Meta announced today that it would sell off its excess computing power. It seems the work of having programmers label data is no longer sustainable.

Unsurprisingly, Meta is the first to surrender in the computing power race.
But I don't think this incident actually has much reference value.
First, this is entirely the result of Meta digging its own grave. They have neither a model nor users, yet they thought they had deep pockets, thought they were awesome, thought they could buy talent with money, and thought users around the world couldn't live without them.
Second, the CEO makes decisions on a whim. The company has no strategy whatsoever, jumping around frantically, and upon realizing they indeed can't make it, they shift to another track and retreat completely.
On AI issues, Meta is just a troublemaker; it does not represent the views of other players.
When LLaMA came out, I already thought they were foolish. Wall Street even took Meta seriously and hyped up their AI value, which is pretty bizarre.
It seems time to adjust the allocation of pension funds and Q.
😂😂😂