Earlier this year, it was announced that artificial intelligence/machine learning (AI/ML) was being used to generate temperatures hotter than the sun’s surface on earth.

DeepMind worked with scientists at the Swiss Federal Institute of Technology (EPFL) to develop a neural network capable of controlling the magnetic fields essential to safely containing the plasma within EPFL’s Tokamak research fusion reactor.

Instead of using 19 magnetic coils, each controlled by a separate algorithm, DeepMind’s neural network can control all the coils at once and automatically learns which voltages need to be supplied to safely contain the plasma, resulting in a more stable, longer-lasting jet. DeepMind’s breakthrough paves the way toward fusion becoming a viable power generation source that can provide abundant clean power indefinitely.

While this is an extreme example of how AI/ML can decarbonise the grid, there are more near-term plays that investors and others in the Climate tech ecosystem should be paying attention to.

We touched on the need for long-duration energy storage using thermofluids solutions in a recent post. While these would usually sit in the high voltage parts of the network and hence their capacity is easy to track, there is an increasing build-up of residential, commercial and industrial (C&I) battery storage and solar photovoltaic (PV) generation in the low voltage segment.

Hence more and more of the data required for AI/ML models to work out what additional clean energy resources can be deployed and when they can be deployed is sitting in this low voltage part of the network, and that data is not actively being captured, analysed and used.

Consequently, it is becoming harder to manage the grid and anticipate when additional supply may come onstream from these sources as the data tracking their generation/resources are not centrally tracked.

For example, in the UK, there is no central “single source of truth” for the amount of electricity being generated from PVs, despite its installed base accounting for c.30% of the UK’s generation mix. In order to match supply to demand, the UK’s National Grid Electricity System Operator (ESO) needs to know how much fossil fuel fired electricity generation to keep ready for load following – hence it needs to know how much PV electricity is being generated at any given moment.

This task will become even more challenging as battery storage expands and can draw down to supply the wider grid beyond an individual household e.g. in the case of bidirectional vehicle to grid (V2G) which Octopus Energy has been trialing in the UK.

While individual players, such as Octopus Energy, have been creating innovative solutions to enable smart grids, grids at a national level lack the infrastructure to collate the data required to train AI/ML models. Gartner estimates enterprise IT spending for the energy and utilities market globally stands at nearly $200bn, but a fraction of that is spent on enabling smart grids to truly achieve their potential.

iPaaS (Integration Platform as a Service) offerings, which connect the data from the Edge to the Cloud, are required to enable that.

There are some stand-out players in the space, such as Greenbird, who enable iPaaS for the utility market. The company’s Cloudwheel solution is an integration Platform as a Service (iPaaS) offered as a managed service that captures data and transforms it into accessible and usable forms. The solution can collate and make sense of the data sitting in the low voltage segments of the national grid network. This data can then be used to train the AI/ML models required to determine what additional clean energy resources are available and when they should be deployed. Others in the space include Sympower, Greencom Networks, and Reactive Technologies, however, on a combined basis, these companies have raised less than $100m of equity funding, highlighting the dearth of attention paid to the space by investors.

From our viewpoint, smart grids will transform power generation and usage that will undoubtedly have a significant role to play as we transition to a cleaner economy. But they are being limited by the lack of investment and communications infrastructure to access the underlying data that can unlock their true potential.

When it comes to investments or the lack thereof, smart grids are inherently capital intensive to rollout, hence governments, utility providers, and ultimately consumers / the public will want to ensure that they are maximising the return on investment of deploying them.

Transitioning towards a cheaper, cleaner, and more secure renewable energy future that can simultaneously provide grid stability will require a more significant shift in how data is viewed. Companies must begin treating data as a valuable asset and utilise it to increase profits. By using internal and external data, companies have massive opportunities to create new and innovative solutions.

DAI Magister has an extensive track record of working with players who facilitate the transfer of data from the Edge of the network (in this case, the low voltage segments) to the Core (National Grid ESO) and then onto the Cloud (to enable data transfers between generators, wholesale suppliers, power traders, etc.) We recently advised Bright Computing, a provider of Enterprise-grade MLOps, on its sale to Nvidia.

We have also worked with players directly operating in the data transfer space within the electricity network. We advised *Pod Point, one of the UK’s largest charging infrastructure providers for EVs, on its sale to EDF (Britain’s biggest generator of zero-carbon electricity). In addition, we also advised *SteamaCo, a UK-based IoT smart metering technology company, on its financing round led by Shell and *Depsys, on its round led by BNP Paribas.

Find out more about us.



Say Hello