Smart Utility.gif

In his famous article back in 2011 “Why software is eating the world”, Marc Andreessen was making the case that all industries will eventually be disrupted by software technologies (now also know as “digital”). He did not mention electric utilities at the time, but now we are clearly there.


Here are a few reasons why.

Electrical systems are puzzles made of heterogeneous pieces: generators, grids, loads, etc…

What’s more they are not static but dynamic, their configuration changing all the time, with loads coming on-line, generators being disconnected or new cables being laid.

Complexity is now increased by the higher variability of consumption (under the influence of active prosumers, extreme weather and soon the moving loads that are electric vehicles) and of generation (renewable sources, ageing generator fleet). On top of that, the system that used to be distributed only on the demand side, is now also distributed on the generation side.

At the same time demand for reliability & safety have never been so high in our “always on” world, where no one wants to be armed using electricity.

Moreover, electrical systems are not only physical systems but are also interdependent with other sub-systems (markets, regulation, human organisation, environment, etc.). Predicting their evolution only on past performance is not enough.

Heterogeneous, dynamic, complex, distributed systems of systems require more intelligence to be safe, reliable and efficient.

This and the nature of electricity that requires split-second reactions in real time systems, puts their design and control beyond the reach of the human mind alone. The needed intelligence has to be embedded in software.

This comes at a time when the proliferation of sensors (starting with smart meters) generates a deluge of data that would submerge any human brain. Software not only keeps one afloat on data lakes but can navigate you through oceans of information. Although at one point there might be too much for the telecom infrastructure to efficiently move around this data, leading to decentralized intelligence approaches such as edge computing. For example energy consumption data can be processed on a local concentrator in the sub-stations and only the load curves are sent to higher level systems.

More than improvements, software opens new realms of possibilities

Software is not only about what utilities have been doing but better. Indeed grids can be operated closer to their physical limits, for example with analytics to detect patterns precursors to failures.

Software models for better planning can also be used at the design stage to right size investment in infrastructure (eg. define the optimal capacity of a battery to perform profitably arbitrage and ancillary services on a given market). Once modelled, the system’s behaviour can be simulated to understand interactions and support decision making.


Software is about what can be done now that was simply not possible before. For example flexibility from behind the meter will not be accepted by consumers unless it is imperceptible, thus automated.

Self-learning algorithms make that possible by observing usage habits and working for example on inertia of loads such as HVAC of buildings. Software that gives deep insights on customer behaviours and new ways of interacting with them also allows new services yet to be imagined for energy suppliers to sell far more than just electrons by addressing the needs for security, comfort, efficiency, maintenance, resilience, performance, banking, communication, mobility, ...


So while batteries are currently all the rage, watch out for the software…