• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

News AMD 4Q25 Earnings

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
No, profits are down the absolute toilet due to capex.

I don't think probability machines can do anything like that.

Profitability it fine, it is free cash that's getting clobbered. You do realize that even if you spend $200 billion on data center buildouts this year you aren't getting hit with $200 billion in expense on your balance sheet, right? These datacenters, AI chips, etc. are being amortized over years.

What I think is increasingly likely is that companies that turn out to have overinvested in AI will do massive writedowns down the road. We may see multiple new records set for quarterly losses due to those writedowns before the AI mania ends if it takes a while for the bubble to burst. The one thing that may save them from themselves is lack of sufficient power to continue building at the rates that are being projected for the next few years.
 
What I think is increasingly likely is that companies that turn out to have overinvested in AI will do massive writedowns down the road. We may see multiple new records set for quarterly losses due to those writedowns before the AI mania ends if it takes a while for the bubble to burst. The one thing that may save them from themselves is lack of sufficient power to continue building at the rates that are being projected for the next few years.
OR..they increase productivity per worker (with much less workers) by an insane amount and dont have to write down anything.
 
AI in its current form isn’t it. Don’t know how but the whole industry fell for the scam.

especially Microsoft, ruined their entire product and quality.
LLMs are useful for automating simple things but idk if ROI's there in general.
MS is just an OpenAI victim.
 
OR..they increase productivity per worker (with much less workers) by an insane amount and dont have to write down anything.
If they're increasing productivity per worker logic says to hire more people and grow as far as the market allows. Net layoffs suggests much worse.
AI in its current form isn’t it. Don’t know how but the whole industry fell for the scam.

especially Microsoft, ruined their entire product and quality.
Not the entire industry.
 
Last edited:
I'd certainly be interested to hear of successful applications.

I know it's fun to play around with, and can be useful for providing a base template for trivial coding tasks. But everyone I know working in computer hardware and software engineering just rolls their eyes and ignores the management directives to use AI to increase productivity. Not particularly surprising that those directives are most prominent at the companies talking up AI the most. It's like the CEOs have forgotten that their employees aren't investors that they need to talk up AI to at every opportunity.
 
I'd certainly be interested to hear of successful applications.
Beside code it is successful in generating engagement bait and comments. Especially on "platforms" where the "content" is promoted by a stochastic, multi-armed bandit.

That's why I am sure Google and Meta aren't entirely mistaken in spending so much on this. It can yield results for their businesses until pigs become picky in what they eat.
 
Just to update on status of Ai usefulness, I used GPT's image editor a couple of days ago to edit images of me in some safety glasses, and it did very well, beat nano banana for a particular look I was going for (Macho Man Randy Savage)...

But, I then loaded a .gif and asked it to edit each frame of the .gif with a specific tweak and it responded like it understood exactly what I wanted, but it spit out a gif that wouldnt animate and didnt even have the tweek visible on the one visible frame it had. I tried 3 more times to tell it what it spit out wasnt correct, and it responded as if it saw its mistake, but it failed miserably each time.

Then again, Waymo is actually real and driving on the streets alongside actual drivers, and theres been limited fatalities, so what do I know?
 
But, I then loaded a .gif and asked it to edit each frame of the .gif with a specific tweak and it responded like it understood exactly what I wanted, but it spit out a gif that wouldnt animate and didnt even have the tweek visible on the one visible frame it had. I tried 3 more times to tell it what it spit out wasnt correct, and it responded as if it saw its mistake, but it failed miserably each time.
Try it with Codex.
 
Last edited:
AI in its current form isn’t it. Don’t know how but the whole industry fell for the scam.

especially Microsoft, ruined their entire product and quality.
Scam? Insane statement to make.

95% of the code I push to production for my company is written by AI. In the last 5 months, LLMs have completely transformed the software creation industry.
 
Last edited:
If they're increasing productivity per worker logic says to hire more people and grow as far as the market allows. Net layoffs suggests much worse.
The roles they require are different than what the companies need. Companies need fewer software engineers, fewer customer service, more data center related roles.
 
Back
Top