Power Strategy
Utility Power vs. BTM Generation
A level-headed look at when grid power still wins, when BTM is the only realistic path, and when the right answer is both.
The definitions
Utility power (front-of-the-meter) means the data center buys electricity from the grid. The utility, or a transmission owner, builds the substation, transmission tap, or line extension, and the site takes service at a defined voltage under a tariff or special contract. Generation, reserves, and reliability are the utility's job, shared across its whole system. On site, the owner's responsibility starts at the point of interconnection. Backup generators and UPS are there to ride through outages, not to supply energy day to day.
Behind-the-meter (BTM) generation means the data center makes some or all of its own power on its own side of the utility meter, or with no meter at all. Typical sources are reciprocating engines, aeroderivative or industrial gas turbines, fuel cells, or solar plus storage, usually on natural gas. The owner, or a developer it contracts with, is responsible for fuel supply, air permits, redundancy, maintenance, and keeping the lights on. If BTM is the primary source, the owner has effectively become a small utility.
The difference isn't really about technology. It's about who carries the obligation to serve, and who carries the risk when something goes wrong.
When grid power still wins
Grid power is still the default. When it's available on a schedule that works, it's hard to beat:
- Reliability through diversity. A large interconnected grid draws on hundreds of generators and many transmission paths. Matching that on site takes N+1 or N+2 redundancy and fuel security, and that costs real money.
- Lower all-in cost in most markets. A data center that runs flat 24/7 is an ideal customer for an industrial tariff. Once you add capital, fuel, O&M, and redundancy, self-generation rarely beats a competitive large-load rate unless gas is very cheap and the grid is very expensive.
- Simpler permitting. A substation and a service agreement avoid the air permitting path that BTM requires: state permits, NSR or PSD review, and possibly Title V. That path can take longer than expected and attract opposition.
- Sustainability claims. Grid supply combined with PPAs, RECs, or clean transition tariffs supports hyperscaler carbon commitments far better than on-site gas generation can.
- Financing and offtake. Lenders and tenants understand utility service. A campus that depends on a private gas plant has to explain its fuel, emissions, and operations risk at every diligence stage.
If the utility can deliver the capacity you need, close to when you need it, at a rate you can live with, take it.
When BTM is the only realistic path
BTM makes sense once that "if" fails, and right now it fails often:
- The interconnection timeline breaks the business case. Large-load studies, transmission upgrades, and generation shortfalls in a utility's IRP can push energization well beyond when the tenant needs capacity. If the grid date is later than the lease date, the grid is not an option, however good the rate.
- The utility can't or won't serve the load. Some utilities are capping new large loads, demanding big collateral, or adding minimum-take tariffs that shift stranded-asset risk onto the customer. Some jurisdictions have formally paused data center growth.
- The site has gas but no wires. A location near a major pipeline but far from high-voltage transmission can support gas generation years before a transmission line could be built.
- Speed is the strategy. For AI training capacity, a year of delay can cost more than the extra cost of self-generation over the life of the plant.
BTM is rarely the cheaper option. It's chosen because it's the faster option, or the only one available. Go in knowing that you're taking on fuel, permitting, emissions, and operations risk that the utility would otherwise carry.
When the answer is both
The most common outcome today is a hybrid. There are three main patterns:
1. Bridge-to-grid
BTM generation powers the campus until utility service arrives. After that, it moves to backup, peaking, or demand response, or gets redeployed elsewhere. The key design question is the exit: what happens to the assets, the permits, and the gas contract once the grid is live?
2. Grid primary, BTM for flexibility
The site interconnects as a curtailable or flexible load. It agrees to drop off the grid during system peaks and run on-site generation instead. Utilities and grid operators can often connect flexible loads much sooner than firm ones, because the load doesn't add to peak demand.
3. BTM primary, grid as backstop
On-site generation carries the base load, and a smaller grid connection supplies standby, startup, and maintenance power. This can cut down on on-site redundancy, but standby tariffs and interconnection rules decide whether the numbers work.
Hybrid designs are more complex to engineer, permit, and contract. In exchange, they give you a delivery date you can promise and a long-term cost you can defend.
A quick decision frame
| Question | Points toward grid | Points toward BTM |
|---|---|---|
| Can the utility energize by your required date? | Yes | No, or uncertain |
| Is there firm, affordable gas supply at the site? | No | Yes |
| Is there high-voltage transmission nearby? | Yes | No |
| How hard are the carbon commitments? | Strict | Flexible, or offset elsewhere |
| How hard is the air permitting? | Difficult or contested | Workable |
| How much is a year of delay worth? | Modest | Very large |
| Will the owner or developer run a power plant? | No | Yes, or through a partner |
If your answers point both ways, which is common, you're looking at a hybrid. The work then becomes designing the transition from BTM to grid on day one, not as an afterthought.