The global leader of BCG’s Cloud Advisory Business talks about why cloud transformation is more than just moving data to a new platform—it’s about revolutionizing not only all infrastructure but also the mindset of a business. Understanding the economics of cloud and having a comprehensive plan for exiting legacy systems are essential steps for any organization looking to succeed in a cloud-driven future.

Filippo Scognamiglio

Managing Director & Partner

New York

BCG: There are a lot of misconceptions about cloud. Can you talk about the key myths you’ve encountered?

Filippo Scognamiglio: Yes, there are a couple that stand out. First, people often believe that cloud operations are riskier than on-premises operations. However, the opposite is true. The reality is cloud is generally more reliable and secure than most on-prem environments, except for a very small number of unique setups. People tend to fixate on high-profile cloud security incidents, but similar problems happen on-prem, often with worse outcomes. We just don’t hear about those as much.

Another myth is that cloud prices are increasing. They’re actually not. Cloud prices have remained pretty flat. However, many clients come to me saying, “My cloud costs are out of control.” What they’re missing is that their expenses are growing because they’re ramping up cloud usage without a clear strategy to reduce legacy costs. So while the cloud line item in their budgets is becoming increasingly significant, the real issue is that they aren’t retiring their old systems fast enough. This imbalance creates the perception that cloud is getting more expensive, when it’s really a matter of planning, executing, and getting the other line items on the budget down.

When it comes to managing these cloud costs, what should companies be asking themselves in order to avoid this kind of imbalance?

Filippo Scognamiglio: Many companies are already thinking about cloud costs, but it’s also important to consider the full picture—looking at the end-to-end costs of both cloud and on-prem operations and how they evolve over time. A key question is How far and how fast should we move to cloud? Not everything will migrate, so understanding what a hybrid future looks like—and how to manage the “bubble costs” of such a transition—is essential.

How does generative AI fit into the cloud conversation?

Generative AI at scale will almost exclusively run on the cloud. More than 80% to 90% of corporations will train and run their AI models in cloud environments. But what’s often overlooked is the economic impact. For example, running GPT-4 on cloud can be 5 to 10 times more expensive than earlier models. Businesses will need to shift from focusing on having the best model—the latest shiny object—to having the one that’s actually “fit for purpose” and cost-efficient. The cloud will be the backbone of AI, but success hinges on balancing innovation with disciplined cost management.

What’s the right strategy for cloud adoption?

First and foremost, there is no such thing as a standalone “cloud strategy.” What companies need is an “infrastructure strategy” in which the role of cloud is clear and well defined. Cloud should be seen as part of the broader architecture, not the sole focus. The critical path for transformation—the real challenge—goes straight through legacy data centers. Without a clear plan for how to decommission legacy systems, cloud transformation will struggle to be sustainable.

Who in the organization should be leading this effort?

A decade ago, people would say, “Talk to my CIO” about cloud. This is still a common misconception today and a real symptom of why many transformation efforts fail; cloud isn’t just an IT issue anymore. Cloud transformation is a shared responsibility across the C-suite. The CFO is key to managing financial outcomes, while the CIO steers technical design and execution. The chief risk officer must also be involved, especially in regulated industries, due to operational risks. And the CEO is the ultimate sponsor, driving the transformation from the top down. So whose job is it? Its everyone’s job—and different leaders play different roles.

What are the key steps companies need to take for effective cloud transformation?

The first step is having a solid business case and infrastructure strategy, which I mentioned earlier. You need to define what you want to achieve and have a clear understanding of the financial commitments required. The next step is choosing the right cloud partner—whether it’s Microsoft, Amazon, Google, or another provider—because this is a long-term relationship that will last decades. It’s a big decision.

Once you’ve selected your partner—or in some cases, more than one—the key is building a detailed execution plan with clear milestones for your migration journey. The most important milestone is setting the date when you’ll hand over the keys to your data centers. Without that, overall expenses will inevitably spiral out of control, and things will veer off track. Execution is critical, as getting those data centers shut down is the linchpin to cloud migration success.