Industry executives and experts share their predictions for 2026. Read them in this 18th annual VMblog.com series exclusive.
By James Tomko, SVP Sales at Zayo
The conversation around AI infrastructure has evolved rapidly. Early AI infrastructure decisions were dominated by the need to secure compute capacity. Today, as AI continues to scale, attention has shifted to how data moves between systems.
The network, once assumed to simply support innovation, is now foundational to AI performance and adoption. In 2026, the most consequential shifts in digital infrastructure will be driven by how networks are designed, built, and operated.
1. Bandwidth Will Make or Break the Next Phase of AI
AI is only as powerful as the networks behind it. Yet unprecedented data volumes, the need for ultra-low latency, and demand for rapid scalability are pushing legacy infrastructure to its breaking point. AI-driven data center capacity is expected to grow 3-6x by 2030 and require approximately 200 million new fiber miles of long-haul and metro fiber to support.
At the same time, delivering this fiber infrastructure is a long game, often taking two to four years from planning to execution, even as bandwidth is assumed to be available on demand. If more fiber is not built now, the market could face a bandwidth gap that could stall AI progress. In 2026, digital infrastructure providers that treat network investment as an afterthought risk being constrained in how and where they scale, while those that prioritize future-ready networks will lead the next wave of AI-driven innovation.
2. The Long-Haul Construction Ecosystem Will Narrow-For the Better
AI’s insatiable bandwidth demand has brought many new entrant to the long-haul market, eager to capitalize on what seems to be a “sure bet” infrastructure investment. However, despite continued demand expected, we’ll start to see the long-haul ecosystem narrow in 2026. The surge of new capital into long-haul fiber will spark innovation and much-needed expansion, but not every entrant will be equipped for the operational realities of large-scale builds. The gauntlet of permits, jurisdictions, vendor chains, and environmental factors can derail projects that already cost tens of millions of dollars, leaving some developers in the red and delaying critical progress.
Over the next two years, the market will self-correct as experience and execution take precedence. At the same time, the pace and scope of long-haul construction will force deeper aggregation across the ecosystem to accelerate delivery at scale. This natural shaking-out of the market will create a more disciplined ecosystem capable of building at the scale, speed, and quality AI requires.
3. For Data Centers: AI Advantage Will Be Won Beyond the Rack
As AI workloads become inherently distributed, data center competitiveness will no longer be defined by compute density inside a single facility. The advantage will come from how well data centers are designed to connect that compute beyond their four walls. AI training, inference, and data pipelines increasingly span multiple sites, metros, and regions, making inter-data-center connectivity a core differentiator. Data centers that integrate metro-scale and long-haul fiber into their designs will be better positioned to support distributed AI. In this next phase, leadership will depend on architecture that treats connectivity as a first-class design requirement, not an afterthought.
4. Interconnection Is Becoming the Defining Constraint for Distributed AI
AI training, inference, and data movement are no longer confined to a single environment. GPU clusters, neocloud platforms, and data centers increasingly operate as parts of a distributed AI ecosystem. As these distributed AI fabrics take shape, success will hinge on the ability to interconnect compute environments across locations, allowing models and data to move efficiently at scale. In this next phase of AI, the challenge will shift from simply adding more compute to ensuring that distributed systems can function as a cohesive whole.
5. Enterprise networking will become cloud-like, programmable, and agentic
As multi-cloud and AI-driven workloads accelerate, regulations continue to shift, and security threats evolve in real time, enterprise networks are under constant pressure to adapt. Organizations relying on static infrastructure are increasingly constrained by the gap between how quickly their business needs to move and how quickly their networks can respond.
The impact shows up as persistent friction: delayed deployments when capacity can’t be scaled in time, cost and compliance risk when data paths can’t be reconfigured quickly, and security exposure when networks lack the visibility or control to isolate emerging threats. In this environment, enterprises will require a more agile and intelligent network experience. Networks will need to support real-time scaling, rapid reconfiguration, and deep observability, with those capabilities exposed to AI-driven operations. Without that shift toward programmability and autonomy, the network risks becoming a drag on growth rather than a catalyst for it.
6. Emerging technologies will create new demands on networks
AI is the immediate driver of change today, but it won’t be the last. Networks will need to be built to support emerging technologies still taking shape, with quantum computing representing just one example of what’s ahead. Within the decade, the race for quantum advantage will become a race for quantum-ready networks. As quantum computing moves from lab environments into distributed, hybrid infrastructure, the bottleneck will shift from qubits to connectivity. Legacy network infrastructure cannot keep pace with the extreme bandwidth, ultra-low latency, and uncompromising reliability that quantum workloads demand. While adoption will remain limited in the near term, early preparation will determine which organizations are positioned to scale as the technology matures.
As AI and emerging technology innovations continue to reshape how data is created, moved, and consumed, the network will increasingly determine what is possible at scale. Organizations that design, invest in, and operate their networks with long-term foresight will be best positioned to adapt as new technologies and demands continue to emerge.
