Opinion Leaders
AI and energy, a political issue
-
Enguerrand Artaz
Strategist
La Financière de l’Échiquier (LFDE)
With mid-term elections looming, rising electricity prices have become a major political issue in the United States.
More specifically, there is concern about the risk of prices skyrocketing due to a sharp increase in demand, against a backdrop of rapid expansion of data centres for AI. Several Democratic Party figures have voiced strong criticism, with Senator Bernie Sanders going so far as to call for a nationwide pause. This concern has been echoed in the Republican camp, with Donald Trump explicitly calling on technology companies to pay the additional energy costs generated by their consumption, without this affecting American households.
However, the various states of the union were not satisfied with these statements and have already taken action. Six of them, including Georgia, Virginia and Oklahoma, have proposed moratoriums on the construction of data centres to give themselves time to establish regulatory frameworks. This has led to a significant increase in the number of data centre construction projects being postponed or cancelled at the national level. The most common response, however, is to make the companies that own the data centres pay the additional costs, using a number of mechanisms. The simplest and most widespread is for federal states to create a specific regulated tariff (large-load tariffs), distinct from the traditional industrial tariff, which allows the costs generated by this additional demand – particularly in terms of network reinforcement and the creation of new capacity – to be reflected in the price without placing a burden on other categories of consumers. This mechanism is often associated with long-term commitments and minimum bills, backed by financial guarantees. The aim here is to secure network investments and prevent them from weighing on other consumers’ bills if data centres ultimately consume less than expected for various reasons. Ohio, for example, imposes a minimum billing of 85% of a reference demand.
These measures are found in many US states. However, technology companies are increasingly being encouraged to directly finance a significant portion of new production capacity, or even to set up dedicated supplies. This is the case, for example, with the Socrates South gas-fired power plant project in Ohio, which is intended to exclusively supply a Meta Platforms data centre without being connected to the physical grid. Other examples include Amazon’s partnerships for the construction of small modular nuclear reactors and Anthropic’s recent announcement that the AI company has committed to covering the full cost of upgrading the electricity grid for its data centres, as well as contributing significantly to the commissioning of new production capacity. With its “Community-First AI Infrastructure” plan announced earlier this year, Microsoft is following the same logic of fully financing the network upgrades required for its data centres. And in most cases, the political dimension is very much present, with a strong emphasis on the commitment not to burden households’ energy bills with the accelerated development of AI.
For technology companies, access to sufficient energy resources is a crucial issue and a prerequisite for achieving their objectives. Their commitments to finance a large part of the increase in electricity production capacity nevertheless represent an additional burden on the colossal expenditure incurred, even though a large part of this is supposed to be billed to the end customer. This reinforces the imperative of pricing power[1] . And at a time when these companies are shifting from models of abundant cash flow and strong shareholder returns to models focused on capital expenditure, thereby changing their status in the eyes of investors, this is not insignificant. From the perspective of the US economy, however, the positive spillover effects could be significant. These plans to increase energy capacity will promote a notable increase in activity – and undoubtedly in employment – in related sectors, particularly construction and energy engineering, while helping to develop and modernise the US electricity grid – all with a modest impact on public spending. Ultimately, the entire economy could benefit. All of these factors reinforce the idea that the US economy is likely to accelerate again and, on the stock market, that the rotation will continue, with large technology stocks less popular but a clear resurgence of appetite for small and mid-cap stocks.
__
Disclaimers
These data and opinions from LFDE, as well as the stocks and sectors mentioned, are provided for informational purposes only and do not constitute an offer to buy or sell a security, investment advice or financial analysis. Past performance is not indicative of future results.
[1] A company’s ability to raise its prices without affecting demand.
