Hello from New York, where we have had another chaotic weekend of politics. Former president Donald Trump was safe on Sunday after another apparent assassination attempt. There are 49 days until the election.
In today’s edition, I report on the latest developments in artificial intelligence and increasing electricity demand. You might recall Simon’s recent piece on the rise of virtual power plants and UK-based Octopus Energy. Today, I look at the companies that could profit from the power-hungry growth of AI.
ESG investing
Electricity providers are ‘next derivative on AI’
As demand for artificial intelligence technologies continues to grow, a new class of companies are starting to emerge as a way to play the sector: electricity providers.
“Investors are looking for the next derivative on AI,” James West, a senior analyst at Evercore ISI on sustainable technologies energy, told me. “The technology investors that are calling us are asking about power.”
“This is the next big bull market, especially as you have some of the other AI derivatives like the chips running out of capacity,” he added. Nvidia, the stock market darling of the AI phenomenon, saw its shares sink after its latest earnings report in late August. “It is hard for Nvidia to grow earnings further because their capacity tightens,” West said.
If this shift occurs, West said the companies that are poised to do well include GE Vernova, the power and renewable energy divisions of General Electric that have been spun out into a separate company, or Fluence, a battery provider competing with Tesla.
With data centres’ energy demands accelerating, renewable energy development is happening at a rapid scale, he said. Renewable electricity generated worldwide in 2025 is expected to surpass coal power for the first time, according to the IEA.
But that might not be enough. There are two broad approaches to meeting AI’s rapidly growing power demands, experts reckon. One path is “re-carbonisation” — restarting or maintaining fossil fuel power plants. This path exposes the major risk that AI and data centres will ultimately drive up carbon emissions. Microsoft’s emissions jumped 30 per cent between 2020 and 2023, largely due to data centres for its AI development systems, the company said in its annual sustainability report this year.
AI data centres demand “99.99 per cent reliable electricity,” Thomas McAndrew, founder and chief executive of Enchanted Rock, a Texas-based microgrid provider, told me. This demand strains electricity grids further and requires increased reliance on existing coal as well as new natural gas plants, he added. The AI data centres’ demand is causing higher electricity costs for residential and higher carbon emissions, McAndrew said. “Speed to power is crucial in the AI arms race.”
An alternative to ‘re-carbonisation’
But there is a second path. If the technology companies can offset power gaps with natural gas microgrids and battery storage, then “AI data centres can ease grid pressure and provide surplus power back to the grid, supporting the expansion of wind and solar, thus reducing costs and carbon emissions,” McAndrew said.
While hardly a zero-carbon fuel, natural gas can be used more efficiently to reduce emissions and fuel data centres, KR Sridhar, founder and chief executive at Bloom Energy told me.
Bloom provides back-up energy sources for data centres, and has been one of the star portfolio companies for Kleiner Perkins, the blue-chip venture capital firm that backed tech giants such as Amazon and Google. San Jose-based Bloom can take the heat from natural gas energy and recycle that to power the cooling systems for data centres, Sridhar said.
If Nvidia and other leaders in the AI space are looking overvalued to some investors, there are other options to ride the AI wave. Electricity infrastructure companies may not be as flashy as Nvidia’s semiconductors, but they could become an AI investing theme for 2025.
Smart read
More and more Chinese women are graduating from university — but it can still prove hard for them to advance on the corporate ladder.
Read the full article here