For the last fifteen years, the mantra of enterprise IT has been “move everything to the cloud.” Data centres got larger, more centralised and increasingly concentrated in a handful of regions — Northern Virginia alone hosts roughly 70% of all US internet traffic. But a counter-trend is gaining momentum: edge computing brings processing power back to where data is actually generated.
The edge computing market was valued at $15.7 billion in 2023 and is projected to reach $110 billion by 2030, according to Grand View Research, with a compound annual growth rate of 32%. What is driving this shift? It comes down to three things: latency, bandwidth and sovereignty.
Latency: Why 50 Milliseconds Matters
An autonomous vehicle travelling at 100 kilometres per hour covers about 28 metres per second. The latency of a round-trip to a cloud data centre 500 kilometres away is roughly 50-100 milliseconds. In that time, the vehicle has moved 1.4 to 2.8 metres — more than enough distance for a split-second decision to mean the difference between braking in time and a collision. Edge computing processes sensor data locally, in the vehicle or at a nearby roadside unit, reducing latency to single-digit milliseconds.
The same logic applies to industrial robotics, augmented reality and real-time financial trading platforms. Virtually any application where milliseconds matter is a candidate for edge processing.
Bandwidth: The Cost of Sending Everything to the Cloud
A single autonomous vehicle generates approximately 4 terabytes of data per day from its cameras, lidar and radar sensors. Uploading all of that to the cloud is not just slow — it is expensive and often impossible in areas with limited connectivity. Edge computing solves this by processing data locally and sending only the relevant, compressed results to the cloud. This “filter and forward” model dramatically reduces bandwidth requirements.
Even consumer applications feel the bandwidth pinch. A smart security camera that uploads 24/7 high-definition video to the cloud consumes roughly 400 GB of data per month. On-device AI processing — Apple’s Neural Engine, Google’s Tensor chip — handles motion detection, face recognition and alert filtering locally, cutting that upload volume by 90% or more.
Sovereignty: Where Your Data Lives Matters
The geopolitical dimension of data processing is becoming impossible to ignore. The European Union’s GDPR imposes strict rules on where personal data can be processed and stored. The US CLOUD Act creates conflicting obligations for companies operating across borders. China’s data localisation laws require certain categories of data to remain within Chinese borders.
Edge computing provides a technical answer to a legal problem: process sensitive data locally, within the jurisdiction where it is collected, and only transfer aggregated, anonymised results across borders. This is particularly relevant for healthcare data, financial transactions and government services.
Who Is Building the Edge?
The edge computing landscape is fragmented, with no single company dominating the way AWS dominates the public cloud. Amazon itself is investing heavily through AWS Wavelength, which embeds compute and storage at 5G network edge locations, and through its network of Local Zones — smaller, localised data centres in metro areas.
Microsoft is pushing Azure Edge Zones and Azure Stack Edge, hardware appliances that bring Azure services to on-premise locations. Google’s Distributed Cloud Edge extends Google Cloud infrastructure to edge locations. Meanwhile, telecom companies see edge as their opportunity: Verizon’s 5G Edge, AT&T’s partnership with Microsoft and Vodafone’s edge computing platform all aim to make use of their existing network infrastructure to host edge workloads.
A parallel story is unfolding in the hardware market. NVIDIA’s Jetson platform dominates edge AI, providing GPU-accelerated computing for robotics, drones and industrial inspection systems. Intel’s OpenVINO toolkit optimises AI models for its edge processors. Qualcomm’s Snapdragon platform powers edge AI in billions of smartphones and an increasing number of IoT devices.
What’s Next
Edge computing is not replacing the cloud — it is extending it. The emerging architecture is a continuum: device edge (on-device processing), network edge (processing at cell towers and local exchanges), metro edge (small regional data centres) and cloud core (traditional hyperscale data centres). Workloads will move fluidly along this continuum based on latency requirements, bandwidth constraints, regulatory obligations and cost.
For technology investors and enterprise architects alike, the key takeaway is that edge is not a single product category — it is a design pattern. The winners in this space will be the companies that make edge computing invisible: just another tier of infrastructure that developers can target without thinking about where the compute actually happens.