(Bloomberg) — Last week, French utility Engie SA announced it would use Google’s AI-powered wind forecasting capabilities to optimize operations at its German wind assets. The pilot program is an extension of Google’s internal work that it says enables it to generate higher revenue by scheduling hourly commitments of wind energy on the grid up to a day in advance. A Google executive calls this offering “a business recommendations tool”, which it is – but it’s also a source of intriguing and important strategic questions for Google, Engie and Big Tech in energy in general.
Google’s AI sister company, DeepMind, continues to expand its reach, scale and capabilities. Essentially, feed DeepMind a set of large, hard-to-solve problems and it can often solve them. In 2020, for example, DeepMind’s AlphaFold initiative determined the structures of proteins, a problem that a scientist in 1969 predicted would take longer than the age of the known universe if done by force. brute.
At the same time, this remarkable capability is at best incidental to the core search business of Google’s parent company Alphabet Inc., which generated more than $90 billion in operating revenue in 2021, while its groups cloud and “other bets” work on health and transportation and AI had negative operating income.
Perhaps pointing out that, less than a year after DeepMind announced its success in protein folding, it made AlphaFold free for the world. Better wind forecasting capabilities might mean almost nothing to Alphabet besides allowing it to sell more cloud services to utility customers.
For Engie, the first and most obvious application of Google’s AI is to better understand what the utility might expect from wind patterns in the future. It is essentially about avoiding risks. The 1.5 day forecast will allow Engie to predict when the wind is available and, just as importantly, to plan when it is not. This should allow the utility to better schedule its other generators to meet demand when wind supply is low.
But there are other approaches these abilities could unlock for Engie as well. Knowing the wind patterns 36 hours in advance could also allow him to take on more market risk. Engie might be more willing to commit to when wind projects will produce power, and might do so further into the future than before. The company could become more confident playing in the spot electricity markets at times when prices are very high. In theory, it could even use its wind assets in a purely “market” way, completely exposed to the market price of electricity, with all the potential advantages to be captured and the disadvantages to be foreseen.
Anticipating the future for a DeepMind wind application forces us to ask ourselves several series of questions. The first set is technical. How can this technology improve? How far in time can his predictions go and how much more accurate can they become? Does Google’s set of predictions integrate well with a utility’s or independent power producer’s own planning and prediction systems?
It should be relatively easy to answer these kinds of questions, from company to company. Engineers meet engineers, software developers talk to each other, enterprise venture deals meet, and more.
The second set of questions concerns the energy market, and these are very different in nature. To what extent is an electricity market open to merchant generation? How far in advance do grid operators plan their electricity distribution? State or national electricity regulators To allow businesses to take the market risk with variable renewable generation? These questions may take some time to answer.
A final set of questions relates to regulation, politics and perhaps politics as well. Will a state utility commission decide that DeepMind-based wind forecasting technology is impossible to assess and ban its use? Will there be a political return to the influence of even narrowly and technically defined Big Tech in electricity markets? There may not be clear answers to these questions, at least not at first. But maybe they don’t matter so much in the end.
In his 2021 essay “Software Development”, independent analyst Benedict Evans states that “when software eats the world, the issues that matter stop being software issues.” In other words, the technology can penetrate a market like books, music, retail or movies – but ultimately the defining issues are those rooted in the existing industry. Take Netflix, for example. He “used technology as a wedge to get into the TV industry,” Evans writes, but “all the questions that matter to his future are TV questions,” like how long his shows will last and what will happen with sports rights.
I imagine it will be the same for artificial intelligence in energy. A better forecast should make it possible to better manage the markets. But at the same time, technology can’t change everything – and the most important questions that technology will have to answer will be the energy industry’s fundamental questions, not its own.