Discover more from Principles of Bitcoin
Bitcoin and Energy @ Texas A&M
Voltage Interruptions on the Texas grid
This last year, I helped launch the Texas A&M Blockchain and Energy Research Consortium. This is an interdisciplinary group of faculty from business and engineering that seeks to understand the intersection of energy and Bitcoin. Our first workshop was last Wednesday, where we hosted 50 people from across Texas to showcase our research group. Here are the highlights:
We first presented a tutorial on voltage ride-through issues in the Texas grid. Occasionally there are irregularities in the supply of power across Texas. For example, a tree may fall on a power line, or a wind turbine on a wind farm may malfunction. The question is how to respond to these interruptions in service.
This matters for Bitcoin mining because an industrial miner is a Large Flexible Load (LFL). They are large because industrial Bitcoin mining in Texas is no small potatoes: publicly traded companies like Riot have up to 1 gigawatt facilities across Texas. What's unique about these loads is that they are flexible: they can turn on or off on a dime. Historically this has been a huge benefit for the Texas grid. When demand spikes, prices start to climb, and Bitcoin miners will rationally stop mining when their costs increase. This is a “demand response,” a rational change in behavior to a market price.
The flip side is that LFLs, like Bitcoin miners, can amplify these voltage interruptions across the grid. Because their load is large, a 600-megawatt interruption in power can have a ripple effect. If the interruption is prolonged, the miners can elect to turn off their machines temporarily, which also destabilizes the grid because it can quickly shift a large load offline.
One of the Ph.D. students in our group ran several experiments on our ASIC S19 that the Texas Blockchain Council donated to our lab last year. He found that an individual ASIC can “ride through” a voltage interruption on its own. But the question is still open for how hundreds of ASICs tied together can. There may be a hardware innovation here that could assist the Bitcoin miners to be able to ride through larger and longer voltage interruptions.
Miners vs. Data Centers
Other large loads on the grid, like data centers and hospitals, have solved this by installing backup generators precisely for these voltage interruptions. This makes sense for them because their end users (internet consumers and patients) demand uninterrupted service. But miners serve not Bitcoin users directly, but rather the Bitcoin network, which operates on a global scale. Even though 17% of the worldwide Bitcoin hash rate comes from Texas, an interruption in voltage for industrial miners in Texas will not meaningfully change the security of the blockchain.
To be even more explicit, miners receive the majority of their revenue today from the block subsidy issued by the protocol, not directly from the users of Bitcoin. The protocol will allocate its block subsidy to whomever on the global Bitcoin network mines the next block. If a Texas miner loses power for a moment, it temporarily stops participating in the global Bitcoin lottery, or temporarily stops contributing hash power to its mining pool. Either way, the miner can jump back online and resume hashing with little penalty from the market.
You might think that the penalty from a voltage interruption would increase when the miner receives more of its revenue from transaction fees, which will happen over time as the block subsidy wanes. Even still, the protocol awards the block reward only to the miner (or mining pool) with a successful block. This “pay for performance” prize is a feature, not a bug. It ensures the sender of the bitcoins pays the transaction fees only once the miner appends the block to the ledger. The mining lottery ultimately insulates the user of the Bitcoin network from the details and vagaries of mining. Voltage interruption, put plainly, is the miner’s problem, not the user’s.
This is the key difference between a miner and a data center or hospital. A power interruption could wreck a surgery or a complex AI computation on an AWS server. So the downside costs are much greater. But because miners do not face such a downside, they have no incentives on their own to pay for expensive backup generators.
The concern of ERCOT, the Texas grid operator, is that these flexible loads can amplify the ripple effects of voltage interruptions, imposing negative externalities on the rest of the grid. As such, ERCOT is considering mandating the miners to install backup generators. This would be a mistake, as it would impose an onerous regulation that would penalize the miner for acting in its own rational self-interest, while not clearly identifying the cost of the negative externality.
The standard economic solution would be to measure that externality precisely and implement it through a price mechanism, so that the miners can internalize any externality they are imposing, if any. This would at least employ prices to induce behavior rather than imposing mandates that impose certain costs. Whatever policy ERCOT selects will tilt the economics slightly in favor of off-grid versus on-grid mining.
Stay tuned as this new policy agenda unfolds.