Cléber Alexandre Agazzi is a Head of Infrastructure & IT Operations, at Sicredi. With a career spanning over 15 years in IT leadership, he specializes in orchestrating transformative initiatives that drive organizational growth and innovation. As the Head of Infrastructure & IT Operations at Sicredi, Agazzi is dedicated to crafting executive-level strategies, optimizing digital infrastructures and fostering a culture of innovation through strong leadership and strategic vision. Managing infrastructure as a product (platform engineering), he focus on providing agile and scalable solutions to support the evolving needs of the business, ensuring seamless integration between technology and operations to drive efficiency and resilience while enabling sustainable growth.

The Shift from Cloud Optimism to Repatriation

Cloud migration has transformed the way companies manage their operations, promising greater agility, scalability and cost reduction. However, as these strategies mature, many CIOs are questioning whether the cloud is truly the ultimate solution for all workloads. While some applications thrive in this environment, others have shown significant challenges, leading many organizations to rethink their strategies.

Historically, data centers were considered reactive, slow and inefficient. The rise of cloud services changed this reality, enabling companies to meet their business demands with greater agility. Consequently, most rushed in this direction are attracted by promises of ease, speed and efficiency.

However, over time, it became clear that not every workload is efficient in the cloud. "Repatriation" movements emerged, as demonstrated by the 2024 Barclays CIO survey, published by Michael Dell on his LinkedIn profile, where 83 percent of companies plan to move workloads back to their private clouds. But why is this happening?

The transition to the cloud is not always optimized. The "lift and shift" approach, inaccurate cost control, inadequate sizing choices and the unpredictability of data growth can create significant issues, particularly in terms of cost. Public cases support these statements, such as Dropbox in 2016, which realized it would be more economical, in the long run, to build and operate its own data centers, or Bank of America, which decided to keep most of its infrastructure in its own data centers, despite the growing use of public cloud in the financial sector.

At the same time, there has been an evolution in data centers. Today, tools and platforms simplify and automate infrastructure delivery. Private clouds have significantly advanced, along with new consumption-based contract models (As a Service). It's already possible to achieve better costs with private clouds, even when compared to workloads deemed efficient in the public cloud, such as containerized and auto-scalable applications.

“The initial expectation that cloud would be the ultimate solution for all IT challenges has been replaced by a more balanced approach, recognizing the limitations and hidden costs of a purely cloud model”

That initial vision of the cloud absorbing all types of services has been replaced by a hybrid strategy, where the benefits of an on-premises structure coexist with what really makes sense to be in the cloud, after all, cost is only one of the pillars that should guide this decision.

On the other hand, when it comes to innovative and specialized technologies, such as Artificial Intelligence (AI) and Analytics, it makes more sense to use ready-made services available in the cloud, rather than investing time and money in acquiring and implementing disruptive or specialized technologies internally.

The trick lies in keeping in-house the systems that do not fully benefit from the cloud, such as consolidated applications with predictable growth, while ensuring agility in meeting business demands. At the same time, the cloud should be leveraged for innovative and highly scalable services, designed to thrive in that environment.

However, managing a data center requires attention to several aspects – availability, resilience, capacity, scalability, accessibility, speed, efficiency and risk. In the cloud, most of these issues come “packaged.” For companies with a mature and evolving structure, the transition to a hybrid model or the repatriation of workloads is a less painful process. On the other hand, for organizations whose infrastructure limitations prompted the migration to the cloud, or that were born in the cloud, this journey can be much more complex.

Some factors are crucial in this journey, such as the adoption of open technologies, instead of proprietary solutions that may lock the company into a single provider. Ideally, applications should be agnostic, capable of operating both in private and public clouds, with scalability between them in some specific cases.

Infrastructure and Platform are becoming commodities, even the container orchestration layer is becoming a market standard and some data center providers have realized they can compete at this level, offering these types of service. In the coming years, the real competitive edge will come from offering specialized products, such as specialized services in AI or ML, Big Data, IoT, E-mail and collaboration and platforms like SAP or Salesforce, regardless of the cloud provider behind the solution.

At a time when cloud dominates infrastructure discussions, it has become clear that not all workloads benefit equally from this model. The initial expectation that cloud would be the ultimate solution for all IT challenges has been replaced by a more balanced approach, recognizing the limitations and hidden costs of a purely cloud model. Workload repatriation is a response to this reality, driven by the pursuit of greater cost control and operational efficiency.

Hybrid Strategies

A well-planned hybrid strategy emerges as the ideal solution. It allows companies to internally manage predictable and consolidated workloads, while leveraging the agility and scalability of the cloud for innovations and highly dynamic demands. For organizations that have achieved sufficient maturity in their infrastructures, the combination of public and private cloud can ensure the best of both worlds: efficiency, agility and innovation. In the end, the secret is not in choosing between data center or cloud, but in finding the right balance between both, based on the specific needs of the business.