TL;DR
TVA filed a resource plan that keeps coal running to 2039, could double gas capacity, and drops wind entirely. The audience composition skews toward the industries on both sides of the ratepayer risk that plan creates.
Six grid operators lived through the same heatwave last July. A wholesale price spread separated the cheapest from the most expensive, and the post drew public engagement intensity well above the topic norm.
Gas turbine backlogs at GE Vernova and Siemens Energy are pushing data center developers toward unproven hardware. The audience composition tilted toward both the physical-build side and the alternative generation technologies competing on the same timeline.
Behind-the-meter economics, sodium-ion grid storage, and the AI compute landlord model round out a week focused on who builds what, how fast, and who pays when the bet doesn't land.
Three threads ran through the week's coverage: speed as a selection criterion for generation assets, fuel diversification as a hedge that either works or doesn't under stress, and the question of who absorbs the risk when utilities and developers optimize for timeline over optionality. Every signal this week touches at least two of those threads.
Coverage This Week
Crusoe's $1.25 billion bet on an unproven gas turbine: what happens when backlog scarcity at GE and Siemens pushes buyers toward uncertified hardware. Read →
Grid power by 2027 vs. fuel cells and CCGTs: the behind-the-meter cost case is no longer just about speed; it's about economics. Read →
SpaceXAI leasing Colossus to Anthropic and Google: the AI infrastructure landlord model and circular financing underneath it. Read →
Peak Energy breaks ground on the first US sodium-ion grid storage factory: Sacramento, $71M, 4 GWh/year, no critical mineral exposure. Read →
Six ISOs, one heatwave, a wide wholesale price spread: clean energy share correlated with lower peak prices, except where it didn't. MISO is the outlier. Read →
TVA's resource plan bets on gas and coal to power AI load: wind dropped entirely, gas could double to 26 GW, and 10 million ratepayers carry the exposure. Read →
This Week’s Signals
Each signal below traces practitioner debate and audience movement on the week's most-debated posts, what got challenged, who composed the audience, and what that pattern indicates.
TVA Dropped Wind. Who Carries the Exposure?
The ratepayer risk allocation question is the real story in TVA's resource plan, not the headline politics. The nation's largest public utility filed a plan that keeps coal running to 2039, could double gas capacity to 26 GW, and drops wind from the generation mix entirely. Data centers already consume 18 to 20 percent of TVA's industrial load and are projected to double by 2030. The board's own stated logic is timeline: gas plants build in about two and a half years, everything else takes five to ten. An IEEFA analyst who reviewed the plan called the retained coal fleet both uneconomic and unreliable. Both characterizations can be true simultaneously. The speed argument is legitimate as far as it goes. But speed and diversification aren't the same trade. Most grids racing this demand curve are hedging with a mix. TVA concentrated risk instead of spreading it, and the 10 million ratepayers carrying this fleet for the next two decades had the least say in the plan.
On a constrained impression base, save activity ran at 3.33x the typical rate for wind content — reference-dominant. The composition mirrors both sides of the ratepayer risk allocation question: the industries making resource plan decisions and the generation-side firms whose technology was excluded from it both concentrated at elevated rates. Renewable Services viewers landed at nearly 3x the topic baseline, and Renewable Energy Power Generation viewers at nearly 3x as well, suggesting the audience shape matches the structure of the argument itself. (Composition: Renewable Services 3% vs 1.04%; Renewable Generation 5% vs 1.84%; saves 3.33x; CXO/VP 18% vs 16.79%.)
Same Heatwave, 3.5x Price Spread
Six grid operators lived through the same heatwave last July. Their wholesale prices diverged by a factor of 3.5. NYISO, running almost no wind or solar, averaged $140/MWh in real-time. ERCOT, above 30% clean power, paid $40. The post lays out the full lineup: NYISO at 1% clean and $140, ISONE at 1% and $95, PJM at 4% and $82, MISO at 11% and $134, SPP at 28% and $58, ERCOT at 34% and $40.
MISO is the outlier that earns its own paragraph. Double-digit clean share, yet priced close to NYISO. The post flags why: MISO's own reliability monitor has noted that wind output falls to a fraction of nameplate during summer peak hours, exactly when the grid needs it most. Clean percentage on an annual slide isn't the same as clean megawatts clearing at 5 PM in July. The analysis frames this not as a clean-versus-fossil argument but as a fuel diversification question: grids running wind, solar, and storage that actually perform during peak hours have more cheap megawatts ahead of expensive peaker plants in the dispatch stack.
Reactions ran +90.0% vs the topic norm (n=341), which suggests this framing cut through in a way that typical solar or renewables content does not. The composition aligns with an audience whose project economics are directly shaped by whether clean capacity clears under peak stress: Services for Renewable Energy viewers concentrated at nearly 2.5x the topic baseline, and Renewable Energy Power Generation viewers at over 2x. (Composition: Renewable Services 3% vs 1.24%; Renewable Generation 4% vs 1.84%; CXO/VP 16% vs 16.79%; reactions +90.0% vs topic norm (n=341).)
The Turbine Bottleneck Pushed Crusoe Toward Unproven Hardware
Crusoe committed $1.25 billion to a gas turbine from Boom Supersonic, the supersonic jet company, for 29 turbines built on an aircraft engine core still finishing certification. Not a single unit has shipped. Two years ago, this deal would have been unfundable at scale. Today, missing the compute buildout is a bigger career risk than betting on hardware that ships late.
The post grounds this in the supply constraint. GE Vernova's backlog exceeds 100 GW, tightening through 2030. Siemens Energy holds a record €154 billion order book with a four-year wait, the longest in company history. Turbines remain the only generation technology that can be built on the timeline the data center buildout actually needs. SMRs don't enter meaningfully until after 2030. Solar and wind still need battery backup at a scale that doesn't exist. Geothermal has real capital but is early. Meta already runs its El Paso site on 800+ mobile mini turbines rather than waiting on GE. The analysis frames the broader pattern: buyers don't take risks because they've gotten braver. They take risks when the safe option runs out and every alternative is still years from ready.
The audience that showed up matches firms with structural exposure to both the physical-build side and the alternative generation technologies competing on the same timeline. Construction viewers concentrated at nearly 3x the typical rate for gas content, and Renewable Energy Power Generation viewers at nearly 2.5x. The post drew 48 comments and impressions +75.0% vs the topic norm (n=255), pointing to elevated engagement intensity on the supply constraint thesis. (Composition: Construction 4% vs 1.47%; Renewable Generation 4% vs 1.62%; CXO/VP 16% vs 16.79%; Enterprise 30% vs 19.71%; impressions +75.0% vs topic norm (n=255).)
Field Notes
Sodium-ion grid storage: Peak Energy broke ground on the first dedicated US factory in Sacramento. No lithium, cobalt, or nickel means no critical mineral exposure, but the chemistry is under 1% of global production. The bet is on where the cost curve goes, not where it is today. Read →
AI compute as a landlord model: SpaceXAI leasing Colossus to Anthropic and Google inverts the infrastructure thesis. When capacity built for Grok ends up running its two biggest rivals, the question shifts from who trains the model to who owns the physical layer underneath it. Read →
Behind-the-meter economics: The behind-the-meter cost case is shifting from a speed argument to a cost argument. PJM capacity price increases between delivery years accelerate the math. Read →
The through-line across this week's signals is risk concentration. TVA concentrated generation risk in gas and coal. Crusoe concentrated technology risk in uncertified hardware. Grids that concentrated in one fuel source paid the most when demand spiked. The pattern is consistent: speed as a selection criterion works until the bet doesn't land, and the question of who absorbs that downside is rarely answered before the commitment is made. If that dynamic is shaping procurement, resource planning, or capital allocation decisions at your organization, that's a conversation worth having before the exposure is locked in.
Reply if any of this is playing out at your company, or contradicting what you're seeing on the ground. Every reply goes directly to our analyst desk and feeds our intelligence.