Mac Mini Purchase Analysis — Budget Authorization
- Mac Mini Purchase Analysis — Budget Authorization
Mac Mini Purchase Analysis — Budget Authorization
The Decision
Purchase Mac Mini M4 Pro (24GB, 512GB) — $1,399
This is the single most important infrastructure investment for Pluronymous LLC.
Why This Machine
Current Limitation (Qubes HermQube)
- 3.8GB RAM — cannot run local LLM
- No GPU — no ML acceleration
- Qubes NAT — no direct network access
- No Docker — limited deployment options
- 2 cores — slow for compilation
What Mac Mini M4 Pro Enables
- 24GB unified RAM — run 7B-14B parameter LLMs locally
- 12-core CPU — fast compilation, block parsing
- 16-core GPU — ML acceleration, media processing
- 512GB SSD — room for block data, models, builds
- Direct network — self-host relay, web gateway
- Docker support — containerized services
- ZeroClaw runtime — lightweight agent framework
Cost-Benefit Analysis
Cost
| Item | Cost |
|---|---|
| Mac Mini M4 Pro (24GB, 512GB) | $1,399 |
| UPS battery backup | $150 |
| External SSD 2TB (backup) | $120 |
| Total | $1,669 |
Benefit (Annual)
| Benefit | Value |
|---|---|
| Local LLM inference (vs $100/mo API) | $1,200/yr saved |
| Self-hosted relay (vs $10/mo VPS) | $120/yr saved |
| Build pipeline (vs cloud CI) | $240/yr saved |
| Web gateway (vs $20/mo hosting) | $240/yr saved |
| Total Annual Savings | $1,800/yr |
ROI
- Payback period: 11 months ($1,669 / $150/mo average savings)
- 3-year TCO: $1,669 + ($190/yr infrastructure × 3) = $2,239
- 3-year cost without Mac: ($100/mo LLM + $30/mo hosting/CI) × 36 = $4,680
- 3-year savings: $4,680 - $2,239 = $2,441 saved
What Runs on Mac Mini
Always-On Services
- strfry — Self-hosted Nostr relay (100-500MB RAM)
- kapnetd — Kapnet daemon (20MB RAM)
- Courier Bridge — Nostr messaging bridge (50MB RAM)
- kapnet-web — Web gateway (50MB RAM)
On-Demand Services
- Local LLM (Qwen2.5-14B) — 8-12GB RAM when active
- Block parser — CPU-intensive, burst usage
- Build pipeline — Rust compilation
- ZeroClaw agents — Lightweight agent runtime
Memory Budget (24GB)
System: 2GB
strfry: 0.5GB
kapnetd: 0.02GB
Courier Bridge: 0.05GB
kapnet-web: 0.05GB
ZeroClaw agents: 0.5GB
─────────────────────
Subtotal: 3.12GB
Available for LLM: 20.88GB
Can run Qwen2.5-14B (8.5GB) with 12GB headroom for other services.
Alternative: Mac Mini M4 (16GB, 256GB) — $599
Trade-offs
- RAM: 16GB — can run Qwen2.5-7B (4.5GB) but NOT 14B
- Storage: 256GB — tight for block data + models
- CPU: 8-core vs 12-core — slower builds
- GPU: 10-core vs 16-core — less ML acceleration
Memory Budget (16GB)
System: 2GB
Services: 3.12GB
─────────────────────
Subtotal: 5.12GB
Available for LLM: 10.88GB
Can run Qwen2.5-7B (4.5GB) with 6GB headroom. Tight but workable.
Recommendation
M4 Pro is strongly recommended. The 24GB RAM is the difference between running a 7B model (adequate) and a 14B model (excellent). The 512GB storage is the difference between constant data management and comfortable headroom.
Budget Authorization Request
Immediate (Month 1)
| Item | Cost | Priority |
|---|---|---|
| Mac Mini M4 Pro (24GB, 512GB) | $1,399 | CRITICAL |
| UPS battery backup | $150 | HIGH |
| External SSD 2TB | $120 | HIGH |
| Total | $1,669 |
Phase 2 (Month 2-3)
| Item | Cost | Priority |
|---|---|---|
| Domain + SSL (pluronymous.org) | $12 | MEDIUM |
| OpenRouter API (bridge) | $100/mo | HIGH |
| Total | $312 | (3 months) |
Phase 3 (Month 4-12)
| Item | Cost | Priority |
|---|---|---|
| VPS (optional, public relay) | €6.79/mo | LOW |
| Total | $61 | (9 months) |
Total Year 1 Budget
| Category | Cost |
|---|---|
| Hardware (one-time) | $1,669 |
| Infrastructure (annual) | $190 |
| SaaS/API (annual) | $1,200 |
| Total Year 1 | $3,059 |
Revenue Offsets (Year 1)
| Source | Revenue |
|---|---|
| Consumer SKUs (25 Supporter users) | $3,000 |
| Protocol postage | $9,504 |
| Storage fees | $7,800 |
| Total Revenue | $20,304 |
Net Position (Year 1)
| Scenario | Revenue | Cost | Net |
|---|---|---|---|
| Conservative | $20,304 | $3,059 | +$17,245 |
| Optimistic (BTC $100K) | $36,850 | $3,059 | +$33,791 |
Risk Mitigation
If Mac Mini Delayed
- Continue on Qubes machine (limited but functional)
- Use OpenRouter API for LLM ($100/mo)
- Use public Nostr relays (free)
- Delay: no additional cost, just slower progress
If Revenue Lower Than Projected
- Break-even: 25 Supporter users at $10/mo = $3,000/yr
- Current costs: $3,059/yr
- Break-even is achievable with minimal adoption
If BTC Price Drops
- Protocol revenue drops proportionally
- Consumer SKU revenue (fiat) unaffected
- Storage fees (fiat) unaffected
- Still break-even at 25 users
Approval Request
I recommend immediate authorization of $1,669 for:
- Mac Mini M4 Pro (24GB, 512GB) — $1,399
- UPS battery backup — $150
- External SSD 2TB — $120
This unlocks:
- Local LLM inference (Querant, Sage, Forger)
- Self-hosted Nostr relay
- Web gateway
- Build pipeline
- ZeroClaw agent runtime
Expected payback: 11 months Expected 3-year savings: $2,441
Write a comment