Industry executives and experts share their predictions for 2026. Read them in this 18th annual VMblog.com series exclusive.
By Michael Reid, CEO at Megaport
As 2025 comes to a close, compute power continues to anchor AI workloads. However, even the most advanced AI systems cannot operate at scale without robust, interconnected networks. Here are our predictions for 2026 and the role networking will play in powering AI at scale.
AI bubble and the infrastructure reality
The AI industry isn’t going to form a typical bubble because there’s real revenue being generated by companies like OpenAI and Meta. Unlike the dot-com bubble, these are large companies with established business models and AAA-rated partners. Even if things cool, there will be a prolonged growth story because data centers take years to build and power takes even longer.
AI looks more like the rise of the cloud than the early beginnings of the internet. Companies like Nvidia, Microsoft, and OpenAI are taking on major commitments backed by strong deals. These are not small startups that could disappear overnight; this is a market built to last.
Networking emerges as the next bottleneck – and opportunity
While power and space created the first bottlenecks for AI, networking is becoming both the next constraint and the next opportunity. Global inference and sovereign AI initiatives require massive, interconnected pipes, and that’s exactly what Megaport provides, serving as the hidden engine behind the AI movement.
Networking is one of the most critical elements of AI, but it often goes overlooked. Power and capacity are incredibly important, but without connectivity, they can’t deliver results. Network as a Service (NaaS) changes the equation: automated, fast, scalable, and edge-connected, NaaS enables businesses to run AI workloads more efficiently and competitively.
Take Australia as an example: companies are building 1.2-gigawatts of NVIDIA chips in data centers, some even in remote areas across Tasmania. AI doesn’t care where it’s trained, but without high-performance networks, it simply cannot function. NaaS turns networking from a bottleneck into a strategic enabler, unlocking the next generation of AI performance.
Networks as the highways of the AI economy
To effectively implement a NaaS-based strategy, you need a network to connect everything to anywhere. You can have servers and storage, but without the network, it’s like owning a hypercar surrounded by dirt roads and potholes; faster or higher-performing cars don’t matter if nothing can move. Networks are the highways that will always be needed-they won’t go away.
As AI deployments accelerate into 2026, scalable, high-performance networking delivered through NaaS will become a key differentiator. Companies that elevate connectivity to a strategic, service-driven capability will be best positioned to unlock AI’s full potential and outpace slower-moving competitors.
