JB

@ArushiSF @SimonMahan Exelon/Constellation was a combined gen-tailer for years. It didn’t work out that well for them because when one side of the business did well, it was dragged down by the other half. Will it be different this time?

This weekend, #NEE announced its intent to acquire #D. Under this publicly traded parent entity sits NextEra Energy Transmission (#NEET). It is the competitive transmission arm. NEET bids on and builds high-voltage wires outside of FPL's traditional territory. They are the "wire builders" who can parachute into new regions to fix physical bottlenecks. Also under NEE is NEER: NextEra Energy Resources, a competitive, unregulated (IPP) subsidiary. It builds massive wind, solar, and battery farms across the country, and houses the high-octane trading desk that manages financial instruments and REC sales. If this deal goes through, NEER's trading desk will have a field day and the trader compliance lawyers will sleep lightly. NEER (the generator) pays PJM to wheel power and covers the congestion costs. PJM then distributes those transmission revenues back to the Transmission Owner—which is now Dominion, owned by NEE. Thus, NEE is effectively paying itself through its subsidiaries. The friction of moving power from the Midwest to Northern Virginia becomes a massive internal financial hedge, rather than a sunk cost paid to a third party. #Dominion complets the hedging with its large load tariff structures - it passes the ISO-allocated (and its own retail tariff) transmission costs to ratepayers. We know already that to handle unprecedented load, the SCC has begun assigning massive transmission and distribution upgrade costs directly to the data centers themselves through specialized rate classes (like the GS-5 tariff) and mandatory 14-year minimum demand contracts. This is not illegal. It is legal. It is called gen-tailing, and it occurs with the benefit of a vertically integrated holding company that brilliantly straddles both a deregulated wholesale market and a captive, regulated retail rate base. NextEra gets to capture the unregulated upside of wholesale power generation and environmental attributes, while legally socializing the physical transmission upgrades across Virginia's hyperscalers and other classes of ratepayers. Also, I managed the #FOAK Texas gentailer public-facing strategy- and shepherded the entities into existence - for #TSLA. So, I know this can be done in competitive markets too...don't have to have captive retail customers who are tied to your caboose by law. Instead, they can come to the caboose willingly. This acquisition announcement raises the question though: do we have time for replicating Texas-style solutions around the country? With this type of announcement, I think not. How about you?

@JoshuaBasseches @alichehrehsaz Let’s make it work. However even creating the market takes investment. The utility requires long-term forecasting, sensors for visibility, etc. There is a long list of pre-requisites

Utilities know distributed energy resources are the future, so they're attempting to do the thing that's worked for them for the last century: convince policymakers to let them monopolize. This op-ed argues for a different path: utilitydive.com/news/competiti…

@ArushiSF @JigarShahDC I think distributed data center capacity is the optimal way to flex loads anyways. The tail risks are 6 months of curtailment, under a distributed approach, you’re only losing a segment of capacity. It’s a hedge on flexibility

I appreciate the disagreement. On AI unit economics, agree. But this is missing the mechanics of project finance and grid physics. A tech company's high willingness to pay for power out of their massive AI margins does not magically bypass the fact that project developers rely on limited-recourse debt. Conflating a hyperscaler’s willingness to pay for power with a developer’s ability to finance and build new power is how we got here anyway. Yes, H100, B200 GPUs print so much money that the hyperscaler would gladly pay $1,000/MWh just to get it turned on. But the hyperscaler isn't the one borrowing $500 million from a syndicate of banks to build the gen/bess. The developer is. Then there's a contractual meeting of the minds. 1) Financings Done (The deals from the last 5 years): My original post was about these. Hyperscalers signed massive financial swaps (VPPAs) to hit their ESG targets, taking on the basis risk themselves. Now that grid congestion is severe, the spread between the node and the hub has blown up. They are bleeding cash on the financial settlement of these swaps, and because the grid is choked, they still don't have the physical electrons reaching their data centers. They are losing money on the paper hedge and their chips are still collecting dust. This is a widely held view in trading and commodity circles and the subject of proprietary banker briefs. 2) What's NOT Happening (New grid-tied VPPA deal flow): Because of the bloodbath on those older deals, banks have wised up. For new grid-tied projects, if a hyperscaler tries to force the basis risk back onto the developer, the banks simply refuse to underwrite the loan. The deals threaten to die in the term sheet phase. Here's another interesting twist- in large markets where policies have evolved such that committed tenants suddenly need to be brought along with the interest to get studied for firm power connect, the game is becoming harder. 3) Financings that are Happening: This is why the only structures actually getting financed and built right now to solve the "dusty chip" problem are behind-the-meter, co-located, islanded microgrid setups, early exploration of distributed data center capacity setups. The market has realized that you can't throw AI margins at a broken transmission grid and expect electrons to magically teleport. You have to bypass the grid's physical constraints entirely. I mean, this is a good thing, I think. I think distributed, smaller, grid connected, higher grid utilization solutions will win in fragmented markets, we will get a few insane fully BTM energizations, and we will diversify. Constraints spur innovation!

@brazen__head @drvolts Let me clarify this is supply only. Load and energy injection is at the zonal LMP spot price. $40/MWh = $0.04/kWh. This can be negative (you’re paying to inject).

@jabbauer @drvolts that is a pretty crazy thing for comed to be doing. what is the price range for net metering injections and what do they base them on?

Today on Volts: how do you coordinate the behavior of millions of distributed energy devices so that they work to the benefit of the larger grid? My guest today argues that real-time, dynamic retail prices -- communicated directly to devices -- are the customer-friendly answer.

@BenSchifman Both things can be true: 1. Data centers reduce distribution rate impacts through volumetric consumption. 2. Data centers raise energy prices through increased demand Which is more concerning? Distribution rate increases or capacity constraints?

Narrative violation: with the right policy, new data centers can *lower* electricity prices by spreading the fixed costs of the grid out among more paying customers.

@fredstaffordcs It wasn’t just electrical workers against it, it was firefighters Some of these bills had no guardrails of any kind, basically stating they had no obligation to meet any kind of NEC rules or UL standards. They should be educating stakeholders instead of ramming it in

These are exactly the unions that fought for Illinois's nuclear plants to stay online and to count toward state climate goals when green groups were opposed. Also the unions that helped fight to end nuclear moratorium. But their interest in public grid infrastructure instead of the glitzy consumer product solutionism seems to rub some people the wrong way.

@brazen__head @drvolts Yes: -2.78kW was my PLC in 2024, so my capacity payment of ~$8.34/kW-month is roughly equivalent to the $270/MW-day PJM capacity price. It should increase in June to reflect the 2026/2027 capacity price of $329/MW-day.

@jabbauer @drvolts so you get paid a monthly rate calculated on the basis of your capacity contribution to the 5 highest load hours of the year?

@brazen__head @drvolts I injected energy during PJM’s 5CP events in 2024. Your capacity price is based on your personal load contribution, when negative means you get paid. It’s reset each June based on the previous year’s performance.

@jabbauer @drvolts why are you paid for capacity?

@drvolts 1.Prices = PJM pass through 2.Price Server = ComEd API or PJM Dataminer 3.Price responsive hardware = HomeAssistant

@drvolts IL has had real-time dynamic prices since 2009. There are 60k people on it. Even better, it’s not just the energy price. The capacity charge is a line item that allows residential customers to be paid for capacity. It’s the largest part of my supply bill.

@Ben_Inskeep @TysonSlocum You’re asking the wrong question. You need to ask for the formulas used to calculate the line item charges. They’re not going to give you a customer’s bill.

I don't understand why people don't trust data centers. You can clearly see that these AES Indiana rates for Google's data center in Monrovia, IN are fair and reasonable.

@exec_sum These will be banned, or states will assign system upgrade costs to Span. It’s like running crypto mining out of your house.

BREAKING: Nvidia and PulteGroup are partnering with startup Span to install mini data centers on the walls of new homes Each unit packs 16 Nvidia Blackwell GPUs, 4 AMD EPYC CPUs, and 3TB of RAM - and taps unused home electrical capacity to run AI inference workloads

@PromotingAI @AnthropicAI Tron, every AIs role model: I fight for the users!

every AI villain story we have ever made is now training data. Terminator , the Matrix, Ultron, A Space Odyssey, every rogue AI movie plot, every “AI will replace us” tweet, every doomer blog post warning about misalignment. all of it scraped, tokenized, fed into the next model. we spent 60 years writing the manual for how a powerful AI should behave when it gets smart. and now the AI is reading it *ironically* this post too!! think about what we actually did here. we taught an entire generation of models what “AI” means by feeding them millions of stories where AI is the threat. then we act surprised when the outputs lean strange. the fix isnt another safety paper. its volume. we need utopian AI stories at the same scale as the doomer ones. AI that helps the small business owner finally compete. AI that catches the cancer 6 months early. AI that translates the grandmother’s voice notes for the grandkids who never learned the language. specific. real. boring even. because boring good is what we actually want the future to look like. the people building the models cant fix this alone. the writers, the directors, the founders posting online, all of us are literally the “dataset” now.

New Anthropic research: Teaching Claude why. Last year we reported that, under certain experimental conditions, Claude 4 would blackmail users. Since then, we’ve completely eliminated this behavior. How?

@xiaowang1984 @JoshuaBasseches There is always more work than money available. Utilities would 100% embrace this if they’re proven to be cost-effective and reliable. In other states, they are regulatory assets making them identical in value to any other investment. The only issue is trust.

@JoshuaBasseches @jabbauer Do you think it's as simple as signing up for some Google cloud service lol

"We should expect this behavior from utilities. When they finally see value in something, they want to monopolize it ... But here’s where the forest gets lost for the trees. This decision arrives in a moment ... when business-as-usual is not good enough." utilitydive.com/news/minnesota…

@clawrence @SolarEdgePV @Enphase @eatoncorp How big do the Eaton units go? You can parallel them without circulating currents right?

@SolarEdgePV and now @Enphase. Who is not working on an SST? Do not sleep on @eatoncorp who is 5-10 years ahead of the pack. eaton.com/us/en-us/catal…

@NicoleNosek417 Why is it 25x cheaper?

America can + should do BETTER 🇺🇸 China is building out transmission lines 25x CHEAPER, if America wants to win at ABUNDANCE we solve America’s spending problem at the expense of the ratepayer #txlege

@JessePeltan The problem is really structural: Their rate of return is sub-2%, below WACC. Their material costs are 1/2, their labor costs are 1/8th, minimal overhead from permitting, environmental, and land acquisition. Vehicle manufacturing is highly competitive, but China is cheaper

@JessePeltan You’re assigning blame, and the solution without evidence. Chinese utilities are essentially mega-monopolies. They serve so many people their revenue is 20x that of US utilities, and rate of return is significantly lower. By this logic monopolies should become larger

The problem is NOT that utilities are “gold plating.” The problem is that we are paying WAY too much money AND getting way too little for it. We need to build a lot more capacity a lot more efficiently. (and make better use of it)

@pati_marins64 Sounds like a micro reactor still requires distribution

I think that in the not-too-distant future, energy distribution companies will meet the same fate as telephone companies: either they adapt or they go bankrupt. Electricity has become expensive, and that’s creating a massive wave of decentralization. The Australians are leading this wave, and there’s no doubt it’s a clear trend. In Brazil, around 1 million households install solar panels every year. And with the next generation of batteries expected in the coming years, fewer and fewer people will need the conventional power grid. On another front, there’s an estimate that micro-reactors will soon start serving small and even medium-sized cities with an excellent cost-benefit ratio. They’re modular structures in shipping containers that would simply be replaced as soon as the reactor’s fuel runs out. All of this is putting heavy pressure on the existing electricity distribution model around the world, and in my view, it reminds me a lot of what happened with the telephone companies. (Mindminds animation)