2025 Predictions: What’s Ahead In Infrastructure And Operations – Part 1

Managing cybersecurity and technology risks are top-of-mind CIO priorities. What does that mean for your infrastructure and operations? How will you help your infrastructure and networking teams do more with less in the coming year? 

A panel of Sayers expert engineers and architects have answers they’re ready to share as part of their 2025 predictions. They met recently to offer insights to help you make the best infrastructure and ops decisions for your organization. 

From consumption-based services to AI-enabled networking platforms, their predictions include:

1. Generative AI Will Influence Infrastructure Decisions And Drive As-A-Service Models.

Companies looking to gain additional value from their data are turning to Generative AI. With GenAI, organizations can generate new content, product designs, business processes, and more. 

Before GenAI can transform industries, companies must make critical decisions about their infrastructure to support their AI endeavors. Some organizations are still in a wait-and-see mode when it comes to GenAI and are spending time on securing their data and drafting corporate AI policies to prepare for future GenAI initiatives. Other companies are deciding whether to buy or build their GenAI capabilities and are potentially building initial GenAI proof-of-concept environments in the public cloud before making large, on-premise infrastructure purchases.

Vendors such as HPE and Dell are making GenAI as consumable as possible in the data center. For example, the HPE Private Cloud AI turnkey offering is available in four right-sized configurations (think S, M, L, and XL) to support a broad range of AI workloads and use cases. If a company wants only to do a small test bed, they don’t have to make a huge capital purchase and leveraging as-a-service options like HPE GreenLake and Dell Apex help companies start small and grow as needed without having to manage any of the backend hardware.

From cost considerations to sustainability challenges, GenAI will influence organizations’ infrastructure decisions in 2025. Key factors include:

Cooling systems. Processing massive volumes of data generates heat that must be cooled to prevent hardware damage. Steve Johnson, Senior Solutions Architect at Sayers, says:

“The average server still can’t handle that much additional heat, especially when some put four to eight GPUs in a server. With the heat that generates, you’ll need additional liquid cooling.” 

Power and sustainability. On top of the cooling power needed to control data centers’ temperatures, GenAI’s need for increased computing power drives additional energy consumption. As data centers expand to meet the demands of GenAI, power requirements will continue to rise.

AI as-a-service. The rising costs of powering and cooling on-premise data centers may drive a shift toward AI or GPU as-a-service. That’s especially likely for smaller and medium-sized enterprise companies that don’t need a large footprint. Johnson says: 

“With the as-a-service model, you can have an uptick in GPU processing when you need it, and then a down-tick when you don’t so you can save some cash.”

Shared infrastructure models will help more companies leverage GPU processing for GenAI. Mark McCully, Director of Modern Data Center Engineering at Sayers, says:

“We will see more consumable GPU resources you can use for whatever compute you need for your inferencing and training models. You can pull that back when it’s not needed, and let somebody else manage the back end. It’s more of a rental model, rather than trying to build your own AI-ready data centers.” 

2. Pre-trained LLM Models Will Spur GenAI Rollout. 

Training an AI model is a difficult, complex process. A scarcity of people with the skill set to ensure you have the right volume, quality, and diversity of data, for example, has hampered companies’ AI progress.

According to Kevin Finch, Senior Business Continuity Architect at Sayers:

“As companies try GenAI and figure out ways to use it, training will be a big factor in how they do it. For some, training their AI model is a real problem in getting the results they want. Microsoft has pre-trained many of their LLM models, and there’s a lot of value in that.” 

Companies can customize a pre-trained language model using prompt engineering, fine-tuning, and other techniques to adapt the model to specific tasks. Microsoft recently announced new, adapted AI models using industry-specific data for top use cases in healthcare, manufacturing, financial services, and other areas.

3. More Companies Will Pursue Full-Stack Observability For The Network And More.

Modern observability platforms, exemplified by market leaders such as Dynatrace, New Relic, and LogicMonitor, offer several advantages over legacy monitoring tools. With full-stack observability, you can know what’s going on with your applications and infrastructure performance, and quickly get to the root cause faster if there’s an issue.

Observability platforms aim to get full-stack, end-to-end visibility across not only the application layer but also the database, infrastructure, and network layer. Observability attempts to consolidate the capabilities of multiple legacy monitoring tools into one platform that can do it all, while adding AI and machine learning components to be more proactive and intelligent. 

An observability platform’s tool consolidation leads to cost savings, and full-stack visibility enables teams to recover more quickly after an incident. These platforms also help teams identify and address potential issues before things can go wrong.

McCully says:

“We expect more observability conversations across the board, with customers wanting observability not only on the network or infrastructure side, but also across cloud-based platforms and applications.”

4. AI Will Help Drive Digital Experience Assurance.

Thanks to integrated applications and systems, today’s networks no longer consist of siloed workloads. But throughout the life of the data packet as it traverses the network, how can you provide a consistently reliable digital experience for your employees, customers, and other stakeholders?

AI will play an important role in providing predictable user experiences throughout the different network domains. Olczak says:

“Companies are still struggling with correlating and contextualizing the data to take the insights, do something productive with them, and provide a predictable experience for the end user. That’s where AI has moved into, and we’re seeing more tools that can take the data and provide better insights. Many organizations will adopt these tools over the next few years.”

5. Companies Will Get Back To The Basics While Transforming Architecture.

Organizations that have kicked the can down the road since the COVID-19 pandemic are ready to start dealing with their accumulated technology debt, whether it’s legacy routing and switching gear or wireless infrastructure. 

Next-generation solutions have evolved with additional functionality, and many companies still lack some of these basic solutions with broadened capabilities. 

Expect more companies to turn to networking platforms that offer a centralized single-pane-of-glass experience. Such platforms help simplify operations, improve network performance, and strengthen security posture. 

Networking platforms have evolved with AI-native, unified functionality that offers automation and efficiencies to help organizations do more with less. Cloud-managed networking platforms such as Alkira and Aviatrix help simplify connectivities between multiple regions in one cloud, or make it easier to set up multi-cloud networking.

Roland Olczak, Senior Network Engineer at Sayers, says:

“During the pandemic, companies deployed critical, time-sensitive solutions that left them in an inefficient and complex state of infrastructure. Now they’re looking for an architectural transformation to provide them greater end-to-end visibility, control, and efficiency across multiple network domains.” 

Questions? Contact us at Sayers today to discover extensive technology solutions, services, and expertise to cover all areas of your business.