GM... You can waste your time, or you can forge it into something that outlives you. That’s the only choice. image
Quiet men with discipline move mountains. Loud men with none get buried under them. image
Most men fear failure. I hunt it. Each scar is proof I built something worth bleeding for. image
Now I do Asics: Bottom line: M50S β‰ˆ 26J/TH vs S19j Pro β‰ˆ 29.5 J/TH. MicroBT wins on efficiency and field reliability. Bitmain wins on parts availability and swap speed. On 240 V, both are happy; Whatsminer PSUs are 220–240 V only and use one C19 lead; S19j Pro uses APW12 (200–240 V) and needs two power cords. Core specs MicroBT Whatsminer M50S: ~126–130 TH/s, ~3.3–3.4 kW, β‰ˆ26 J/TH, PSU P221B/P222B AC 220–240 V, single C19. Bitmain Antminer S19j Pro: 96–100 TH/s, ~2.8–3.1 kW, 29.5 J/TH, APW12 PSU AC 200–240 V, two power cords required. 240 V electrical differences Whatsminer: Integrated top-mounted PSU (P221/P222). Input 220–240 V only. One C19 β‰₯16 A per unit simplifies cord sets and PDU design. Antminer: APW12 modular PSU. Input 200–240 V. Two cords to the PSU; more outlets per rack, but easier PSU swap. Data-center ops: pros and cons Whatsminer M50S Pros: Better efficiency; strong stability reports at scale; single-lead power; fewer DOA/failures vs recent Bitmain lines per operator reports. Cons: PSU and chassis integration means heavier units and fewer third-party spares; vendor-specific parts (P221/P222) and firmware; slower init than some Antminers in tests. Antminer S19j Pro Pros: Huge aftermarket for hashboards, fans, APW12; fast boot; lots of repair docs and vendors; easy PSU swaps. Cons: Worse efficiency; community reports higher DOA/failure rates on several Bitmain 19/21-series batches; more cords and outlets to manage. Failure rate and labor cost model (use your scale) Field chatter shows Whatsminer ~2% annual failures vs Antminer ~8% depending on batch; S19j Pro is older but still sees board/PSU swaps. Treat these as assumptions, not guarantees. Example for a 1,350-unit farm, $150/hr tech time, 2.0 h per incident (R&R, test, ticket): Whatsminer @ 2% β‡’ 27 incidents/yr β‡’ $8,100 labor/yr. Antminer @ 4% β‡’ 54 incidents/yr β‡’ $16,200 labor/yr. Delta β‰ˆ $8,100/yr more labor for Antminer. Scale linearly with your unit count and your measured failure rates. Quick take If you want fewer headaches: M50S. If you want fastest swaps and cheapest parts cabinet: S19j Pro. On power distribution: both standardize cleanly at ~240 V PDUs; Whatsminer’s single-cord layout simplifies outlet density. My recommendation is MicroBT all day View quoted note β†’
Someone asked me what it takes to put on a 5MW infra plan so i wrote it up with cost avg.... here ya go world. 1,428 MicroBT ASICs at 3.5 kW each on a raw 5 MW bus. If you budget a realistic PUE 1.05 - 1.10, that’s 1,360–1,300 miners powered concurrently. Maths: Raw count: 5,000 kW / 3.5 kW β‰ˆ 1,428 units. Container plan Spec’d 4 Γ— 1.5 MW containers = 6 MW of rack capacity. Your generation is 5 MW. Oversized here for 80% infra to mitigate heat issues durring summer months. Run 1.5 MW container at ~1.25 MW per container (even split). a 1.25 MW container 357 asics. Electrical one-liner (clean and cheap) shouldn't be more that few hundred dollars to electrical engineers office. Best practice to kill copper costs and voltage drop: 1. Paralleled gens: 2 Γ— 2.5 MW turbines at 480 V into a 480 V generator switchboard, each on a ~3,200 A breaker (2.5 MW/(√3Β·480Β·0.95) β‰ˆ 3,165 A). 2. Step-up to MV: 2 Γ— 2.5 MVA 0.48 kV to 4.16 kV transformers (or 13.8 kV). Tie to an MV bus. 3. Distribute MV to pads near each container. 4. Step-down at the edge: 4 Γ— 1.25–1.5 MVA MV to 415Y/240 V pad-mounts, one per container. 5. LV switchboard at each container feeds PDUs. Why I recommend this: pulling 7,300 A at 415 V across site is a cable, cost crime. MV distribution fixes that. Main 480 V board to 2 Γ— 2.5 MVA 480 to 415Y/240 V transformers in parallel β†’ common 415 V board. From 415 V board to each container: 1.5 MW feeder current β‰ˆ 2,200 A (1.5 MW/(√3Β·415Β·0.95)). Expect 5–6 runs/phase of 500 kcmil Cu or 7–8 runs/phase of 750 kcmil Al per container to stay within ampacity and voltage-drop limits (exact run count depends on length, ambient, grouping, and NEC derates). Transformer sizing and notes Edge pad-mounts: 1.25–1.5 MVA, MV primary (4.16 or 13.8 kV), 415Y/240 V secondary, Z ~ 5–6%, ONAN/NEMA 3R, taps Β±2Γ—2.5%. Central step-up: 0.48 feed to MV at 2.5 MVA each, or buy turbines with MV alternators and skip the step up. If staying LV-LV: dry-type 2.5 MVA 480Ξ” to 415Y/240 exists off the shelf. Switchgear and breakers (typical) 480 V generator breakers: 3,200–3,500 A each. 415 V main bus rating: β‰₯8,000 A if centralized. Container feeders: 2,000–2,500 A frame per 1.25–1.5 MW container, adjustable trips set by cable study. Conductor quick math (centralized LV case) Site-total at 415 V: ~7,300 A (5 MW/(√3Β·415Β·0.95)). Per container: ~2,200 A at 1.5 MW; 1,830 A at 1.25 MW. Feeder count guide (THHN/XHHW-2, 75–90 Β°C, in multiple conduits): plan 5–6Γ— 500 kcmil Cu/phase for 1.5 MW over modest distances. Check voltage drop and derates before final. Rough capex ranges (equipment only, excludes site/civils/MEP/labor) Gas turbines 2.5 MW class: equipment-only $700–$1,200/kW at this scale. Two units avg $3.5–$6.0 M range pending make/model and package. Capstone path is modular (C1000S = 1 MW blocks), often priced used/new per unit; multi-MW achieved in parallel. Pad-mount transformers 2.5 MVA: $90k–$150k each depending on voltage and options. 1.25–1.5 MW mining containers (shells): wide spread. Air-cooled 20–40 ft units typically $50k–$150k; high-density or hydro options can push higher. Cabling/switchgear: highly distance-dependent. Centralized LV design balloons copper. MV distribution plus edge step-downs usually cuts conductor cost by 40–60% vs long 415 V pulls. (Engineering judgment; validate with your actual runs.) Deliverables you can lift into RFQs Generation: 2 Γ— 2.5 MW natural-gas turbines, 480 V alternators, continuous duty, black-start package, utility-grade sync/protection. Transformers: Option A(my recommendation): 2 Γ— 2.5 MVA 480β†’4.16 kV step-up; 4 Γ— 1.25–1.5 MVA 4.16 kVβ†’415Y/240 V pad-mounts at container pads. Option B (LV-LV): 2 Γ— 2.5 MVA 480β†’415Y/240 V near gens; 4 Γ— 2,000–2,500 A 415 V feeders to containers. Switchgear: 480 V gen board with two 3,200 A breakers and tie; MV switchgear with four feeders; or 415 V board with four 2,000–2,500 A feeders. Conductors: Per container feeder sized for 1.25–1.5 MW at 415 V. Plan 5–6 Γ— 500 kcmil Cu/phase (or 7–8 Γ— 750 kcmil Al/phase) for short-to-moderate runs; finalize by NEC 310 ampacity, 250.122 EGC, and voltage-drop calcs. COSTS ASSOCIATED NEW VS USED: clean parts-only build sheet for 5 MW at 3.5 kW/ASIC. Count = 1,428 ASICs $285,600 at $200 each. Everything else is line-itemed New vs Refurb with assumptions stated. Assumptions Power topology: 2Γ— 2.5 MW gens, central 480 to 415/240 step-down (2Γ— 2.5 MVA), then 415 V feeders to 4Γ— containers. Feeder length placeholder = 200 ft per container. Change L and multiply. Feeder sizing (415 V) per 1.25–1.5 MW container: 5 parallel runs/phase of 500 kcmil Cu (or 750 kcmil Al) with neutral and EGC in each conduit. That’s 25 conductors/container. Electrical labor $150/hr. Crane allowance shown. Transport $50k total. Gas hookups $25k. Unit prices (to calc conductors and options) 500 kcmil Cu THHN: ~$12.5–$17.1/ft. Use $15/ft midpoint. 750 kcmil Al XHHW-2: ~$6.8–$7.7/ft. Use $7.25/ft. 15 kV MV-105, 350 kcmil (for future MV option): ~$32/ft. 2.5 MVA LV-LV 480Ξ”β†’415Y/240 (new, NEMA 3R/dry or pad-mount oil): budget $220k–$260k each; use $240k. Example new 2.5 MVA pad-mount pricing shown. Refurb comps in the $20k–$100k band depending on type. 2.5 MW turbines/gens (continuous duty): New $1.3M–$1.8M each (use $1.6M mid); Refurb/used $0.6M–$1.0M (use $0.8M). Comps shown (CAT 2.5 MW new, MTU 2.5 MW new, Jenbacher 2–2.5 MW used). Mining containers 1.5 MW: Used $50k, New $150k. Crane: allow $25k–$35k total for 2–3 days of 100–175 ton with mobilization. Conductor maths (per container @ 415 V, 200 ft) Copper option: 25 conductors Γ— 200 ft = 5,000 conductor-ft Γ— $15/ft = $75,000 per container is $300,000 for 4. Aluminum option: 5,000 conductor-ft Γ— $7.25/ft = $36,250 per container is $145,000 for 4. (Formula to adjust: Total $ = containers Γ— 25 Γ— length(ft) Γ— unit $/ft.) Line-item quote (NEW build) 2Γ— 2.5 MW gas turbines/gens, new @ $1,600,000 … $3,200,000. 2Γ— 2.5 MVA 480 step down 415/240 transformers, new @ $240,000 … $480,000. 4Γ— 1.5 MW mining containers, new @ $150,000 … $600,000. 415 V feeders (Cu, as specced) = $300,000 at 200 ft/container. Main/switchgear & distribution (480 V + 415 V boards, breakers, protection): $250,000 (allowance, new). ASICs 1,428 @ $200 … $285,600. Electrical labor allowance 1,200 hrs Γ— $150/hr … $180,000. Crane allowance … $30,000. Transportation … $50,000. Gas hookups … $25,000. NEW total β‰ˆ $5,401,000. Line-item quote (REFURB/USED build) 2Γ— 2.5 MW turbines/gens, refurb @ $800,000 … $1,600,000. 2Γ— 2.5 MVA 480β†’415/240 transformers, refurb @ $80,000 … $160,000. 4Γ— 1.5 MW mining containers, used @ $50,000 … $200,000. 415 V feeders (Al option) β‰ˆ $145,000 at 200 ft/container. Switchgear refurb allowance … $120,000. ASICs 1,428 @ $200 … $285,600. Electrical labor $180,000. Crane $30,000. Transportation $50,000. Gas hookups $25,000. REFURB/USED total β‰ˆ $2,796,000. Notes that save you money If you flip to MV distribution (480β†’4.16/13.8 kV near gens, pad-mounts at each container), your feeder copper drops massively. Example: 15 kV 350 kcmil at $19.2k/container for 200 ft**, vs $75k copper at LV. Net can remain lower even after adding four pad-mounts.