[aurareturn]

I don't think companies like Nvidia, OpenAI, Microsoft, Apple, Google, Meta are in a bubble. They're making real profits or they are foundational. Likewise, I don't think S&P500 companies who are adopting LLMs in a big way are in a bubble.

I do think startups with multi-billion dollar valuations but mostly relying on OpenAI/Claude are in a bubble. There's no reason a company like Windsurf was worth $3b or Cursor should have $10b.

[goalieca]

What happens when LLM companies stop funding their services at a loss? The downstream parts and services will stop seeing revenue. Those people are making a killing off AI since they are so expensive to build and run.

[aurareturn]

They're already profitable per token in terms of inference.

They're losing money because of R&D and spending capex to buy GPUs.

[bayindirh]

While I make a lot of anti-LLM comments here, my qualm is with user facing "AI companies" using internet as their scrape-ground to detriment of everyone.

From what I see, the real AI, the ones users don't see, but the ones powers research infrastructures or working on private data are doing some real work and creating real results is the future. The LLMs many see are toys when compared to these ones.

I have seen AI models run experiments in lab, the boring vision models like YOLO creating real predictions and insights about our ecosystem, and fine tuned local LLMs provide real insights.

Hardware companies are not in a bubble because of this, but user facing LLMs are in a bubble.