Durable Concrete, Fragile Statute

The second power grid has a defect the first one does not share: it can be rewritten between breakfast and lunch. A transmission line, once energized, tends to stay energized for decades regardless of who wins the next election. A tax credit exists at the pleasure of a Congress that is never more than two years from its next composition. This module is about that asymmetry, because underwriting it correctly is what separates a practitioner from a headline-reader.

Consider the two kinds of commitment a data-center project relies on. The first is a negotiated local deal — a property tax abatement from a county, an economic development agreement with a municipality, a special utility rate. These are contracts or ordinances with named counterparties and, usually, defined terms of years. They are the product of a specific negotiation, and their durability comes from that specificity: a signed abatement is hard to claw back because doing so breaches a deal and spooks every future investor the county hopes to attract. Local governments guard their reputation as reliable counterparties because their entire economic development strategy depends on it. The deal is negotiated, and precisely because it was negotiated, it is durable.

The second kind of commitment is a federal tax credit. Nobody negotiated it with you. It was enacted by a legislative majority, and it can be amended, phased out, or accelerated toward expiration by a later majority through the ordinary budget process — often on a party-line vote, with an effective date that can reach backward toward projects already in motion. The credit made you no promise. It expressed a policy, and policies are revised. This is not a hypothetical risk. The clean-electricity credits enacted in 2022 were written with long horizons, and subsequent legislation has already moved to compress those horizons — tightening the deadlines by which a project must begin construction or be placed in service to qualify, and layering on eligibility restrictions that did not exist at enactment. My read is that the exact contours will keep moving; what will not change is the direction of the lesson, which is that the fiscal grid carries statute risk and election risk in a way the physical grid does not.

Hold the two side by side. The negotiated-but-durable local deal is narrow and slow to obtain but sturdy once you have it. The legislated-but-fragile federal credit is broad and available to anyone who qualifies but revisable by people you will never meet. A serious underwriting model treats them differently. It does not book twenty years of federal credit as if it were a signed contract, because it is not one. It asks what happens to the project if the credit tier degrades, if the deadline moves, if an eligibility screen is added mid-stream. The projects that survive contact with reality are the ones whose sponsors stress-tested the fiscal assumption rather than treating it as furniture.

There is a subtler point here that repays attention. The fragility is not symmetric across the credit's life. A credit that rewards being placed in service by a date creates a cliff: a project that misses the date by a quarter can lose a slice of its economics that no amount of operational excellence recovers. This turns a tax provision into a construction-schedule risk, which is a genuinely strange thing for a tax provision to be, and it is exactly the kind of second-order effect a practitioner is paid to see. The prevailing-wage and apprenticeship conditions I mentioned earlier compound this: they are not just a higher credit tier, they are a documentation-and-labor obligation that a project must run correctly for years, and a project that stumbles on the paperwork can find the rich tier reclassified down to the base. Fragility, in other words, lives inside the credit as well as around it.

So the frame for the course is now visible in outline. Federal fiscal policy sets the hurdle rate for building at all — that was the first lesson. And federal fiscal policy is fragile in a way that local deals are not — that is this one. A practitioner holds both truths at once: the federal layer is powerful enough to determine whether the power exists, and unreliable enough that you must never mistake it for a promise. Everything that follows — the investment and production credits, the transferability market that turned them into cash, the semiconductor credit sitting upstream, and the stroke-of-the-pen change to how research costs are treated — is an elaboration of these two sentences. Read the fiscal weather first, and never confuse a policy with a contract.