AI has transformed the technology ecosystem over the past year, pushing networks to become faster, more agile and more scalable to support these workloads. In parallel, many operators are working to integrate AI into their own operations, from predictive maintenance to customer service chatbots, to keep pace with the escalating scale of modern infrastructure.

Given their inherent complexity, technological evolutions are always difficult to predict. Despite this uncertainty, operators can prepare for tomorrow’s demands by understanding market conditions and technological requirements throughout the ecosystem. With these shifts, there are a few key trends that may rise in importance as AI applications proliferate.

More fiber: Build it and they will come

A few larger customers are asking for more and more fiber pairs to support the capacity requirements of AI and cloud applications, with these applications driving value creation across the global business landscape. Now that we’ve reached Shannon’s limit (the amount of capacity that can physically fit on one fiber pair), more construction is needed.

This era requires careful consideration and planning, driving operators to balance their builds with business opportunities and other factors. Operators can now no longer mostly rely on leveraging or revitalizing existing infrastructure.

However, these new builds require massive investment. Meanwhile, every operator wants to win new customers, but they must balance these goals within a hypercompetitive industry, where they risk losing customer opportunities to competitors even if they do build new infrastructure in a high-demand location.

AI data centers are being built in locations with affordable, green power, but these are not normally the places where cables currently exist, which also calls for new cables to be built. Operators will build more in 2026 but must also differentiate themselves from competitors through customer service excellence and other factors, particularly in a saturated market where many providers offer similar solutions.

With some experts proclaiming we’re in an “AI bubble,” similar to the dot-com era of the late 1990s, the current landscape is being shaped by emerging AI neoscalers (with high bandwidth consumption patterns but less predictable longevity) and well-funded hyperscalers.

So, this time around, the “bubble” will include a few household names as well. Together, these groups are driving growth in network capacity needs, creating new business opportunities for operators and additional pressure to expand infrastructure.