This article was originally published on Fool.com. All figures quoted in US dollars unless otherwise stated.
The artificial intelligence (AI) revolution will be an ongoing focus for investors, given its transformational potential. However, the winners and losers in the market could change rapidly in this fast-evolving space.
In a recent interview, Microsoft (NASDAQ: MSFT) CEO Satya Nadella mentioned something that could be concerning for 2023 and 2024's big winner, Nvidia (NASDAQ: NVDA), which has run away with a lot of AI value gains thus far. But Nadella's statement could be very bullish for a certain commodity whose associated stocks often come with high dividends for passive income.
'Not chip constrained' any more
Earlier this month, Nadella appeared on a podcast with venture capitalists Brad Gerstner and Bill Gurley. The wide-ranging, hour-plus interview dealt with AI. But while the interview was largely bullish on the prospects of AI broadly, one topic did come up that seemed to put a damper on sentiment around Nvidia.
When asked if Microsoft was still supply-constrained for Nvidia chips as it was throughout 2024, Nadella noted:
I am power [constrained], yes, I'm not chip supply constrained ... We were definitely constrained in '24. What we have told the street is that's why we are optimistic about the first half of '25 which is the rest of our fiscal year. And then after that I think we'll be in better shape going into 2026 and so we have good line of sight.
Since that statement, Nvidia's stock has been somewhat weak. That's not surprising. Since 2023, there has been far more demand for Nvidia's chips than it could supply, leading to big revenue increases and high margins for Nvidia's graphics processing units (GPUs). And Microsoft has been Nvidia's biggest customer by far, with some estimating Microsoft accounted for 20% of Nvidia's sales over the past year.
On recent earnings calls, Microsoft noted it had been supply-constrained; otherwise, its Azure cloud growth, especially for AI workloads, would have been even faster. So now, Nadella's hint that those supply constraints may be coming to an end could mean one of three things: Demand for AI is slowing, chip supply is improving, or a bit of both is happening.
Are AI improvements slowing? Is Maia ramping?
There have been some rumbles that improvements to AI large language models may be harder to come by, and the pace of innovation may slow down. These rumours have been denied by some major industry participants, but they could have an effect on AI chip purchases.
After all, if the projected returns on AI experimentation and applications are slow to show up, demand could slow down. Even if Microsoft has plenty of demand from enterprise customers, it's possible that smaller buyers of GPUs, such as mini-cloud CoreWeave or others that supply capacity to riskier AI start-ups, may be seeing less demand.
Still, most companies in the space are still pretty bullish on AI demand. One other possibility is that Microsoft sees in-house-designed Maia accelerators ramping to greater volumes in mid-2025.
Microsoft was well behind the other cloud giants, which have been engineering their own custom chips for years and use them internally to avoid being as dependent on Nvidia's expensive GPUs. This is why Microsoft buys so many more Nvidia GPUs than its cloud-computing rivals.
Microsoft, however, first introduced its Maia accelerators and Cobalt central processing units (CPUs) at the end of 2023, one year ago. So, with a year for Microsoft to hone its design and perhaps ramp up a new manufacturing supply, Nadella may merely be seeing Maia chips ramp to higher volumes in the new year, alleviating its chip constraints.
Nadella's commentary about being energy constrained seems to indicate demand is still there, at least for Microsoft's enterprise customer base, and this could be more of a Maia ramp than a downturn in AI chip demand.
Could natural gas stocks beat Nvidia in 2025?
Concerning AI opportunities that could present themselves in 2025, let's turn to the phrase that Microsoft is "power-constrained."
It has been posited that the United States will have unprecedented demand for electricity in the years ahead, with the growth of that demand outstripping that of the past 10 years, in large part due to AI data centres.
What types of clean energy can be brought on quickly to meet demand? Renewables will no doubt have a role here, but renewables don't run 24/7, and solar arrays and wind farms may be constructed in faraway places that need to be connected to the grid.
In addition, some of the biggest 2024 winners were nuclear stocks. Microsoft itself inked a deal to bring on power from the shuttered Three Mile Island facility to feed its AI data centres for 20 years.
However, shuttered nuclear facilities take a long time to get up and running. Three Mile Island itself won't be able to return to service until 2028. So, imagine the cost and time lags to get new nuclear facilities fully operational. With nuclear-oriented stocks having rocketed higher this year, there could be some disappointment in the offing.
Famous hedge fund manager David Tepper also doesn't think the AI revolution can be satisfied by nuclear for all these reasons. That leaves only one other option to quickly deploy at existing or easy-to-build facilities: natural gas. Back in September, Tepper warned, "If you're going to meet the power needs of what they need for AI, you're going to have to use natural gas."
Morgan Stanley energy sell-side analyst Stephen Byrd recently echoed that sentiment. Last year, Byrd predicted the rise of nuclear stocks, which came to pass. This year, Byrd expects natural gas stocks to benefit from the inevitable demand for power to AI data centres.
He even sees new natural gas facilities being built on the same land as AI data centres and connecting directly to them, thereby bypassing the grid and the long transmission-approval process that comes with traditional power deals.
If natural gas sees a new surge in demand, the sector's biggest stocks could do very well.
The quintessential natural gas play
If Nvidia is the quintessential AI stock, EQT Corporation (NYSE: EQT) might be the quintessential natural gas stock. EQT has amassed the largest acreage in the Marcellus and Utica shale formations of the Appalachian Basin, which is home to the largest, low-cost natural gas reserves in the US.
With the highest acreage and lowest drilling costs in the Appalachian Basin, EQT just lowered its break-even costs even more with the acquisition of midstream company Equitrans. The deal closed at the end of July.
The Equitrans acquisition turns EQT into the only vertically integrated play in the Basin, not only with production but also with gathering, processing, storage, and pipelines. This acquisition should be beneficial in several ways.
EQT will now not have to pay a pipeline operator margins to take away its gas, lowering its break-even costs to the lowest of its peer group. In fact, EQT says that post-acquisition, its break-even price is now about $2 per metric million British thermal unit (MMBtu), which is down from about $2.50 for EQT as a stand-alone corporation and roughly equal to the lowest prices seen for natural gas during the pandemic, as well as during a downturn earlier this year.
So, while most natural gas companies have to hedge natural gas prices to protect solvency in a low-price scenario, EQT says it will hedge a lot less after 2025, when its current hedges roll off.
This is because EQT is confident it can break even at low prices, whereas competitors can't afford to operate at those levels. Not having as much downside protection means EQT can realise higher prices should the price of natural gas surge, as it won't have capped upside. And if the AI revolution, coal replacement, and LNG export markets increase, there could be substantially higher natural gas prices through the rest of the decade.
While natural gas prices have just about doubled off their lows to just under $4 per MMBtu today, they did go over $9 when Russia invaded Ukraine. Trading at just 18 times 2025 earnings expectations, EQT could be one of the biggest 'AI winners" in the year ahead -- maybe even bigger, stock appreciation-wise, than Nvidia.
This article was originally published on Fool.com. All figures quoted in US dollars unless otherwise stated.
