The internet boom of the 1990s led to a bubble in power production. Will history repeat itself with this AI boom?

Jinjoo Lee, an analyst from The Wall Street Journal, recently published an article analyzing the similarities and differences between the power demand during the current AI boom and the power demand during the Internet boom of the 1990s.

Lee stated that, similar to today, in the 1990s the expected returns of independent power producers' stocks were much higher than those of regulated utility companies' stocks. Back then, independent power producers built power plants on a large scale, resulting in oversupply and many companies going bankrupt. However, now it seems that power demand will inevitably increase and should not face similar issues. But the stock prices of independent power producers may have risen more than they should have.

Lee also warned that over the past two decades, the power industry has matured significantly, but the basic lesson still applies: high expectations may bring pain.

With recent massive investments in AI by US technology companies, power demand has reached unprecedented levels. A report from Wells Fargo & Co predicts that after a decade of weak power growth, by 2030, US power demand will increase by 20%. Only AI data centers are expected to add about 323 terawatt-hours of power demand in the US. Boston Consulting Group predicts that by the end of 2030, the electricity consumption of US data centers will triple that of 2022, mainly due to AI model training and higher-frequency AI queries.

Some analysts have also pointed out that while utility stocks may struggle to match the high valuations of tech giants, their investment value has been rediscovered as defensive stocks with high certainty and stable dividends.

Will this power surge lead to corporate collapses?

The last time independent power producers were this excited was in the late 1990s and early 2000s when the rapid growth of Silicon Valley drove expectations for increased power demand.

Similar to today, at that time, the expected returns of independent power producers' stocks were much higher than those of regulated utilities' stocks. Independent power producers claimed that their newly built gas-fired power plants would replace the old and inefficient generators of monopoly utility companies.

SSR's co-head of utilities and renewable energy research, Hugh Wynne, said that developers at the time assumed that long-term electricity prices would match the long-term marginal cost of new power plant capacity, covering variable operating costs and capital costs.

However, aggressive construction led to excess capacity in many markets, with electricity prices dropping significantly, only sufficient to cover the operating costs of these power plants, unable to repay the debt they borrowed for construction. Major power developers including Calpine and NRG Energy later went bankrupt.

Will the current market face similar issues? Jinjoo Lee says it may not.

First, the driving factors for this electricity demand growth are more explicit. Large U.S. technology companies are investing heavily in building artificial intelligence infrastructure; at the same time, the energy demand for transportation and industrial applications is steadily shifting towards electricity.

Secondly, independent power producers are more familiar with the operation of competitive electricity markets than before. Wynne said that during the last power heat wave, independent power producers and financiers funding their projects suffered significant losses, so they are now less likely to invest in new power plants without long-term power purchase agreements. Data centers require electricity available at all times, and technology companies are willing to sign long-term contracts at a premium. On September 20th,$Microsoft (MSFT.US)$The last Friday and $Alliant Energy (LNT.US)$The company signed a 20-year power purchase agreement.

The stock gains of independent power producers may exceed their expected gains.

Although the demand for electrical utilities seems to inevitably increase, the scale and timing of the growth are still uncertain, with differing market views. For example, the Arizona Public Service Company predicts in its latest planning document that in the coming years, demand from high-power users (such as datacenters) may range between 100 megawatts to 1000 megawatts.

It is worth noting that public utility companies' long-term electricity demand forecasts have undergone significant changes in recent years and may continue to be revised. The Rocky Mountain Institute projected at the beginning of 2021 that the total utility load would increase by 8.2% by 2035; by June 2024, it was already expected to grow by 23.9%.

Therefore, while a complete collapse seems unlikely, the stock gains of independent power producers may exceed their expected gains.

A well-known power generation company in the USA $Vistra Energy (VST.US)$ This year, $S&P 500 Index (.SPX.US)$nvidia is the best-performing company, second only to the second-best performing $NVIDIA (NVDA.US)$ company. According to the forward PE ratio, Vistra's valuation is about 26% higher than its historical average level. Alliant Energy is the fourth best-performing company, with a forward PE ratio of about 32 times, 57% higher than the historical average level.

As electricity demand rises, in addition to independent power producers, regulated utilities will also benefit. They have been working with technology companies to develop new generation power projects, including NV Energy supported by Berkshire Hathaway and $Duke Energy (DUK.US)$.

Editor/new