Marginal cost of adding Tier 3 to existing institutional infrastructure is effectively zero

The cost case against Tier 3 deployment usually treats it as new infrastructure. It is not. Most institutions already operate the substrate Tier 3 needs, at substantial idle capacity:

• More than half of institutional server capacity sits idle; only ~40% of data centers measure server utilization at all and ~25% of physical servers are entirely comatose. Where utilization is measured, on-premises enterprise servers run at 12-18% of capacity (S-0006).
• Networks run at ~26% average utilization globally; Internet2 maintains over 50% headroom by policy and flags concern at 30% (S-0007, S-0008).
• Internet2 bandwidth contracts are flat-rate based on institutional scale, not traffic volume (S-0008).

Against that idle capacity, Tier 3 protocol nodes are operationally trivial. A BitTorrent seeding daemon averages 9-14 MB resident memory; a Tor relay needs 512 MB and 10-16 Mbps (~0.016% of a 100 Gbps campus connection); Forgejo runs as a single binary at 100-150 MB resident memory; BitTorrent's WebSeed specification allows any existing web server to function as a seed with no software modification (S-0097, S-0098, S-0099).

Universities are already running these protocols as routine background operations: TU Dortmund's student-run Matrix homeserver (S-0100); TU Dresden's Matrix for 18,000 users (S-0101); MIT's SIPB-operated Mastodon and Forgejo (S-0102); Tor relays at over 45 universities (S-0098); 298+ TB of research data on Academic Torrents at zero central infrastructure cost (S-0097).

For an institution starting from nothing, a Hetzner CX23 instance costs ~$46/year. For an institution with existing infrastructure, the marginal cost is closer to zero — invisible in the noise floor of the existing operating budget. The 1:10:100 cost heuristic (M-0004) sets prevention as the cheapest column; this Claim documents that the cheapest column is, in this case, effectively free.