The AI industry is boosting energy companies due to surging demand for electricity, and Constellation Energy (NASDAQ:CEG) could be a major beneficiary if this AI narrative succeeds in the long run.
Nuclear power seems to be the perfect fit for AI, since it can maintain a very stable power output and do so with solid cost efficiency. There are no ups and downs here like renewables, and nuclear energy itself is pretty green.
According to industry analysts, AI data centers could add approximately 323 terawatt-hours of electricity demand by the beginning of the next decade, potentially representing 8% of total U.S. electricity consumption by 2030. Wedbush’s Dan Ives also went on CNBC earlier to say that the use cases for AI have been exploding, and that the regulations have been too slow to catch up.
If this continues, even better results are likely in the coming quarters. For Q1, adjusted EPS came in at $2.14, up from $1.82 a year ago, but GAAP net income dropped sharply due to one-time items. Management sees full-year adjusted EPS at $8.9 to $9.6, and there is also a pending $26.6 billion Calpine acquisition to add natural gas and geothermal assets to the business.
Revenue growth is real and likely to continue as data center demand explodes. However, with shares already up sharply and profit growth lagging, the stock looks fairly valued for now.
If you own CEG, it makes sense to hold and watch how the Calpine deal and new AI-driven contracts play out.
It doesn’t make sense to me to buy CEG stock at current valuations, since there are more aggressive AI stocks with less downside at less than 29 times earnings. With CEG, you are essentially betting that AI will forever be a power hog, and that’s still something that hangs in the balance post-DeepSeek.
If anything, AI models have continued to become more efficient. DeepSeek R1 showed efficiency that was thought to be impossible previously, and the company is apparently working on DeepSeek R2 now. The efficiency gains from R1 also translated into most other AI models people use today, since DeepSeek is open source.
Many people are now deciding to use AI models on local hardware instead of paying API costs to companies, and this trend could accelerate if open-source AI models keep getting better. On top of that, AI is shifting towards inference instead of training over time. Inference is much less power hungry than training.
With all that in mind, I wouldn’t add more CEG stock to a portfolio at this point.
