The Federal Bill That Taxes the Exact Behavior States Are Rewarding
Two branches of the same government, acting in good faith, can look at the same behavior and reach opposite conclusions. Pennsylvania just told developers that bringing their own power earns them a faster permit. Five days earlier, a member of Congress introduced a bill that taxes that same power regardless of where it comes from. Nobody coordinated those two positions, and I do not think either side realizes yet how directly they are pulling against each other.
What the Bill Actually Says
Representative Andrea Salinas introduced the Data Center Community Reinvestment Act, a proposed federal excise tax of one cent per kilowatt hour on electricity used by any data center exceeding one megawatt of capacity. Projected to raise roughly $1.76 billion annually, the revenue would split across five federal programs, including the Land and Water Conservation Fund and the Housing Trust Fund. The stated goal is straightforward. Salinas has said communities should not be left footing the bill for infrastructure built to serve the AI boom.
Here is the detail that changes how every developer should read this bill. The tax applies whether the electricity comes from the grid or is generated on-site. Not just grid draw. Not just utility purchased power. Any electricity the facility uses, full stop, regardless of its source.
Why This Contradicts Everything Else Happening Right Now
Every meaningful policy move over the past several weeks points the same direction. PJM's own reforms reward large loads that bring their own capacity. Pennsylvania's new GRID requirements give faster permitting specifically to developers sourcing new power supply rather than relying on the grid. The entire industry has spent this year absorbing one consistent signal, build your own generation, secure your own supply, and you get treated better by the system.
The Data Center Community Reinvestment Act does not care which side of that line a project sits on. A developer who spent months and real capital securing on-site generation specifically to earn the faster permitting track in Pennsylvania gets taxed at the exact same rate as a developer who never bothered. The behavior every other policy lever in this space is currently rewarding gets no relief under this bill's current language.
Why I Use GTSQT to Read This, Not Just Tax Policy
I use GTSQT to describe how power feasibility actually behaves, and the Generation letter is usually treated as a physical and commercial question, does the power exist, can you secure it, does it show up on schedule. This bill is a reminder that Generation is also a policy surface, and policy surfaces do not always move in the same direction as each other just because they touch the same physical fact. A developer can solve the physical Generation problem completely, build the plant, sign the interconnection agreement, hit every timeline, and still discover that solving it did not solve the policy exposure sitting on top of it.
That is the actual decision error worth naming plainly. Bringing your own power protects a project from curtailment risk, because PJM's reforms exempt self-supplied load from priority cutoffs. It protects a project from permitting delay, because Pennsylvania's GRID requirements reward exactly that choice. It does not, under this bill as introduced, protect a project from federal taxation, because the bill was written specifically to close that door before anyone could walk through it.
Why the Bill Being Early Stage Does Not Make This Irrelevant
The bill remains at the introduced stage, referred to three House committees, and plenty of introduced bills never advance. I would resist the temptation to file this away as noise because of that. The bill's own analyst commentary makes an important point, that for the largest AI and hyperscale projects, access to executable megawatts and interconnection timing likely matters more than the tax alone. That is true for site selection. It is not an argument that the tax exposure itself is trivial once a project is built and operating for the next twenty years under whatever version of this bill, or its eventual successor, actually passes.
What This Means for Anyone Underwriting BYOC Strategy Right Now
Separate physical risk from policy risk explicitly in every model that assumes bringing your own power is a clean, complete hedge. Physical risk, whether power exists and shows up on time, gets genuinely reduced by self-supply. Policy risk, whether a future federal or state action taxes or restricts that same behavior regardless of its physical merits, does not automatically shrink just because a developer solved the engineering problem. This bill is the clearest evidence yet that those are two separate categories of exposure, and treating them as one is exactly how a well-engineered project ends up with a cost line nobody modeled.
If you want to walk your own portfolio through which of its BYOC assumptions are actually protecting against physical risk versus policy risk, I would rather do that with you now than have this bill, or the next version of it, make the distinction for you.
If that's where you are, whether you're walking through this yourself or bringing it to a board that's about to lean on one of these assumptions, grab 15 minutes on my calendar and let's set up time to go through it.
PLUS:
Grab the guide. Before advancing a data center site, there are 12 questions I make sure I can answer. I wrote them up here: [The 12 Questions Every Real Estate Professional Should Ask Before Advancing a Data Center Site].
