It is part of the grift. Anthropic and OpenAI want to go public and a huge part of their cash burn is compute expense for training new models. (In public filings they claim they are profitable … if you look at their numbers that’s only if you remove their biggest expense which is model training) At the same time they are seeing fewer gains with new frontier models. Some are also claiming the new models aren’t as good as the ones from a year ago and you are loosing your market advantage with frontier models?
What do you do if your largest expense is generating fewer gains and you want to go public? Get the government to regulate it and build a regulatory mote around your industry and freeze your advantage over your competitors while you have it. Cut that expense and go public as a profitable company!
At this point I only want to see regulations well after the AI bubble pops. Don’t take any government stakes in these AI companies as Bernie has argued for (I agree with his stances on most economic matters but he’s off base with this). Let these AI companies burn in the free market they advocate for soo much.
They also captured the Nuclear Regulatory Commission to allow them to run mini nuclear reactors without any of those pesky safety rules getting in the way.
Running trained LLMs has, more or less, always been profitable. Inference can only run on one GPU efficiently, you know the size of the model, you know how long it takes… etc.
Inference is approximately as expensive and environment destroying as Call of Duty, which don’t get me wrong, is a waste of humanity for sure.
It is part of the grift. Anthropic and OpenAI want to go public and a huge part of their cash burn is compute expense for training new models. (In public filings they claim they are profitable … if you look at their numbers that’s only if you remove their biggest expense which is model training) At the same time they are seeing fewer gains with new frontier models. Some are also claiming the new models aren’t as good as the ones from a year ago and you are loosing your market advantage with frontier models?
What do you do if your largest expense is generating fewer gains and you want to go public? Get the government to regulate it and build a regulatory mote around your industry and freeze your advantage over your competitors while you have it. Cut that expense and go public as a profitable company!
At this point I only want to see regulations well after the AI bubble pops. Don’t take any government stakes in these AI companies as Bernie has argued for (I agree with his stances on most economic matters but he’s off base with this). Let these AI companies burn in the free market they advocate for soo much.
Spot on. Regulatory capture is not usually this transparent.
They also captured the Nuclear Regulatory Commission to allow them to run mini nuclear reactors without any of those pesky safety rules getting in the way.
Wait, they can cover running costs now?
Running trained LLMs has, more or less, always been profitable. Inference can only run on one GPU efficiently, you know the size of the model, you know how long it takes… etc.
Inference is approximately as expensive and environment destroying as Call of Duty, which don’t get me wrong, is a waste of humanity for sure.
Model training however…
Yes, by shoveling investor cash into the generators powering their data centers.