Data centers have money and power. They lack permission.

CVC DIF has taken majority control of Firstcolo, a German colocation operator. PWD recorded the deal on 14 August. It landed alongside a Frankfurt data-center exit, a mid-market rotation that suggests the firm is trading mature digital infrastructure for newer colocation exposure.

Copenhagen Infrastructure Partners has raised $3 billion for its second growth-markets infrastructure strategy. The new fund is three times the size of its predecessor. It will chase the same power-linked digital assets now drawing balance-sheet money.

On the balance-sheet side, Axe Compute is buying a 49% stake in Duos' data-center project instead of renting capacity. The build is 55 megawatts. The neocloud wants the data centers, not just the output. GDS has lifted its sales target past one gigawatt. It guided to $1.4 billion in capital expenditure. Blackstone's BREIT has put $3.3 billion into QTS.

GDS is spending ahead of contracted revenue, building capacity it expects to lease after construction. Blackstone's QTS stake moves a real estate trust into hard digital infrastructure. That position is closer to a utility than to an office building.

CalSTRS is anchoring Nuveen's Energy Infrastructure Credit strategy with up to $2 billion. The commitment is not equity in a data center. It is a loan against the cash flows those data centers are expected to support.

The generation side is moving in the same direction. Blue Energy and GE Vernova Hitachi are advancing a Texas gas-plus-nuclear plant that will feed a Crusoe data center by 2030. Small modular reactors would add capacity in 2032.

The gas turbines are scheduled for 2030. The reactors follow in 2032. The sequencing suggests Crusoe is not building a speculative power plant; it is locking in dedicated electrons for a specific data center.

None of this is early-stage venture money. Institutional capital is moving into the physical layer, and doing so fast enough to erase the capital constraint.

The next phase runs into a wall. PWD's reporting is blunt: seven in ten Americans oppose a local data center. That is not a permitting inconvenience. It is a financing risk that arrives before the offtake contract.

That figure captures sentiment across the inland and exurban markets where much of the new capacity is planned. A data center can lock in a gas interconnect, a nuclear contract, an anchor credit commitment, and still fail at a county zoning board. The approval process is local, discretionary, and increasingly organized.

The opposition is not evenly distributed. Still, the aggregate number means a developer cannot assume consent. Some communities welcome the tax base and the construction jobs. Many do not.

The projects now in diligence have solved for capital and for power. The approval risk remains unpriced. Investors are underwriting megawatts as if consent is a fixed cost. It is not.

Infrastructure, at root, is an exercise in land and permission. The new data-center funds carry construction risk, offtake risk, rate risk. A project that has all three secured can still be stopped by a planning commission.

Community engagement is a core underwriting input, not a public affairs afterthought. The funds that treat local opposition as an engineering problem will find it is a political one. It cannot be solved with a bigger transformer or a more patient capital stack.

Capital and power constraints have been priced and, in many cases, secured. The next dollar committed to a data center will be wagered on whether the neighbors let it connect.