Case studies › ETFs · Policy
Data centre ETFs are now a policy trade, whichever way the vote goes
Forty-six of roughly fifty data centre ballot initiatives passed. Fifty-six per cent of voters told the CNBC poll that AI and data centres were a crucial or major factor in their ballot decision, and sixty-five per cent of those voted Democratic. The Nasdaq fell 5.5% on 5 November, its worst single session in nineteen months, taking US$2.5 trillion off the index.
That is a policy repricing, and it happened in one session on a vote rather than on an earnings revision.
The state lever, and the concentration
Texas hosts 27% of planned US data centre capacity. The state's governor-elect has pledged a moratorium on all data centre approvals until the legislature enacts resident protections. Ohio's incoming senator said plainly that he does not work for data centres. Pennsylvania's governor signed an executive order in August to slow development and framed it as local consent.
| Input | Value |
|---|---|
| Data centre initiatives passed | 46 of ~50 |
| Voters citing AI/DC as decisive | 56% |
| Of those, Democratic vote share | 65% |
| Texas share of planned US capacity | 27% |
| Nasdaq, 5 November | -5.5% |
| Index value erased | $2.5tn |
Twenty-seven per cent in one state is the number that turns a local permitting question into a national queue question. Any developer with a Texas pipeline now has a legislative date in front of it that did not exist last week.
Why the ETF wrapper does not diversify the risk
A data centre fund holds developers, REITs, utilities and the equipment supply chain. Those are different businesses with one shared input: an approved interconnection and a permitted building. A moratorium hits all four legs through the same gate.
The supply chain leg is the worst placed. A chip or transformer supplier does not need a permit, but it ships against orders placed on the assumption that permits arrive. Its revenue lags the gate by a procurement cycle rather than leading it.
The counter-argument, and it is decent
Capital is mobile. Developers facing a moratorium redirect pipelines to welcoming jurisdictions rather than abandon them. Hyperscaler capex is contracted years ahead, so the initiatives largely hit new approvals rather than the multi-year backlog already under construction. Federal posture stays supportive.
That is a coherent case for the shock being a delay rather than a stop. It is also entirely about timing, and a delay is exactly what a discounted cash flow model with a fixed 2028 to 2030 delivery window cannot absorb.
The signal to use instead of the vote
Interconnection queue length, published quarterly by the grid operators. It is duller than an election result and it is a filing rather than a guess. A queue that lengthens while approvals tighten is the arithmetic confirmation that the political risk has become a delivery risk.
The second signal is the split between new approvals and backlog. As long as construction continues on the existing book, revenue keeps coming and the headline risk stays a headline. The moment energised megawatts plateau against announced megawatts, the delay has reached the income statement.
Where the case breaks
The roadblock case breaks if a compromise is struck quickly, which is plausible precisely because AI is now an economic development argument in the same states. A moratorium with a legislative sunset would resolve most of the risk in a quarter.
The speed-bump case breaks if a single large developer publicly defers a Texas project. Disclosure of a deferral is worth more than any ballot count, because it is the first moment the queue is measured in cancelled megawatts.
What the arithmetic says
A 5.5% index session on a vote with no earnings change is the market pricing a regulatory discount into an asset class. That discount is legitimate. It is also a discount on timing, and timing is resolved by filings rather than by politics.
Forty-six of fifty initiatives passed and 27% of planned capacity sits in one newly hostile state. The vote set the discount. The interconnection queue decides whether it was earned.
Sources
Ballot results, voter poll figures, Texas capacity share and the index move are third-party published data. State policy statements are the officeholders' own. The supply chain lag and the queue-versus-backlog test are derived observations on the disclosed process and can be re-run.
Published for information only. Nothing here is investment advice or a recommendation. This note is not prepared in accordance with the legal requirements designed to promote the independence of investment research. No position is held in any security named.