šØ A Bitcoin mining company just signed a 910-megawatt, 15-year data center lease worth an estimated $28 billion, while Vitalik spoke up this week in support of the āanti-data centerā crowd.
Group: éčÆ·ęØå å „ęēē¾¤ē» ęē©ŗē²äøē¾¤
On November 8, Ethereum co-founder Vitalik Buterin publicly pushed back against the heated criticism directed at the anti-data center movement over the past week. He didnāt call for existing facilities to be shut down or propose a government ban. He simply made a measured point: slowing construction down might not be a bad thing.
š„ The catalyst was Riot Platforms. On November 4, it signed a 15-year lease at its Corsicana, Texas, campus, covering 910 megawatts of critical power capacity, with estimated revenue of about $28 billion over the initial contract term. Cipher Digital is also converting mining sites into AI hosting facilities. The power allocations, land, and substations in minersā hands have suddenly become some of the most valuable tickets to entry in the AI era.
šø But Vitalikās argument isnāt that āAI is dangerousā; itās that āalgorithms will get cheaper.ā His reasoning is that tasks requiring bigger and more data centers today could be done with even larger centers in two years, and possibly with existing facilities in four yearsāor even sooner.
ā So his conclusion is that slowing the breakneck expansion of computing capacity would have a positive net effect. What really needs time isnāt model training, but giving society stronger defenses against cybersecurity and biological risks. Research progress also doesnāt depend on computing power alone; AI being a year slower doesnāt mean science will be a year slower.
š These words are especially jarring right now. The business case for miners shifting to AI hosting makes sense, while community opposition is focused on electricity costs, water use, and the pace of construction. The two sides are really talking about different things: one is calculating cash flow, the other is asking who bears the external costs.
ā ļø Thatās where the risks lie, too. If algorithmic efficiency improves more slowly than expected, slowing construction could mean ceding computing sovereignty to others. If it improves faster than expected, the 15-year contracts signed today could leave behind a pile of idle transformers.
Whatās really worth watching isnāt whether AI data centers should be built, but who gets to decide what āfastā meansāthe mining companies signing the contracts, or the residents living beside the substations.
When one of the industryās most aggressive expansionists starts speaking up for its opponents, whose calculations are you more inclined to trust?
Tap the profile picture to watch the livestream + join my fan group for daily strategies š
#ęÆē¹åø #AIē®å #仄å¤Ŗå #InstitutionalCapital
Group: éčÆ·ęØå å „ęēē¾¤ē» ęē©ŗē²äøē¾¤
On November 8, Ethereum co-founder Vitalik Buterin publicly pushed back against the heated criticism directed at the anti-data center movement over the past week. He didnāt call for existing facilities to be shut down or propose a government ban. He simply made a measured point: slowing construction down might not be a bad thing.
š„ The catalyst was Riot Platforms. On November 4, it signed a 15-year lease at its Corsicana, Texas, campus, covering 910 megawatts of critical power capacity, with estimated revenue of about $28 billion over the initial contract term. Cipher Digital is also converting mining sites into AI hosting facilities. The power allocations, land, and substations in minersā hands have suddenly become some of the most valuable tickets to entry in the AI era.
šø But Vitalikās argument isnāt that āAI is dangerousā; itās that āalgorithms will get cheaper.ā His reasoning is that tasks requiring bigger and more data centers today could be done with even larger centers in two years, and possibly with existing facilities in four yearsāor even sooner.
ā So his conclusion is that slowing the breakneck expansion of computing capacity would have a positive net effect. What really needs time isnāt model training, but giving society stronger defenses against cybersecurity and biological risks. Research progress also doesnāt depend on computing power alone; AI being a year slower doesnāt mean science will be a year slower.
š These words are especially jarring right now. The business case for miners shifting to AI hosting makes sense, while community opposition is focused on electricity costs, water use, and the pace of construction. The two sides are really talking about different things: one is calculating cash flow, the other is asking who bears the external costs.
ā ļø Thatās where the risks lie, too. If algorithmic efficiency improves more slowly than expected, slowing construction could mean ceding computing sovereignty to others. If it improves faster than expected, the 15-year contracts signed today could leave behind a pile of idle transformers.
Whatās really worth watching isnāt whether AI data centers should be built, but who gets to decide what āfastā meansāthe mining companies signing the contracts, or the residents living beside the substations.
When one of the industryās most aggressive expansionists starts speaking up for its opponents, whose calculations are you more inclined to trust?
Tap the profile picture to watch the livestream + join my fan group for daily strategies š
#ęÆē¹åø #AIē®å #仄å¤Ŗå #InstitutionalCapital
