That AI isn’t AI at all. I hate when people call it AI - it’s like the whole redundant ‘what is God’ argument.
This shit isn’t intelligent. You don’t ask someone how many “r’s” are in the word “strawberry” and it take 3 hallucinations before it finally “get it.”
Nor do you ask intelligence, artificial or not, how many “e’s” are in the word “seventeen” and it confidently tells you 1, then 2, then reaffirms it’s 2 then finally realizes it’s 4.
This shit isn’t intelligent. It’s a cool useful little search tool or pattern recognition device. It can do a lot. But at what cost?
Companies left and right are cutting token spending because it is simply too much fucking money.
Measuring the ROI is impossible.
It can code somewhat decent big fucking whoop.
It can search human generated content and summarize it, though in my experience it is confidently wrong MANY MANY times - requiring multiples upon multiples of fact checks.
It’s not useless. It’s not some “artificial intelligence” either.
It’s somewhere in between.
Tech has no more revolutionary ideas.
Nobody wants AR glasses. No body wants this bullshit that tech companies are desperately peddling as the next “new” thing.
They’re out of ideas.
AI isn’t going away. LLM’s aren’t going away. But they aren’t consciousness. And they DAMN sure aren’t going to generate the profits required to justify all these INSANE valuations and CAPEX spend.
Eventually Anthropic and OpenAI are going to show their numbers and in time everyone is going to see that there is simply no way to profit off of these tools in the way in which the markets are predicting.
They aren’t going away (the companies probably will the tools won’t). Opensource models, strong balance sheet companies like Google will utilize them, but from what i’m reading and seeing there is simply NO ROI and this will slowly be realized.
Thus, the semiconductor companies will crash massively. The hyper scalers will cut cap ex and the stock market in particular TECH will have a MASSIVE draw down.
AI has been the only thing propping up the markets. If these companies can’t show a MASSIVE ROI then this bubble pops extraordinarily hard.
These companies simply cannot sustain their valuations if earnings don’t continue to keep up. Eventually, this quarter, next quarter, the one after that, EVENTUALLY hyper scalers are going to CUT capex, these LLM labs are going to go public and as soon as SpaceX, Anthropic, OpenAI, Nvidia, etc - as soon as they show a flat quarter or a negative quarter, the ENTIRE semiconductor market and tech market at large and really, the entire market, is going to MASSIVELY implode. Being that AI and the entire infrastructure and supply chain of AI has been the driving force of these overpriced valuations and massive bullrun(s) we have seen in these names that are typically cyclical.
So unless some miracle of god happens and these companies can generate ludicrous amounts of revenue to make up for the massive spending this thing is going to hell in a hand basket.
If you look at TSMC, Nvida, ASML, Micron etc. The chip companies, the semiconductor companies - from top to bottom of the food chain, they have all exploded in returns over the last 6 years when AI began all the hype and the demand for chips and data centers took off.
Now we have “agentic AI”, we’re a good bit of time into that. Now we have OpenAI and Anthropic racing to IPO.
It’s all bullish and vibes right now. AI is this new “revolutionary” technology. Problem is, eventually the hype settles, EVENTUALLY companies need to report earnings to their investors.
Broadcom misses earnings. TSMC misses earnings. Nvidia’s demand haults.
Guess what?
We’re talking 30, 40, 50%+ crashes (or more) in some of these names who have zero moats or zero cash flow business outside of this insatiable demand for AI technology.
This technology, as cool and useful as it can be just simply doesn’t have the ROI for the massive amounts of capital it costs to run and deploy.
I truly do not see this becoming anything but a better search engine. A second pairs of eyes on an MRI scan. A cool text generator or chat app or good for self driving cars or whatever. A good buddy to code with side by side.
But it’s not this AGI nonsense the media and tech bros spew.
Maybe i’m wrong.
But in my opinion, it’s going to be companies like Google with actual business models and profits that can survive and implement these tools.
If you are simply JUST an LLM lab - GOOD LUCK - you have ZERO moat.
Open source models, European models, Chinese models, Gemini, Claude, ChatGPT, Microsofts model, Xai.
If LLMs aren’t that profitable in and of themselves there will be a few models that survive and the revenues will only be a small portion of their balance sheets as these AI models simply ARE NOT providing businesses with the return on the cost of their investments.
THUS, if enterprise AND ordinary users can’t justify token spend, then LLM labs can’t pay hyperscalers to rent data centers, hyperscalers can’t justify datacenter construction spend, memory and chip companies revenues SINK and the whole thing collapses leaving only those with a strong balance sheet to continue keeping the models running if they can find SOME way to justify doing so i.e. PROFITS.
The stock market doesn’t run on hopes and dreams. EVENTUALLY…. E V E N T U A L L Y -whether it is 6 months, 1 year, or 3 years from now, EVENTUALLY the profits and ROI are going to have to be proven and SHOWN to the investors via earnings reports and as soon as their is a CHINK in their hyped armor, this market being propped up by AI hopes and dreams comes CRASHING down in a MASSIVE way.
Not saying it’ll happen tomorrow. But it WILL happen and it is going to be BLOODY.