platforms as the primary factor in how they’re selected, which made perfect sense when storage was primarily about retention and scale. Now, it’s a problem when those same platforms are expected to support active AI workloads, because accessibility, integration, metadata, and governance start to carry as much weight as capacity, if not more.

The Storage Readiness Pitfall

The distance between ambition and readiness isn’t simply a single failure point. It comes from the accumulation of decisions that made sense individually but don’t hold up when the system is asked to operate as a whole. Financially, organizations invest in tools and platforms that promise acceleration, but underinvest in the underlying data foundation that those tools depend on. It’s possible that many organizations are measuring the wrong forms of success as well, with leaders at Harvard Business School recommending measurements of gradual AI adoption as a signal of success, rather than expecting change overnight. Here’s where the issue can appear:

• Architecture: Hybrid environments have become the default, which reflects the realities of financial services where data sovereignty, regulatory control, and scalability all have to coexist. But, those same environments introduce complexity around data movement, latency, and consistency that is easy to underestimate until it becomes the thing slowing everything down.
• Operations: Data exists in abundance but is often fragmented, duplicated, or difficult to access in a way that supports reuse, which leads to pipelines being rebuilt for each new use case instead of serving as part of a continuous system.
• Governance: Controls are often layered on after the fact, which protects data but can also make it harder to use, especially when policies are inconsistent across environments or require manual intervention to navigate.
• Organization: Ownership is distributed across teams that are not always aligned, which means the system itself reflects that fragmentation, even when the intention is to create a cohesive, collaborative solution.

At this point, storage — which is almost never the first thing anyone looks at when AI initiatives stall — starts to become part of the problem, because the job it was designed for is not the job it’s now being asked to perform.

Most object storage environments in financial services were built around cost efficiency and long-term retention, which made them incredibly effective at holding large volumes of unstructured data in a durable and scalable way. But AI changes the expectations entirely, requiring data to be accessible, discoverable, governed, and integrated into workflows that are running continuously.

When that doesn’t happen, you end up with data that is technically available but practically unusable, stored in a way that satisfies compliance requirements but creates barriers to access, organized in a way that reflects where it came from rather than how it needs to be used.

In hybrid environments, this becomes even more pronounced, because data is distributed across locations with different performance characteristics, different governance models, and different access patterns, which means every interaction with that data carries a small amount of overhead, and those small amounts add up quickly when you’re trying to operate at scale.

The organizations that are starting to close this gap aren’t necessarily doing something revolutionary, but they are making a shift in how they think about data and storage, moving away from the idea that storage is simply where data lives and toward the understanding that it is part of how data moves and how it gets used.

That shows up in environments where metadata is treated as a first-class capability rather than an afterthought, where governance is embedded into the data layer instead of applied externally, where data can be accessed and reused without being copied multiple times, and where systems are designed to support continuous execution rather than isolated pipelines.

The Hitachi Vantara Perspective

AI doesn’t fail because organizations lack ideas or use cases or even technology. It fails when the system underneath it can’t support execution in a way that is consistent, scalable, and sustainable. Until that changes, the gap between AI ambition and AI readiness is going to remain exactly where it is — not because organizations aren’t trying, but because they’re not as ready as they think they are.

Through hybrid-cloud architectures, modern resilience frameworks, and AI-ready data systems, financial institutions can gain a data foundation through Hitachi Vantara that supports operational certainty today and scalable innovation tomorrow. With VSP One Object, storage that is not only resilient and scalable, but intelligent, composable, and ready to support modern data strategies.

Learn more about how financial institutions unlock more value from data with object storage in a new report by FSTech.