this is Amodei's position in a nutshell for AI development. We have to go as fast as possible because China. It's not the only frame though. If AI models and warfare (cyber, biological) becomes easily accessible and dangerous enough, there is a strong incentive for the world's leaders to cooperate towards something akin to nuclear non-proliferation.
In fact, there's strong incentives now to slow down AI progress for multiple reasons: de-escalate tension over Taiwan and lessen China's desire to build their own advanced fabs, protect peoples livelihoods by smoothing the AI transition. Except the incentives to bring AI companies public (and maintain some twisted shred of American Hegemony) are greater.
There is no reason to slow down and only reasons to speed up. AI has not had the world shatteringly obvious negative potential as atomics by a long shot.
There is a latent potential for a negative outcome but the surface is showing a relatively benign productivity boost similar to a smart phone.
Only time will tell if the negative impact is on the scale of atomics or manmade bioweapons. Sadly, humans usually need to be burned a few times before learning the lesson. Eugenics seems to be a universal no go publicly across the world. But that was a painful lesson.