A 12-unit Kaspa farm in Texas: the deployment and the cliff decision
This is a composite, drawn from two real Kaspa-focused customers who deployed with us in late 2025 and early 2026. The numbers and the decision are accurate to what we actually saw; identifying details are removed.
The starting position
The operator had access to a light-industrial space in Texas with a power rate around $0.038/kWh, cheap by US standards, and the kind of rate that makes kHeavyHash hardware genuinely viable rather than a hope-the-price-rises bet. The plan was a focused Kaspa fleet rather than a mixed deployment.
The first order was 4x IceRiver KS5L. At roughly 12 TH/s and a 3,400W draw each, the KS5L was the cleanest efficiency tier available at the time for a sub-$0.04 power rate. Starting on one model kept firmware, monitoring, and pool setup simple while the operator learned the facility.
The first 60 days
The same patterns we see on almost every first deployment showed up here:
- Realised hashrate landed about 3% under nameplate after firmware variance and accepted-share latency. Planned-for, not panicked-about.
- Metered power draw came in around 4% over nameplate. The operator had sized circuits at 75% utilisation, so there was margin to absorb it.
- One unit hit a fan controller fault in week two. Replaced under the 60-day warranty, back online inside a week.
After two months of clean data, the operator came back and scaled to 12x KS5L total, plus a single IceRiver KS5M at 15 TH/s to benchmark the higher-hashrate unit against the KS5L fleet in the same facility.
Then the cliff math arrived
We have written separately about the July 2026 Kaspa emission cliff, the protocol-scheduled drop that cuts daily KAS issuance roughly 90% from where it sat in spring 2026. For this operator, the cliff was the single biggest planning variable in the whole deployment.
The honest pre-cliff math at $0.038/kWh: each KS5L drawing about 3.4 kW costs roughly $3.10/day in power. Pre-cliff daily revenue per unit ran comfortably above that, so the fleet was solidly profitable. Post-cliff, holding KAS price and difficulty constant, daily revenue per unit drops toward the power-cost line and the margin gets thin.
The decision the operator made
Three options were on the table: sell the fleet before the cliff while resale was strong, hold and bet on KAS price appreciation, or diversify into a more emission-stable algorithm.
The operator chose to hold the KS5L fleet and add Scrypt hardware alongside it. The reasoning: at $0.038/kWh, even the thin post-cliff KAS margin stays marginally positive under conservative price assumptions, so selling efficient hardware into a soft post-cliff secondary market did not make sense. But concentrating an entire operation on one algorithm facing a 90% emission cut was too much single-point risk. Adding a few Scrypt units (Litecoin and Dogecoin merge-mined, emissions stable through the next LTC halving) hedged the algorithm risk without abandoning a fleet that still worked.
Why this is the useful lesson
The takeaway is not "Kaspa good" or "Kaspa bad". It is that the right call is entirely a function of your power rate. At $0.038/kWh the operator could hold through the cliff. At $0.08/kWh the same fleet would have been a clear sell-before-the-cliff decision. Power rate is the variable that decides almost everything in mining, and any vendor who tells you a unit is profitable without first asking your power rate is selling, not advising.
If you are modelling a Kaspa deployment
If you are weighing kHeavyHash hardware and the cliff is in your planning window, message us with your power rate and target fleet size. We will send a written ROI projection that models the July transition for your specific numbers, no charge, and we will tell you plainly if your power rate does not support the buy.
Talk to us about hardware selection, ROI projections at your power rate, or a written quote for bulk pricing.