Alan Siefert on Nostr: If the AI data centers are allocating about 1 GPU per API user who pays $200/month, ...
If the AI data centers are allocating about 1 GPU per API user who pays $200/month, the H100 costs $30k and is running 24/7 (earning $2400) in a year, the company would make its investment on that GPU back in 12.5 years. What am I missing?
Published at
2025-08-15 19:30:48 UTCEvent JSON
{
"id": "7d975cfd6ee2fd0a92eabef8dfd7125aeb8e2fbc14c718718b034d251d61f87c",
"pubkey": "f26606e331f128325ec99cd38782edbec134a5f911ea9fada7adce2ca9015b16",
"created_at": 1755286248,
"kind": 1,
"tags": [],
"content": "If the AI data centers are allocating about 1 GPU per API user who pays $200/month, the H100 costs $30k and is running 24/7 (earning $2400) in a year, the company would make its investment on that GPU back in 12.5 years. What am I missing?",
"sig": "139e53084ef6b95ee15ef806cda79cce88d68c9ab56f4302dd33abebf8133a9903b37896c033f510c522fb27b5f2bc02a0dc32e01928682dbb7aab03bffccf37"
}
