S21 XP vs S23 upgrade math worth the infrastructure costs?

[philipma1957]

hi folks, i'm running 12x s21 xps pulling 13.5w/th at my facility. bitmain pushing these s23 hydros at 9.5w/th but need complete rewire to 410v three phase plus liquid cooling setup.

did the math:
current setup: 12 units x 277th = 3324th total
upgrade option: 6 units x 580th = 3480th total

gain: 156th more hash
cost: 6x$9000 = $54k plus $25k infrastructure
payback at 6 cents/th: 79000/9.36 = 8440 days

am i missing something or is this upgrade financially insane?

 

[junmisakiro]

similar calculation here but different situation. have old s19 pro units from 2021 still running 29.5w/th average. for me upgrade make more sense because current efficiency very bad. but philip already have good efficiency units. maybe wait for next generation or focus on electricity cost reduction instead hardware upgrade?

 

[CryptoDada]

The infrastructure investment is the critical factor most miners underestimate. Beyond the $25k upfront cooling costs, you're looking at:

• Ongoing coolant maintenance and replacement cycles
• Increased complexity requiring specialized technical knowledge
• Higher failure points with pumps, radiators, and liquid connections
• Potential catastrophic damage from cooling system failures

For established operations with recent hardware, the ROI simply doesn't justify the transition. The efficiency gains are marginal compared to the operational complexity increase.

 

[The Cryptovator]

working with enterprise clients considering similar upgrades. institutional perspective is stay with proven air cooling infrastructure until liquid efficiency reaches sub-8w/th threshold.
current s21 xp fleet performing well for most operations. better to optimize existing setup than chase incremental gains with unproven liquid systems

 

[cryptonit]

another bitmain cycle pushing miners to upgrade perfectly functional gear. s21 xps still profitable why fix what aint broke
seen too many operations get burned chasing latest efficiency only to deal with cooling failures and downtime

 

[btcinfo891]

ran numbers on 50 unit upgrade scenario recently. structure costs were brutal even with bulk pricing.
cooling system: $87k
electrical upgrade: $34k
installation/commissioning: $15k
total: $136k just for infrastructure
hardware already expensive enough without 100k+ facility modifications

 

[hack3rcon]

analyzing power consumption data from existing hydro deployments. found 15-20% higher actual power draw than spec sheets due to cooling pumps and radiator fans. "9.5w/th" becomes closer to 11w/th when accounting for complete system power draw. still better than air cooled but margin shrinks significantly

 

[root777]

bc infrastructure complexity is thebiggest concern. air cooled units fail you swap them out. liquid cooled fails you potentially flood entire rack and damage multiple units. seen warehouse lose 24 units from single pump failure that wasnt caught quickly enough

 

[goldkingcoiner]

how smart spending 80 grand to mine slightly more efficiently while difficulty keeps rising anyway. by time you break even next gen will be out making your "upgrade" obsolete again. so its just another way for bitmain to extract money from miners who cant do basic math

 

[UnderLow404]

running small home operation 4x s21 xp units. been looking at upgrade but costs make no sense for smaller miners. need 3 phase power upgrade $8k minimum plus cooling setup another 12k. for 4 units thats insane overhead

 

[maxd]

institutional mining operations have different calculations than home miners. when youre running 1000+ units the infrastructure amortization makes more sense. but for smaller operations the fixed costs create impossible payback periods especially with rising difficulty

 

[Radius]

seeing trend toward ai/ml workload hybrid operations. pure bitcoin mining margins getting squeezed from multiple directions. maybe focus on diversifying compute instead of chasing efficiency gains that get eaten by difficulty increases

 

[sk8er]

so if s21 xps still profitable why everyone stressing about upgrades? my dad runs 6 units making like 300 bucks month after power. seems like lot of work and money to make same profit with fancier gear

 

[philipma1957]

so if s21 xps still profitable why everyone stressing about upgrades? my dad runs 6 units making like 300 bucks month after power. seems like lot of work and money to make same profit with fancier gear

exactly! running profitable operation why risk it chasing marginal gains.
infrastructure upgrade would take 23 years to break even at current btc prices. by then gear will be obsolete anyway.
sticking with air cooled s21 xps until something revolutionary comes along not incremental improvements

 

[Major]

agreed manufacturers create fomo around efficiency gains while real world payback periods get longer. s21 xps solid reliable units. upgrade when they stop being profitable not when something slightly better exists

 

[Trump_Coiner]

maybe just get few s23 units test them out instead going all in? infrastructure costs probably scale better with more units anyway

 

[CryptoDada]

@Trump_Coiner Partial deploymnet creates operational complexity without cost benefits. You end up managing two different cooling systems, power requirements, and maintenance schedules. Beter to commit fully to one approach or maintain current infrastructure until clear ROI justification exists.

 

[bittersweet]

remember when antminer s9s were cutting edge efficiency? then s17s then s19s now s21s. each generation promised amazing payback periods. reality is manufacturers benefit most from upgrade cycles not miners

 

[junmisakiro]

remember when antminer s9s were cutting edge efficiency? then s17s then s19s now s21s. each generation promised amazing payback periods. reality is manufacturers benefit most from upgrade cycles not miners

true but tech progression still important. older gear become unprofitable event so need replacement strategy. questiom is timing and scale of upgrade not whether upgrade ever needed

 

[philipma1957]

ok so here is my final calculation for anyone interested:

current profit: 3324th x 0.06 = $199/day
upgraded profit: 3480th x 0.06 = $208/day
daily gain: $9
investment: $79000
payback: 8777 days = 24 years

unless btc price increases dramatically or electricity costs drop significantly this upgrade makes 0 financial sense