A Lesson on Energy: Spirit Airlines

What the unfortunate end of an airline says about asymmetric energy price exposure.

I definitely stole this meme from somewhere on the internet. Thought it was too funny…

Spirit Airlines shut down on Saturday.

It’s really unfortunate— my partner is an airline pilot and I can only imagine what that means to the employees and their families. Seventeen thousand employees out of work. The last flight landed at Dallas-Fort Worth just after midnight, and air traffic controllers said goodbye to the pilots over the radio.

There’s debate about what was ultimately to blame. Since the Strait of Hormuz closed, jet fuel prices have roughly doubled, from about $2.24 per gallon to $4.60. JPMorgan estimated the spike would add $360 million to Spirit’s annual fuel costs. That number exceeded their total cash reserves.

Spirit didn’t just die because fuel got expensive, they died because they had asymmetric risk exposure to the upside, and couldn’t absorb it.

The distinction matters, and there’s a parallel here to industrial America and electricity: if you run a factory, a smelter, a chemical plant, or any other operation where electricity is a major input cost, it could matter to you, too.

The Anatomy of the Asymmetry

Spirit’s entire business model was built on being the cheapest option. Ultra-low fares, fees for everything else, margins measured in fractions of a percent. That model works when your biggest variable cost stays predictable. When it doesn’t, you have little buffer.

Their restructuring plan, approved by a bankruptcy court in early 2025, was built on a fuel assumption of $2.24 per gallon. That assumption became obsolete in a weekend, and it wasn’t their fault. But a geopolitical event 7,000 miles away doubled the price of a commodity they were completely exposed to.

Spirit had minimal fuel hedges, so no meaningful protection against the upside scenario.

The asymmetry is what hurts. If fuel stays at $2.24, Spirit survives. If fuel goes to $4.60, Spirit dies. There is no version of $4.60 fuel where Spirit’s business model works. The downside of hedging would have been a modest premium. The downside of not hedging was the end of the company.

Now Replace Fuel with Electricity

There are industrial electricity consumers in the United States right now whose risk profile to electricity looks similar to Spirit’s risk profile to fuel.

They operate on thin margins; electricity is one of their largest variable costs. Their budgets, their capex plans, their board presentations, their procurement strategies are often built around an assumption that wholesale electricity prices stay in a historically normal range: $40 to $60 per MWh at most hubs. And they don’t want to structure long-term deals because they are “conservative” and don’t “speculate”. Which in a normal environment is a fine strategy.

But what if electricity prices don’t stay in the normal range?

Consider the math. An aluminum smelter operates at roughly $60 to $80 per MWh of revenue tolerance. A steel mill sits around $80 to $120. A chemical plant, $100 to $160. These are the prices at which electricity costs consume enough margin to threaten viability. Above those thresholds, the business doesn’t work.

Now imagine electricity going from $50 to $150 per MWh at a hub where these facilities operate. This could threaten the viability of those businesses.

But Here’s Where It Gets Worse

To be fair, Spirit’s fuel shock was a black swan. Nobody planned for the Strait of Hormuz to close. The Iran war was, by any reasonable definition, an exogenous event that disrupted global oil supply in a way that was difficult to predict and impossible to control.

The electricity price risk facing industrial consumers is not a black swan. It is a risk that is already somewhat visible in the data.

AI data centers are entering wholesale electricity markets at a rate and scale that has no historical precedent. The interconnection queues are public. The demand forecasts are published.

And the economic characteristic that makes this demand different from anything the grid has seen before is price tolerance. The Compute Heat Rate™ (CHR)¹, which measures the maximum electricity price an AI workload can profitably sustain, sits at a blended average of roughly $6,350 per MWh. That’s 127 times the current wholesale price at most hubs. Frontier training workloads can absorb prices north of $49,000 per MWh.

GPU economics dictate that the electricity cost of running a data center is a small fraction of total operating cost. A facility running Nvidia H100s spends roughly $13 per hour on GPU depreciation and about $0.11 per hour on electricity at $50/MWh. The rational economic decision is to keep running at any electricity price the grid can produce.

You can learn more about CHR by reading the paper on SSRN or digging into the full methodology at computeheatrate.com.

The Asymmetric Payoff

Spirit’s management could have hedged their fuel exposure. The cost of protection at $2.24 per gallon would have been manageable. If fuel stayed low, they’d have paid a modest premium for insurance they didn’t need. If fuel spiked, they’d still be flying.

The same math applies to industrial electricity consumers today. The cost of locking in a long-term PPA at $45 to $55 per MWh, while those prices are still available, is measurable. If the CHR thesis is wrong and AI demand stalls, you’ve locked in a market-competitive electricity price. The cost of being wrong and hedged is measurable and manageable. The cost of being right and unhedged is potentially the company.

Hans Royal is the originator of the Compute Heat Rate™ (CHR) framework. All views are his own and do not represent those of any employer or affiliated organization.

1. Royal, Hans, The Compute Heat Rate: Quantifying AI-Driven Electricity Price Tolerance
and Its Implications for Wholesale Market Repricing (February 28, 2026).
Available at SSRN: http://dx.doi.org/10.2139/ssrn.6322318