da-ri 리트윗함
Be very skeptical of any claims such as "Bitcoin mining is unprofitable beyond this threshold" or "AI is killing Bitcoin".
Not only is it more nuanced than that, but the research tells us that AI datacenters increasingly need Bitcoin mining (see 7. below)
For example
1. In places such as Europe, where electricity prices are high, mining still exists because often miners view heat recycling as the primary revenue, SATS as the byproduct
2. Bitcoin mining companies increasingly buy generation assets, allowing them to mine for zero marginal cost
3. Bitcoin mining companies can use older mining rigs on intermittent power sources, getting a very cheap rate.
4. Stranded energy such as O&G and landfills provides the chance to mine at ~1c/kWh in return for higher upfront CAPEX
5. Demand response, FCAS, RECS, carbon credits all provide ancillary revenue sources that aren't factored into general "mining is unprofitable below level-x" pronouncements
6. Prices can and often do go negative where there are renewable electricity surpluses, despite the general wholesale price of electricity (Again, a suitable source for older mining rigs)
7. Mining can be used alongside AI for strategic advantages including
- monetizing energy during AI datacenter construction
- using forward-purchased energy that would otherwise be wasted
- smoothing demand patterns of AI load and recategorizing AI datacenters as FLC (flexible load consumers) source: web.archive.org/web/2024033021…. As FLC, AI datacenters become far more likely to win contracts with grid operators, and help grid operators delay infrastructure investment costs that AI would otherwise necessitate (source: nicholasinstitute.duke.edu/sites/default/…)
When assessing claims about Bitcoin mining profitability, two good questions to ask are
"does this person have domain knowledge at the intersection of AI, Bitcoin mining and energy grids?"
and
"Does the claim sound hyperbolic or absolutist?"
English
