For those who live in countries with competitive energy supply markets, the heinous price shocks and the dark magic of market forces have again questioned the role of competitive markets to deliver value for end consumers. In the UK and Australia, price regulation is now entrenched, and as a result some are questioning whether fully competitive markets will ever emerge on the other side of the Ukrainian crisis. European consumers are likely to pay an extra 2 Trillion dollars next year alone, which is bigger than all technology companies in the sector (including Tesla) combined. We should also note this is the first energy transition crisis – there will be many more in the transition as fossil fuel production slows, and we experience the effects of climatic variability.
To cut to the heart of this issue, apart from the construction and maintenance of grid generation systems and network infrastructure, the rest of the industry can be almost completely automated out of existence, shielded from shocks, manage risk, and ensure sound consumer outcomes. This does mean an end to market efficiencies as we know them today. It doesn’t necessarily mean nationalisation of the energy system, nor does it mean being void of healthy competition and innovation.
Can AI replace the retail energy market?
Technically speaking, this is not hard (it comes at substantially lower risk than in autonomous road transport), and no offence to energy companies out there, but people are not going to miss their energy suppliers if they cease to exist, and will be happy to lean more heavily on their own solar and flexibility to meet their local energy needs. To do this, we would need to make clearer the separation of the two functions of risk management algorithms for forward contracting energy to customers, from all of the residential, commercial and industrial technology platforms for billing, energy management and extension services, as well as operation of local energy resources such as rooftop solar PV.
AI trading and increased transparency
As for the problem of managing risk for both generators and consumers, despite what most energy market participants (incumbents really) will argue, this can likely be achieved more efficiently through a different system than what we have in place today. The real truth of today’s energy market is that energy producers inadvertently collude as much as they compete due to the way we have constructed our energy markets.
As a thought experiment, lets now imagine a scenario where we replaced every actor in the current wholesale electricity market by smart AI algorithms. In this world, technical regulators would be required to inspect the behaviour of these algorithms of each actor, to ensure they are consistent with the intent and the rules of the market, and that consumers are not being forced to pay the costs of either market failures or exploitations of rogue algorithms.
This regulated AI approach would force us down a path where the technology systems are regulated to take the actual input costs rather than bidding behaviour that is open to market leverage to arbitrarily set generation prices. One might also expect that regulators would inspect these algorithms to eliminate the unfair use of market power in market trading. Wholesale generation prices would begin to act as if they were price regulated. Insider trading on privileged information would be a thing of the past.
More sophisticated markets
Another potential strategy in this AI world is to give up on our outdated notations of technology neutrality in the energy markets. This could allow the use of algorithms such that scarcity in one fuel source, such as gas, could be decoupled from those of renewables, and thus alleviating price pressures on consumers where it is not warranted. Auctions for battery storage could be run separately from coal or nuclear, each that provide different characteristics to the system but are baselined differently. This could still allow generators to build and operate to maximise profit, but it would restrict the exploits of a shortage in one technology from cascading across the market to push up prices everywhere.
Smarter energy tech and smarter regulation
It is true that we need some forms of regulation and transparency in the energy sector to be sure that customers are not being charged or penalised unnecessarily for energy. But in a well structured market that is built for AI, machines can be goal oriented to optimise for consumers against the wholesale energy traders. With the requirement for transparency and explainability in these AI systems, we can foreseeably create a lower cost outcome for customers, alongside a far better user experience.
For the most part, the smart energy tech for managing solar, batteries and EV charging is already here today, and can optimise for customer outcomes. These will require further transparency and structuring to ensure that conflicting interests are structurally separated in the value chain, but otherwise the shift to full automation in CER is not a big jump.
Energy markets 2.0
The time is rapidly approaching where we need reconsider the design and regulation of the electricity retail markets as a multi-layer, AI driven infrastructure that algorithmically adapts to serve consumers via vetted and transparent algorithms. Foreseeably, these systems could be faster and better at detecting risks (e.g credit, hardship) than our energy retail systems today, and could better provide basic levels of service to all consumers and improve inequity. Whilst this will not eliminate the current problems of consumers not being able to pay for their energy, it bring some large cost efficiencies to the sector, and make the job the regulators inherently more technical, but much easier to enforce.
These are all things virtues that expect from our energy system today. To achieve them, energy regulators need to change their remit and mindset away from being economic regulators, towards technical regulators that require bidding processes to be automated and auditable, and to routinely inspect that AI systems are operating to a mutually agreed set of rules and objectives. AI systems can be audited. Humans cannot.
Some might argue that this type of transition is too costly, and will only push up energy costs. With foreseeable cost increases to consumers and interventions from government in the order of 10’s of billions of dollars, it makes the costs of these reforms look trivial. The only things holding us back is the imagination and foresight of our policy makers, and how dire consumer energy costs become over the next two years.
