人工智能数据中心推动比特币挖矿经济的微妙方式

对廉价电子的竞争可能会为算力价格设定一个底线，或者至少减缓算力增长。

The burgeoning presence of artificial intelligence (AI) data centers might be good news for bitcoin miners — even those who have nothing to do with AI.

人工智能（AI）数据中心的蓬勃发展对于比特币矿工来说可能是个好消息——即使是那些与人工智能无关的矿工。

Here’s why: The competition between AI data centers and bitcoin miners for cheap electricity could wind up putting a floor on hashprice, a key metric used by miners to gauge their revenue.

原因如下：人工智能数据中心和比特币矿工之间对廉价电力的竞争可能最终会给哈希价格设定一个底线，哈希价格是矿工用来衡量收入的关键指标。

“Every potential mining investment now goes through this filter: is it better to use this site for AI purposes or mining,” Spencer Marr, president of bitcoin mining firm Sangha Renewables, told CoinDesk. “Every time they choose AI or other forms of high-performance computing, that means hashrate won’t climb, and hashprice won’t be negatively impacted.”

比特币矿业公司 Sangha Renewables 总裁斯宾塞·马尔 (Spencer Marr) 告诉 CoinDesk：“现在每一项潜在的矿业投资都会经过这个过滤器：将这个网站用于人工智能目的还是采矿更好。” “每次他们选择人工智能或其他形式的高性能计算，这意味着哈希率不会攀升，哈希价格也不会受到负面影响。”

Hashrate is a measure of the total combined computational power backing a Proof-of-Work blockchain, in this case Bitcoin. Hashprice, meanwhile, is the amount of bitcoin that a miner can expect to earn each time their machines perform a certain number of hashes, or computations, on a given timeframe.

算力是支持工作量证明区块链（在本例中为比特币）的总计算能力的衡量标准。与此同时，哈希价格是指矿工在给定时间范围内每次机器执行一定数量的哈希或计算时可以预期赚取的比特币数量。

At press time, bitcoin’s hashrate totalled 770 exahash per second (EH/s), per Hashrate Index data, with bitcoin’s hashprice coming in at $61.12 per petahash per day. Hashprice has been steadily decreasing as mining has grown more competitive. Back in 2017, for example, it wasn’t uncommon for hashprice to clock in above $1,000 by that measure.

根据哈希率指数数据，截至发稿时，比特币的哈希率总计为每秒 770 exahash (EH/s)，比特币的哈希价格为每天每 petahash 61.12 美元。随着挖矿竞争变得更加激烈，哈希价格一直在稳步下降。例如，早在 2017 年，按照这一标准计算，哈希价格突破 1,000 美元的情况并不罕见。

Establishing a floor on hashprice would be valuable for miners as it would guarantee that computing power will never go below a certain threshold in terms of value, no matter the conditions.

建立哈希价格下限对于矿工来说很有价值，因为它可以保证无论条件如何，计算能力的价值永远不会低于某个阈值。

“In the competition for access to cheap electricity, miners are starting to be squeezed out by more willing purchasers in the form of other forms of compute,” Marr said. “It’s a specific game theory, because as the miner, you like to see other people deciding to use cheap electrons for other purposes than Bitcoin compute because of the competitive nature of mining.”

马尔表示：“在争夺廉价电力的竞争中，矿工开始被更愿意以其他形式的计算形式购买的人挤出。” “这是一个特定的博弈论，因为作为矿工，你希望看到其他人决定将廉价电子用于比特币计算之外的其他目的，因为采矿的竞争性质。”

But the squeeze may simply lead bitcoin miners to up and move to other jurisdictions across the globe where AI data centers aren’t sprouting left and right, Jaran Mellerud, co-founder of bitcoin mining hardware and hosting services firm Hashlabs Mining, told CoinDesk.

但比特币挖矿硬件和托管服务公司 Hashlabs Mining 的联合创始人贾兰·梅勒鲁德 (Jaran Mellerud) 告诉 CoinDesk，这种挤压可能只会导致比特币矿工迁往全球其他人工智能数据中心尚未兴起的司法管辖区。

“I don’t think the competition for power from AI facilities will significantly impact hashprice,” Mellerud said. “The Bitcoin mining network is a self-correcting mechanism, so reduced hashrate in one country will simply increase profitability of miners in another country, giving them more room to grow.”

“我认为人工智能设施的电力竞争不会对哈希价格产生重大影响，”梅勒鲁德说。 “比特币挖矿网络是一种自我纠正机制，因此一个国家的算力降低只会增加另一个国家矿工的盈利能力，从而为他们提供更多的增长空间。”

“My thesis is that the U.S. will have less than 20% of the hashrate by 2030 due to competition from AI facilities, while hashrate will grow elsewhere, particularly in Africa and Southeast Asia,” Mellerud added.

Mellerud 补充道：“我的观点是，由于人工智能设施的竞争，到 2030 年，美国的算力将占不到 20%，而其他地方的算力将会增长，特别是在非洲和东南亚。”

Marr agreed these were valid points, though he pointed out that at the end of the day, “there is a finite number of rock bottom cheap electrons.” AI data centers are also trickier to operate than bitcoin mines; they require constant uptime, for example, and are much more expensive to build and run.

马尔同意这些观点是正确的，但他指出，归根结底，“最低价的廉价电子的数量是有限的。”人工智能数据中心的运营也比比特币矿场更棘手；例如，它们需要持续的正常运行时间，并且构建和运行的成本要高得多。

“Maybe in the end the competition for electrons slows down but doesn’t stop hashrate growth,” Marr said.

“也许最终电子竞争会放缓，但不会阻止算力增长，”马尔说。