Yves here. No doubt, many here remember how Enron drove up electricity prices massively in California. If you don’t think that will happen with AI hyperscalers becoming energy traders, I have a bridge I would like to sell you. It is a virtual certainty that, like most large corporate Treasury departments, that these new power traders will be profit centers and the traders’ pay will reflect how they perform.
The Los Angeles Times was one of many that chronicled the profitable games Enron traders played at the expense of California electricity customers. From the start of a 2002 account:
One of Enron Corp.’s favorite trading strategies during the California electricity crisis was like booking an airline ticket for a flight you don’t intend to board.
It’s a waste of time and money unless you’re sure the flight will be overbooked and the airline will have to dish out rewards to passengers who agree to stay home.
Enron–and, possibly, other energy traders–worked variations on this theme to collect special fees from the California Independent System Operator, the embattled traffic cop for the state’s power grid following deregulation.
Sometimes Cal-ISO would pay Enron premiums not to use power that the firm didn’t really need in the first place. Sometimes Enron would exploit California’s emergency price caps, buying power at the capped price and then selling it at huge profit out of state, where there were no price caps.
Enron’s trading strategies were described in memos released Monday by the Federal Energy Regulatory Commission. The memos, written by lawyers for Enron, detailed an array of trading methods that went by such swashbuckling nicknames as Death Star, Wheel Out, Fat Boy and Get Shorty.
California officials have pounced on the Enron memos as proof that energy traders and freelance power generators–mainly from out of state–were manipulating California’s energy markets, raising prices and even triggering blackouts.
Oh, and the attempted defense? These were common trading gimmicks, nothing to see here, move along. Which is even more reason to worry that these new hyperscaler traders will follow in Enron’s footsteps. Admittedly, California made it easier to be fleeced via its poorly-devised deregulation. But how well will the schemes across the US hold up when faced with very large traders who can push prices around?
By Tsvetana Paraskova, a writer for Oilprice.com with over a decade of experience writing for news outlets such as iNVEZZ and SeeNews. Originally published at OilPrice
- Meta, Microsoft, and Apple have either requested or obtained authorizations from the Federal Energy Regulatory Commission (FERC) to sell wholesale power.
- U.S. power utilities are investing a record amount of money into transmission and grid connections, but uncertainty about the size of demand raises investment risks.
- The venture into power trading from hyperscalers could give utilities more certainty that their future new capacity will find customers.
America’s hyperscalers are looking to ensure the electricity for their huge data centers by entering the supply side of the market—power trading.
Meta, Microsoft, and Apple, to name a few, have either requested or obtained authorizations from the Federal Energy Regulatory Commission (FERC) to sell wholesale power.
Meta Platforms, for example, seeks to incentivize long-term commitments in power-generating capacity by expanding into power trading and having the flexibility to contract electricity from future power plants, according to Urvi Parekh, Meta’s head of global energy.
Currently, power plant developers are careful about committing investments in the long term, which is not enough to meet the demand from AI and data centers.
Power capacity developers “want to know that the consumers of power are willing to put skin in the game,” Meta’s Parekh told Bloomberg in an interview last week.
“Without Meta taking a more active voice in the need to expand the amount of power that’s on the system, it’s not happening as quickly as we would like,” the executive added.
U.S. power utilities are investing a record amount of money into transmission and grid connection. But current forecasts of AI-driven power demand vary so much that there is a massive margin of error, analysts and utility officials told Reuters Events in June.
The U.S. market faces “a moment of peak uncertainty,” according to Rebecca Carroll, senior director of market analytics at energy advisor Trio.
Electric utilities face a high degree of uncertainty over future revenues as the boom of AI data centers generates widely varying forecasts of peak demand in many areas across the country.
If utilities overestimate their future demand, they risk overbuilding new capacity that will not be met by consumption. A possible overbuild would come at the expense of the American ratepayers, who have already seen electricity prices rising at a faster pace than U.S. inflation over the past three years.
The era of stagnated power demand in the United States ended about two years ago when hyperscalers started the race to develop AI-driven solutions and build huge data centers across the U.S.
America’s five largest hyperscalers are set to hike spending on data centers by 50% to over $300 billion in 2025, according to estimates by Wood Mackenzie.
U.S utilities have already committed to add 116 gigawatts (GW) of large load to their networks, equivalent to around 15% of U.S. peak electricity demand in 2024, the energy consultancy reckons.
The venture into power trading from hyperscalers could give utilities more certainty that their future new capacity will find customers.
“We’re seeing a breakdown between the demand and supply sides of the market, with the biggest actors playing on both sides,” WoodMac’s Ben Hertz-Shargel told Bloomberg.
“To better orchestrate growth, you need some of the largest buyers of electricity to actively support the buildout of the supply side.”
If Meta and other tech giants commit to long-term purchase agreements, power capacity developers will be more willing to invest in long-lead plant construction.
Meta alone is investing $600 billion in the U.S. by 2028 to support AI technology, infrastructure, and workforce expansion, the company said earlier this month.
The company says it’s making additional investments in energy supply and these have so far helped to add more than 15 GW of new energy projects to the grid across 27 states, representing more than $16 billion in capital investments.
It’s not only clean energy that will power the huge data centers of the hyperscalers—natural gas will play an important role, too.
For example, Meta’s new data center in Richland Parish in Louisiana, worth over $10 billion, will source electricity from three new gas-fired power plants after the Louisiana Public Service Commission this summer approved an agreement that paves the way for Entergy Louisiana to build these plants.
“Importantly, Meta is paying its share of the costs for the infrastructure needed to support its operations, ensuring that other customers are protected from those expenses,” Phillip May, Entergy Louisiana president and CEO, said in August.
12 comments
Comments are closed.
This article considers mostly electricity expenses which comes from new power capacity to be built. Only the last paragraph deals very lightly with infrastructure investments i guess in transport and distribution. It is indeed possible to shield, at least in part, the consumers from the energy appetite of “hyperscalers”. If investments in new utility power and infrastructure goes 100% on hyperscalers ability to pay, then both utilities and grid managers face a risk if the AI boom sours. So here, how these will power traders play? It will almost certainly result in higher electricity prices because the traders will seek sweet deals with utilities as explained in this blog post:
The post titled “What Will Data Centers Do To Your Electric Bill?” By Severin Borenstein merits a full read.
“If investments in new utility power and infrastructure goes 100% on hyperscalers ability to pay, then both utilities and grid managers face a risk if the AI boom sours.”
Do you mean ‘if’ the AI boom sours — or ‘when’?
The AI bubble and its impacts on the u.s. utility companies is beginning to appear to me like a mad competition that could lead to further consolidation of the tech industries and perhaps of the u.s. utility companies as well. While the AI bubble is inflating the management and other people “in-the-know” will be able to make a lot of easy money at the expense of index stock nest-eggs and perhaps the u.s. treasury and some lucky winners can grab up utility companies consolidating them to wreck havoc on entire regions of the Grid.
Its all a great potlatch of burning blankets and killing slaves to bring losers to ruin and raising the grandeur of winners. The u.s. Populace will be the biggest losers.
the great potlatch
in my novice assessment…under the current PE driven system…bubbles in the flutes.
… sans the ‘poison pills’ of yonder years…likely a fleecing scenario… leaving the ‘herd’ again
with the rawness of indignation
Extraordinary evil, to burn through the inherent wealth of the planet, and baste it to boot, in the alleged drive for a machine smarter than a firemonkey. Leveraging AI’s failures into financial control of energy resources makes all-too-perfect sense. Whoever wrote this episode should die
A nugget for the tumbler from John Robb:
> “We won’t reach AGI with LLMs.”
>> He’s right. We can create Social AIs with LLMs though, and they are useful in nearly every setting. In contrast, AGIs are too dangerous for almost all types of work.
It could be that LLMs have reached their peak capability, and all the rest is coprophagia and marketing. If that’s the case, then every bit, every single bit, of energy poured into them is a blow against a habitable planet.
It’s a looting operation. Just like Enron. Why bother with normal business when you can muscle in and effectively poll tax people through increased utility prices.
I’m coming to the conclusion that all these fads and schemes, from Enron to CDS funds to Bitcoin to AI are at their core just marketing scams to get pension funds to shovel money into carpet bags. The taxpayer on the hook for everything by default. These scams get us nowhere but they keep working so they won’t stop. After AI pops there will be yet another glitter ball bubble.
The FIRE sector needs to be crushed with regulation before we are all pauperised.
Couldn’t this possibility be the best reason for capping the power of the big tech conglomerates, reducing their control over their IPRs, and breaking them up into smaller clearly defined units with fewer but clearly defined business constraints on the basis of America First and grounds of national security, if nothing else?
I see the tail wagging the dog here. The massive power requirements necessary were baked in from the get go.
That these companies are now becoming power traders seems a logical progression (if not planned all along), with the “need” being baked into The Narrative. The public be damned.
All at a time when needs be that we consider it necessary to reduce power consumption.
The argument of this article appears to me to be disjoined. I cannot grasp why it is touting the benefits of big AI firms becoming energy traders:
The venture into power trading from hyperscalers could give utilities more certainty that their future new capacity will find customers.
I do not understand how the conclusion is derived from the premises. Why does an increased number of traders raise the demand certainty of utility companies?
“To better orchestrate growth, you need some of the largest buyers of electricity to actively support the buildout of the supply side.”
For this to be true, Microsoft, Meta, Oracle, &co would presumably have to become providers of energy, i.e. build their own power plants, not just trade electricity. Otherwise there is no meaningful “buildout of the supply side”.
Of course there is a hint that they may also go somewhat in that direction:
Meta is paying its share of the costs for the infrastructure needed to support its operations
but what this exactly entails is left undiscussed. Furthermore, this just happens in one example, and I again fail to see the relation with Meta becoming a trader.
I don’t know how the wholesale market functions in the US but if it is similar to European markets then there might be a big difference. As traders they would compete with commercializing companies and by doing so would rise price for all users, specially in data centre peak power hours, specially if and when they make the most expensive utilities to enter and supply the market: high prices shared by all users. If instead the data centres have to negotiate individual contracts with one or few utilities, the quotes they get aren’t negotiated in the wholesale market and consumers might avoid the pooling of the most expensive utilities that have to enter to feed data centre peaks. I am not sure If this explanation makes things clearer or more obscure.
The power consumption is surely the lessor problem created by datacentres.
Millions of gallons a day of water pulled, heated, and discarded makes me worried.
My local power co-op is scrambling to build-out solar micro-grids and is designing a tidal-generator — not to be energy independent but because they expect the public hydroelectric utility that provides the bulk of our electricity to raise prices to nose-bleed levels due to cash-rich AI server operators bidding them up. Co-op members can buy shares in the solar and tidal generation that will give a partial offset of hydro price spikes.
This is looking like Enron and the California ISO all over again, and I strongly doubt that a privileged Meta exec like Urvi Parekh gives a rat’s a** about the poors — because markets!
If there was any possibility that this would benefit the general public it would not be considered because markets.
Look at what has been done by the Trump administration when it comes to regulators, inspector generals of Government agencies, the DOJ…
It’s full on looting and this is simply another example, with a small figleaf.
Once you drown the Government in a bathtub it’s a good idea to drain the water and dispose of the corpse, leaving it around like this,well it smells bad and looks worse.