SCP-4856
Unknown
~
medium confidence
SCP-4856
Expected annual
$16.0M
One-time setup
$61.2M
Annual recurring
$15.2M
Personnel
25
Initial capital and reserve requirements are large (roughly $57.7M one-time in this plan), driven by structural retrofits, Gen(Tau)-class AI hosting and multiple contingency reserves. Recurring annual costs are substantial (~$15.2M/yr), dominated by MTF readiness, specialist staffing, AI operations, and capital refresh amortization.
One-Time Capital Costs
Total: $61.2M
Annual Recurring Costs
Total: $15.2M/yr
Cost Scenarios
📊
Baseline
(baseline)
$15.2M/yr
Normal, uneventful year with standard operations, maintenance, staffing and amortized capital refresh.
no major breaches
routine operations and maintenance
🚨
Minor Incident
$16.7M/yr
Localized breach or iteration-caused damage requiring targeted repairs, short-term surge MTF activity, and increased forensic/medical spend.
single-location breach
limited paradox-related disruption
localized rebuild and investigation
🚨
Major Paradox Breach
$27.2M/yr
Significant causal-chain failure requiring large MTF Bootstrappers deployment, multi-site covert ops, major rebuilds and large covert cover-up operations.
causal chain does not close
widespread temporal/paradox cascade
multi-site covert operations
Personnel
25 total
| Role | Count | Notes |
|---|---|---|
| Security Officer / MTF Agent | 12 | [#6] 12 dedicated on-site guards (rotating shifts) as per staffing estimate used to compute security wages. |
| Site Security Supervisor | 1 | [#6] Site security supervisor to manage guard contingent and QRF coordination. |
| Research Scientist | 4 | [#7, #12] Dedicated researchers tasked with monitoring Forethought outputs and surgical/medical research. |
| Technician / Engineer | 6 | [#7, #15] Technicians and site engineers handling hardware maintenance, inspections and structural fixes. |
| AI Operator / Systems Admin | 2 | [#3, #7, #17] Operators and ML/ops staff for Forethought.aic, secure archival maintenance and cyber-hardening. |
Confidence Notes
Notes provide detailed line-item estimates for most categories, but large uncertainties remain around Gen(Tau)-class AI development costs and the open-ended expenses of paradox resolution; these drive medium confidence.