Oil costs hit five-year lows… don’t anticipate a sustained turnaround anytime quickly… why doesn’t AI vitality demand prop up costs?… electrical energy versus oil… an enormous AI pink flag from Eric Fry
Yesterday, oil costs slid to their lowest degree in practically 5 years.
Brent Crude fell beneath $60 a barrel whereas West Texas Intermediate Crude (WTI) dropped into the mid-$50s.
Each are rebounding as I write on Wednesday morning, however the general pattern right here is weak point. Brent and WTI have shed 23% and 25%, respectively, over the previous six months.
Why?
Overwhelming provide.
Right here’s JPMorgan:
On the threat of flogging a really useless horse, our message to the market has remained constant since June 2023.
Whereas demand is strong, provide is just too ample.
The info backs that up. At this time, we’ve got document U.S. crude output, sustained manufacturing from OPEC+ members, and softer demand alerts – notably from China’s slowing financial system.
Put all of it collectively, and the worldwide markets are drowning in a glut of extra crude.
If you happen to’re holding plenty of oil and gasoline shares, you must know that the massive funding banks don’t anticipate any reduction over the following two years.
JPMorgan says Brent will fall to $58 subsequent 12 months, whereas WTI is predicted to hit $54 (it fell to $55 yesterday). And searching additional out to 2027, JPMorgan doesn’t see a rebound both – simply one other roughly $1 per barrel of draw back, retaining costs depressed for years moderately than quarters.
Goldman’s estimate is comparable – it sees oil hitting $56 and $52, respectively, subsequent 12 months.
What about AI’s ravenous demand for vitality?
In 2023, U.S. information facilities consumed roughly 176 terawatt-hours of electrical energy, about 4.4% of complete U.S. electrical energy use.
Projections from Congress.gov present this might double and even triple by the tip of the last decade as AI workloads develop.
So, how can we’ve got an AI-driven vitality shortfall whereas we’re drowning in oil? It sounds contradictory – nevertheless it isn’t.
Knowledge facilities drive demand for electrical energy, not crude oil.
The overwhelming majority of U.S. electrical energy is generated from pure gasoline, renewables, nuclear energy, and coal. Oil performs a minor function in grid energy era. There merely isn’t a correlation between information heart energy consumption and oil burned at utility scale.
So, when utilities discuss information facilities stressing the grid, they’re speaking about electrical era capability, not crude oil consumption per se.
This is the reason companies just like the U.S. Vitality Info Administration forecast document highs in electrical energy demand – partly pushed by information facilities – even because the oil market is contending with an excessive amount of crude provide and delicate demand development.
So, what does that imply for traders?
It means we should always acknowledge that these two markets are distinct – demand for electrical energy will not be demand for oil.
Zeroing in on the AI/electrical energy a part of the equation, let’s flip to our know-how professional, Luke Lango of Innovation Investor.
He not too long ago laid out a three-layer framework for enjoying AI-driven electrical energy demand:
Utilities / IPPs
Personal the sellers of electrical energy AI will purchase for years. Favorites embrace Constellation Vitality (CEG) and Vistra (VST).
Nuclear & Uranium
Large reactors and SMRs are again. Purchase Cameco (CCJ) for uranium; International X Uranium ETF (URA) for basket publicity; Oklo (OKLO) and NuScale (SMR) as next-gen reactor names; Centrus Vitality (LEU) and BWX Applied sciences (BWXT) as part suppliers.
Vitality Storage / Backup
Knowledge facilities can’t go darkish. Purchase Bloom Vitality (BE) for gasoline cells; Fluence (FLNC) and Eos Vitality (EOSE) for batteries. Storage additionally accelerates time-to-power: construct the battery now, plug into the grid later.
Capex cycles finish, but when grid spend actually doubles into 2030, we’re within the early innings.
Over the previous couple of days, Luke has been zeroing in on alternatives associated to nuclear energy. Particularly, the delay in rolling out new nuclear energy vegetation:
Large Tech goes all-in on nuclear.
However conventional nuclear vegetation take 10-plus years to construct – and AI can’t wait.
Realistically, we want an answer that deploys in two to 3 years, whereas additionally fixing the problem of nuclear waste.
And I’ve discovered the one firm that does each.
I wish to cowl extra floor in at this time’s Digest, however you can watch Luke’s free research video on the company he says has “10X potential” right here.
Returning to the funding implications for oil…
If you happen to had the imaginative and prescient to purchase Large Oil on the top of the Covid panic in 2020, congrats, you’re possible up greater than 200%.
However at this time’s provide/demand economics ought to immediate you to ask…
Is that this nonetheless one of the best place for my cash over the following 12-24 months?
If you happen to’re affected person, don’t thoughts ready for the availability/demand stability to shift, and are keen to sit down by means of additional potential draw back over the following 12 months or two, oil can make sense. Simply acknowledge that costs might fall additional than anticipated.
On that observe, right here’s Yahoo! Finance:
At JPMorgan, strategists predict that with out some market stabilization efforts, Brent may change fingers within the $30s per barrel by 2027, a degree not seen because the depths of the 2020 oil crash at the beginning of the COVID-19 pandemic.
Backside line: Investing in AI/electrical energy is a vivid “inexperienced mild” at this time. However for oil and its provide imbalance, let’s create a brand new investing slogan…
As a substitute of “don’t battle the Fed,” for the foreseeable future, it’s “don’t battle the glut.”
Acknowledge the domino impact of this demand for electrical energy….
In commodities, an excessive amount of provide crushes costs. That’s precisely what we’re seeing in oil proper now.
However at this time’s dynamic within the electrical energy market is the mirror picture…
Whereas crude markets are drowning in extra barrels, the facility grid is going through an excessive amount of demand – and that imbalance is beginning to present up in costs in a really possible way for shoppers.
The chart beneath from JPMorgan tells the story clearly.
Over the previous decade, electrical energy inflation largely tracked general CPI. No drama. No standout strain.
However during the last 5 years, electrical energy costs have decisively damaged away – rising a lot quicker than headline inflation. And the forecast means that hole gained’t be closing anytime quickly.
Supply: Supply: JPAM, RBC
So, what modified to make electrical energy so costly?
You already know the reply…
AI and information facilities.
Hyperscale computing amenities don’t simply devour energy – they demand it relentlessly, across the clock.
And in contrast to oil, electrical energy can’t be stockpiled at scale or shipped in from midway across the globe when shortages seem. When demand surges quicker than era and grid capability can reply, costs rise.
That’s a irritating dynamic for high-price-weary U.S. shoppers.
Sure, drivers might lastly be catching a break on the pump. However that reduction is more and more being offset by increased electrical energy prices, which is hitting households by means of month-to-month utility payments and feeding into every part from hire to providers to native taxes.
Backside line: For tens of millions of households, electrical energy is turning into the following “can’t-avoid-it” expense.
Gasoline down…however electrical energy up. What is going to it imply for the American shopper in 2026?
Shifting from AI and the monetary stress it places on shoppers, to AI and the monetary stress it places on company stability sheets…
Because the AI increase costs forward, clever traders are beginning to ask a vital query…
The place, precisely, is the danger hiding?
In keeping with Tom Yeung, lead analyst at Fry’s Investment Report, the hazard isn’t prone to come from a sudden collapse in AI enthusiasm – however from one thing far quieter and extra acquainted…
Monetary pressure masked by accounting and construction.
From Tom:
The factor that issues us most is debt and debt-like liabilities…
Right here, we’re starting to see some worrying shenanigans that allowed companies like Lehman Brothers and WeWork to briefly cover long-term liabilities.
Tom writes that over the previous two years, a number of main tech corporations – together with Amazon, Alphabet, and Microsoft – quietly prolonged the depreciation schedules on their information heart gear.
By spreading prices over six years as an alternative of three or 4, cloud computing all of a sudden seems to be extra worthwhile at this time – regardless that the money outlays haven’t modified.
Then we’ve got the extra inventive financing constructions…
Meta’s huge $27 billion Hyperion data-center mission, for instance, was positioned right into a special-purpose car, with many of the mission “bought” to exterior investor Blue Owl Capital and leased again for as much as 20 years.
Again to Tom:
That is basically a mortgage, as a result of Meta receives money at this time in alternate for the promise of future funds.
Accounting guidelines permit the tech agency to categorise it as an working lease that is still off the books.
We see related dynamics with Oracle.
The corporate not too long ago disclosed practically $100 billion in off-balance-sheet lease commitments, nearly triple final 12 months’s determine, with little element on what these obligations entail.
Why does this matter?
As a result of generative AI nonetheless isn’t producing significant money circulate.
Actually, OpenAI now expects to burn greater than $100 billion in money by means of 2029. As enlargement accelerates, extra of that funding might come from debt – or debt-like constructions which can be simple to miss in good occasions.
Which brings us to a telling sign…
Again to Tom:
It’s significantly noteworthy that Eric bought Oracle final month for 27% good points – an organization he beneficial in September 2024 and anticipated to carry for a number of years.
That’s an enormous pink flag to me…
An organization he anticipated to carry for a number of years.
Now, that early exit isn’t a name that AI is over – Tom and Eric each consider that AI has lots extra runway.
However right here’s Tom with the takeaway that you simply and I must think about:
The monetary dangers of AI are merely rising too steep… and corporations are turning to some moderately inventive methods of hiding the true prices of enlargement.
In our expertise, it is a recipe for one thing going mistaken down the highway.
As a fast reminder, for months, Eric has urged traders to take earnings on overextended tech names, redirecting capital into earlier-stage, undervalued corporations that may profit from the AI wave with out extreme valuation threat.
He not too long ago launched a “Sell This, Buy That” that identifies a number of high-conviction swaps for navigating this setting (he offers away the names of three of his prime picks without cost).
You can check it out for free right here.
We’ll maintain you up to date on all these tales right here within the Digest.
Have a superb night,
Jeff Remsburg
Disclaimer: I personal Amazon, Alphabet, and Microsoft.
