Technical Advisor · Houston, Texas

We measure ourselves
by impact.

22GW

of data center and behind-the-meter generation capacity advised on or independently reviewed since 2024. Approximately 11,700 MW of it behind-the-meter or permanently islanded.

We make the technical case bankable.

We remove the information asymmetry between the party with power to sell and the party with load to serve, so the deal can be underwritten rather than argued.

Behind-the-meter gas generation
Reciprocating engine and turbine plants, grid-forming BESS
11,210MW
Grid-connected large load
138 kV to 345 kV across ERCOT, PJM, IESO, SPP and the NEM
7,950MW
Utility-scale thermal
Combined-cycle power island serving co-located load
1,200MW
Firm geothermal
Baseload offtake structuring and dynamic modeling
1,060MW
Fuel cell
Solid oxide, islanded and grid-parallel
700MW
Nuclear SMR
Pod-to-grid, islanded through to interconnected
50MW

Megawatts, rounded. Each project counted once under its primary technology. Some of this capacity reached financial close. Some did not. We were at the table for all of it.

Case Studies

Four pairings, one question

Islanded, hybrid, and grid-connected. One question each time: will this power actually serve this load?

The Pairing
A gigawatt-class data center campus served by an on-site fleet of industrial and aeroderivative gas turbines, distributed across multiple load feeders at medium voltage.
What Was at Stake
The offtaker needed to know what happens to the racks when a bolted three-phase fault lands on the load side. Not in principle — in milliseconds, with real machine dynamics, on the actual feeder configuration. Without that answer, no one could sign.
What We Did
Eradeh built a full electromagnetic transient model of the generation, the step-up transformers, the feeders, and the load, then ran the fault at the worst credible location and duration. We measured generator currents at the machine terminals, voltage and current at transmission voltage, and the voltage seen at every load feeder on the campus — capturing exactly how the plant behaves, feeder by feeder, from fault inception through clearing and load response.
The Outcome
A defensible, measured picture of ride-through behavior that both sides of the transaction could rely on — and a clear line of sight to the machine parameters that needed vendor confirmation before final design.
The Pairing
A several-hundred-megawatt AI campus proposed to run islanded on a modular solid oxide fuel cell platform, with no grid behind it.
What Was at Stake
Two questions the market had not answered. Can a fuel cell platform follow AI workload transients? And what does the plant actually have to cost to deliver the availability the campus requires? The vendor had evidence. The arithmetic behind it had not been tested by technical experts.
What We Did
Eradeh reviewed the platform architecture, the load-following evidence, and the pricing basis, then developed a medium-voltage distribution concept against the campus load. We graded every finding by confidence — supported by vendor material, derived by our own calculation with the arithmetic shown, or unresolved and tracked to a formal RFI. One finding moved the deal, because a fraction of a point of assumed cooling efficiency turned out to be worth several additional power blocks and hundreds of millions in capital — a number that had gone unexamined.
The Outcome
The client entered the next phase knowing which assumptions carried the money, which vendor claims held up, and which questions had to close before design freeze.
The Pairing
A hyperscale data center with a multi-year wait for utility interconnection, served in the interim by a hybrid islanded plant: combustion turbines, reciprocating engines, and a grid-forming battery system, exporting into the campus substation at distribution voltage.
What Was at Stake
A bridge-to-power plant has to deliver utility-grade reliability with no utility behind it. The battery is doing four jobs at once — forming the grid, absorbing fast data center load swings, covering the loss of the largest unit, and black-starting the plant. Every one of those has to be right, and the redundancy philosophy across the prime movers has to add up to the availability the offtake contract promises.
What We Did
Eradeh reviewed the design basis for the load side: generation redundancy across both prime mover types, the grid-forming control philosophy, the black start sequence, protection and controls, and the interface to the campus substation — including the provisions that keep a future utility interconnection viable rather than foreclosed.
The Outcome
A clear read from our technical experts on whether the plant as conceived could hold the reliability commitment at the point of delivery, and where the design had to tighten before it could.
The Pairing
An AI campus in the high hundreds of megawatts to gigawatt class, connecting directly to the transmission system at high voltage, with a market operator as the party that has to say yes.
What Was at Stake
Large loads now face the scrutiny that only generators used to get. The operator needs to see how the campus behaves during a system disturbance — whether it rides through, how it recovers, and what happens to the network if it drops. An AI campus is not a static load, and a model that treats it like one will not survive review. A weak submission means conditions, delays, or a connection that does not happen.
What We Did
Eradeh developed an aggregated large load model representing how the campus actually behaves electrically, using composite and user-defined load modeling for system-level dynamics alongside electromagnetic transient modeling for the fast behavior that positive-sequence tools cannot capture. Our technical experts then reviewed the full set of dynamic simulation results before submission, identifying and correcting multiple technical findings while there was still time to fix them.
The Outcome
A model and a study package the operator could work with, submitted right the first time — and a client who knew what their load looked like to the grid before the grid told them.

We work both sides of the seam between generation and load — and we prove the pairing holds before the first breaker closes.

Bring us the pairing. We'll make the technical case bankable.

Start a conversation →