tty5

I spend significantly more time keeping AI at bay than using it.

• It is not good enough to write code at the expertise level I'm usually required to work at. I've spent more time fixing generated code than it would have taken me to write it from scratch.
• It hallucinates too much for me to trust any information provided by one
• Security of AI companies is about the same as IoT companies with the difference being that if IoT leaks my data it's going to be incompetence and not malice - I don't trust AI with any of my local data.
• AI agents require to be given even more access and permissions and that's just not happening.
• I contact support when I've exhausted what I can do myself and as a result AI chatbots are an annoying obstacle that can't help me and I have to waste my time going through to reach a person that actually has power to help me

Slightly different part, same process. The bigger bottleneck is packaging - HBM is 3d stacked.

Same memory production capacity can be allocated to ddr5 or to hbm and openai signed contracts with sk hynix and samsung, the two largest ram manufacturers in the world, and bought a significant percentage of next year's production.

DDR5 prices started spiking as that deals impact propagated through the supply chain. I bought a 2x32 6800 Cl30 kit for 195 euro 12 days ago. It was 330 euro 4 days later.

CATL wholesale pricing per kWh is already almost 50% below lifepo with a goal of sub $20/kWh pricing in coming years.

More durable, cheaper, can be operated at a wider temperature range and much safer, but at a cost of lower energy density.

They look like a big step forward for uses where density matters little, like grid energy storage or small scale home backups.