As we dust off the crystal ball and look at what might be coming in the coming year, one thing is clear – 2026 is shaping up to be the year that will significantly rewrite the rules of the networks that fuel our digital connectivity. AI isn’t just coming; it’s already taking control and transforming how we build, operate, and manage the network. Next year, we will see this shift ripple across every layer of connectivity, reshaping expectations, unlocking new possibilities, and forcing businesses to rethink what a network can really do.

AI forces a redesign of the network

2026 will mark the point where AI shifts from experimentation to fully commercialized, autonomous decision-making at scale. The acceleration in inference traffic alone will expose the limits of network architectures designed for linear data flows and predictable consumption. AI-driven workloads will generate volatile east-west traffic patterns, machine-to-machine exchanges, and microburst dynamics that current networks were never built to accommodate. Ultra-low latency, deterministic performance, and the ability to dynamically allocate bandwidth in milliseconds will move from “nice to have” to critical requirements.

The drive to generate ROI from AI will also put a bigger spotlight on the network. Today, some networks don’t allow GPUs to run at full utilization, which will no longer be acceptable. Those that cannot unlock full GPU performance will become a bottleneck, financially and operationally. The market will favor architectures that are elastic, programmable and power aware, aligning infrastructure resources and cost with real-time demand.

Interconnects will shape data center scale in 2026

The next wave of AI-driven innovation won’t be defined by GPUs alone, it will be defined by how efficiently we move data between them. The interconnect now becomes the gating factor for scale. AI fabrics will emerge as a competitive differentiator in 2026 as organizations confront the realities of AI cluster sprawl, stranded capacity, and input/output limitations in AI computing.

The industry will accelerate its shift from monolithic to distributed, fabric-first architectures where silicon, optics and link technology matter as much as the GPU itself. In 2026 we will see increased investment in new optical interconnects technology (co-packaged optics, photonic-integrated circuits), coherent technology pushed closer to the server, advanced copper interconnects, on-package optical connections that remove distance, power, and thermal constraints inside the data center. The strategic winners will be the companies — cloud, semiconductor, OEM, and hyperscalers — that treat interconnect as a first-order architectural decision, not a late-stage engineering problem.

There will also be a sharper focus on energy per bit as AI scales. In many environments, more energy is now spent moving data to and from GPUs than on the computation itself, putting metrics like picojoule per bit on the agenda. Interconnect choices will be judged not only on speed, but on how efficiently they move every bit of data.

Edge computing will become a strategic niche, not a mass-market solution

In my personal view, 2026 will be the year the industry recognizes that a broad, ubiquitous rollout of Edge computing will shift to a more pragmatic approach, focusing investment on targeted locations that deliver greater value. Certain locations will absolutely require real-time, latency-sensitive compute, particularly for sovereign data, regulatory obligations, and mission-critical applications, such as financial services and autonomous operations. But mass, distributed Edge computing will take some time to materialize.

Telcos, many of which own the network Edge, have a significant strategic opportunity to provide targeted, high-value Edge capabilities in the right locations. They will benefit by focusing on fewer, carefully chosen sites, where latency, data sovereignty, and regulation truly demand it.

Satellite networks will play a bigger role in connectivity than anticipated

The industry has long viewed non-terrestrial networks (NTNs) as a means to fill coverage gaps where terrestrial connectivity is too impractical or costly. However, conversations from recent industry meetings and events tell me that NTNs are set to play a far more important, and potentially disruptive role than originally expected.

Tens of thousands of new satellites are set to launch in the coming years, with Musk alone securing licenses for 10,000 additional units. This rapidly expanding mesh of networks is evolving at pace and will soon reach a point where direct-to-cell services can offer performance competing with terrestrial coverage. It is important to note, however, that NTNs will never be able to compete on peak data throughput. They will be part of the broader connectivity ‘coverage package’.

For decades the workhorses of satellite communications have been Medium Earth Orbit (MEO) and Geostationary Earth Orbit (GEO) satellites. They have provided broad coverage, stability and high service level agreements (SLAs) essential for mission-critical applications, including government communications, broadcasting and remote connectivity service. The emergence of Low Earth Orbit (LEO) with low-latency capabilities and higher capacity adds another dimension and creates a package that helps these systems to compete for the very first time (albeit at a limited scale) with terrestrial service offerings.

Satellites will begin underpinning global enterprise connectivity and increasingly integrate with terrestrial systems, with offerings such as Amazon’s LEO network integrating seamlessly with AWS to deliver both cloud and communications services. So yes, NTNs will continue to be important for rural and hard-to-reach areas, but I predict 2026 will see their influence go far beyond that.

These are just a few of my observations around some of the key trends and technology events for the coming year. There is no doubt in my mind that 2026 will stand out as a pivotal year in the networking sector. AI Edge computing, NTNs/satellites and more will continue to make headlines, and the network itself will continue to underpin everything we do and get smarter and adapt to these emerging technologies.