Global IT spending will grow 9.3% year over year in 2025 to $5.74 trillion, according to a Gartner report. The market will achieve a 7.2% increase this year, topping $5.3 trillion, the analyst firm said.
John-David Lovelock, distinguished VP analyst at Gartner, explained CIOs are coming out of 2024 with many proof-of-concept projects having been completed, and many of the projects being a failure.
“GenAI PoCs failed most often due to corporate data inadequacies, inability of the organization to change and lack of ROI,” he said.
In 2025 and 2026, CIOs will be reducing their expectations for what GenAI can accomplish while simultaneously increasing the spending.
Lovelock predicted there would be a transition away from self-development of GenAI projects toward a reliance on GenAI enabled features delivered by incumbent software vendors.
“After 2025, IT spending increases at over $500 billion dollars every year,” he added. “This is the surest sign that IT has moved from being a back-office cost center to be controlled to being a foundation component of corporate strategy.”
Mitigating AI, IT Costs
Derek Ashmore, application transformation principal at Asperitas, said higher demands on processing power will lead to higher costs, so there will be a significant focus on mitigating those costs.
He noted training and running AI is highly compute-intensive, so he expects to see more focus on the efficiency of the AI models.
“This is especially likely due to the many different AI models coming to functional parity with each other,” Ashmore said.
With rising energy consumption driven by AI demands, data centers will increasingly focus on renewable energy sources, such as solar, wind or hydroelectric power to support their operations.
“Data centers will likely prioritize energy-efficient servers, power supplies and cooling systems and adopt intelligent management software that dynamically adjusts power usage,” Ashmore said.
Network Architectures Challenged
Amir Khan, CEO at Alkira, said the projected $500 billion increase in IT spending highlights a critical challenge, which is that traditional networking architectures weren’t designed for today’s distributed, data-intensive workloads.
“As companies scale their IT infrastructure, the network becomes a potential bottleneck,” he said.
He added organizations are discovering that traditional hub-and-spoke networks can’t efficiently handle the massive data flows between clouds, data centers and edge locations that modern applications require.
Meanwhile, the substantial growth in data center spending presents both challenges and opportunities from a sustainability perspective.
“While increased power consumption is a concern, particularly with GenAI workloads, modern networking architectures can actually help reduce environmental impact,” Khan said.
By optimizing network paths, organizations can minimize unnecessary data movement and reduce their carbon footprint.
From his perspective, the key is intelligent workload placement and network optimization.
Rather than building new physical infrastructure everywhere, companies can use cloud-native networking platforms to dynamically route workloads to the most efficient locations, whether that’s for power usage effectiveness or data locality.
“This approach helps balance performance requirements with sustainability goals,” Khan said.
GenAI Adoption Strategies Evolve
Ashmore said CIOs are beginning to understand what to expect from GenAI as the hype starts to subside.
“Some original expectations were that artificial general intelligence was just around the corner and that most knowledge worker jobs would be affected,” he said.
He said with GenAI solutions including Microsoft’s Copilot and OpenAI’s ChatGPT, CIOs now understand what GenAI can and can’t do much better than a year ago.
Ashmore added the market also saw many GenAI competitors on par with ChatGPT, such as Anthropic’s Claude, Meta’s Llama, X’s Grok, and Google’s Gemini.
“We’re also seeing significant impact in business intelligence and analytics with solutions like Microsoft Power BI and Tableau GPT,” Ashmore said. “In cybersecurity, platforms like CrowdStrike Falcon and SentinelOne are leading the way in advanced threat detection and response.
