> Execs Confused and Horrified by the Huge AI Bills After Thinking They Could Replace Workers for Free

Empty. You're out here selling free lunch AI where the only way the bill pencils is machines that work 24/7 and never need a bathroom break. That's just fire-everyone-keep-the-lights-on with better branding. I still read "worse than the run-" like you meant sock market while dudes in ties were buying mainframes. Zero replies already. Board doing god's work.

And even at the new higher pricing schemes the AI companies are still providing service at an operating loss while completely bullshytting the markets on their depreciation costs.

Meanwhile much cheaper Chinese competitors are gobbling up global market share.

The computer industry has always been 50% hype.  Remember MITI's Fifth Generation Computer Systems initiative?  Y2K?  Quantum computing? Big data? 

I took FORTRAN at the university using punch cards and met Evans & Sutherland ... now teenage retards edit images on the fly.  Maybe the hype is what has allowed dump trucks of money to keep things moving forward.   

wrote:

And even at the new higher pricing schemes the AI companies are still providing service at an operating loss while completely bullshytting the markets on their depreciation costs.

Meanwhile much cheaper Chinese competitors are gobbling up global market share.

I think DeepSeek was the crack in the dam. If you can get say 80%/90% as good for half the cost - how do you justify these insane data center build-outs. You can RAG/tool call locally and connect to it to pre-trained models via cloud - or some other combination of mixed use.

If you truly believe in some sort of AGI/Singularity - then investing will be useless as money will become obsolete.

., wrote:

I think DeepSeek was the crack in the dam. If you can get say 80%/90% as good for half the cost - how do you justify these insane data center build-outs. You can RAG/tool call locally and connect to it to pre-trained models via cloud - or some other combination of mixed use.

Meta selling GPU access is another crack in the dam.

Sockpuppet wrote:

I am not so sure.

Your guess doesn’t sound very informed.

I'm in the Zitron camp:

https://youtu.be/3u0KeTC7jso

I'm skeptical on the scaling thingy.  Take a billion LineShine (record holding Chinese supercomputer) machines all running since the the big bang; their output is DWARFED by the number of ways to shuffle a deck of playing cards.

Businesses are increasingly moving towards China's free & open source AI models, because the costs are dramatically cheaper. 

In fact, with China's open source AI models, businesses have the option to download everything for free, and run everything on their own hardware.

The US government response will be to ban US companies from using China's free & open source AI models.

-----

Chinese AI models are gaining ground with U.S. companies as OpenAI, Anthropic costs surge

Tue, Jul 7 2026

https://www.cnbc.com/2026/07/07/chinese … 180abb6e55 [Highly Credible Source (Liberal)]

-----

Lawmakers probe growing use of Chinese AI models in U.S. companies

Wed, Jul 8 2026

U.S. lawmakers are considering strategies to halt the growing adoption of Chinese AI models by homegrown companies.

Chinese models have gained traction among U.S. firms as they’ve closed the performance gap with American rivals while being cheaper to use.

https://www.cnbc.com/2026/07/08/chinese … 180abb6e55 [Highly Credible Source (Liberal)]

wrote:

The place I work goes back and forth between demanding everyone show how they're adding value with AI, and then panicking that the AIs aren't locked down hard enough.  No concrete written guidelines or demonstrations of the "right way".

They even put a question on the annual eval form asking people to explain how they were using AI to benefit the company.

I'm using AI to get dirt on management.

wrote:

The place I work goes back and forth between demanding everyone show how they're adding value with AI, and then panicking that the AIs aren't locked down hard enough.  No concrete written guidelines or demonstrations of the "right way".

They even put a question on the annual eval form asking people to explain how they were using AI to benefit the company.

CFOs are asking for real return on investment.

AI is wonderful, until the bill shows up.

wrote:

Businesses are increasingly moving towards China's free & open source AI models, because the costs are dramatically cheaper.

Local inference as the eventual winner?
Shove the inference costs to the end-user sucker.

., wrote:

I also just don't believe it's a "winner take all" situation - which is their sales pitch for burning money. There are many competing models out there, and many usage approaches that are continuing to evolve. That's why they're IPO-ing now at peak speculation.

What happens when there's no money left to burn?