Skip to main content
PJM Data Center Study → Interactive

Build a Policy Package

Configure the coal moratorium, DC interruptibility program, and self-generation structure. See the cost, reliability, and emissions impact update in real time. Can you find a package that's reliable, affordable, and clean?

Coal Moratorium 165GW firm

How much coal capacity stays online. More coal = more firm capacity, but more emissions.

2032 plan retires all coal. 2028 plan keeps 10 GW. Current fleet = 20 GW.

DC Interruptibility 18% enrolled

Fraction of 30 GW DC load that agrees to curtail during grid stress events. Higher enrollment = fewer shortage hours.

0% 10% 18% 30%

5,400 MW enrolled — eliminates all deficit hours

Self-Generation Structure

How data centers source their power. Behind-the-meter removes load from PJM but may cannibalize existing generation.

Grid-Connected PPA: 30 GW new generation added to PJM supply stack.

Quick Presets
Hours of Unserved Energy per Year
--
hours / year
9 hr
100 hr
RELIABLE (0-9) AT RISK (10-100) FAILING (101+)
Annual Policy Cost
Total Package Cost --$B/yr
Coal moratorium (RMR payments) --
Interruptibility payments --
Gas-only buildout alternative $3.22B/yr
Coal Emissions Impact
--
MT CO2 / year
Social Cost @ $51/ton
--
Social Cost @ $200/ton
--
--
Cost vs. Gas-Only Buildout ($3.22B/yr)
$3.22B
--
Screening-level estimates. Costs use Brandon Shores RMR rates ($114/kW-yr) for coal retention, $40B capital for 20 GW new gas (0.0806 CRF), and interruptibility payments scaled to enrolled MW. Reliability hours are interpolated from stochastic capacity-expansion model results across 200 draws and 9 weather years. Actual outcomes depend on interconnection queue timing, weather severity, and market dynamics.
Why This Matters

The Three-Way Tradeoff

Start with "Do Nothing" and watch the grid fail: 324 hours of unserved energy, zero cost, zero emissions. Now hit "Gas-Only" — reliable at 29 hours deficit, but $3.22B/year and you still have coal emissions. The "Recommended Package" gets you to 0 deficit hours for a fraction of the gas cost, using a coal moratorium plus an 18% interruptibility program.

The slider reveals a sharp nonlinearity: interruptibility buys almost nothing from 0-5%, then eliminates nearly all deficit hours by 18%. Beyond 18% you're paying for curtailment capacity you'll never use. That 18% sweet spot is the single most important finding in this study — and it only emerges when you model the full stochastic dispatch, not deterministic peak-hour planning.

Now try "BTM Taking Nuclear" and watch what happens to reliability. Behind-the-meter sounds clean until data centers cannibalize existing nuclear capacity, removing 5 GW of firm generation from the grid. The cure becomes worse than the disease.

The Fidelity Lesson
A deterministic model would tell you "build 20 GW of gas." A stochastic model reveals that a coal moratorium + 18% interruptibility program achieves the same reliability at 14% of the cost — a policy invisible to simpler methods.