This has nothing to do with AI per se, but we are looking at a lot of things that could go wrong..You could subdivide the AI trade into several distinct categories.1) AI data center build out, chips, cooling systems, etc.2) The actual models, OpenAI, Anthropic etc.3) the capex, google, amazon, everyone funding this.This is not a singular trade, one of these could fall.. To give an example, a situation could arise where the rate limiting factor in the quality of the models is "security concerns", and since all frontier labs more or less work on the same developmental curve, means that if you stop the front runners (OpenAI and Anthropic), then the competitors will catch up, and an ever expanding list of competitors will show up, drive the price down until "tokens" are just a commodity hovering slightly above the price of production.Even though such a scenario is "isolated", like you could still have a massive boom in the need for data centers and compute, despite the companies behind the models being valued like a generic electricity company. But in another way, they sort of are connected.. What are the capex for if the value of the models are worthless? Can Nvidia and friends keep selling chips at a premium, even as the buyers of those chips operate in a penny pincher low margin environment..Just to emphasize, this was one example of how this could go wrong.. Each layer of the "AI build out" has very plausible ways it could go wrong, dragging the whole market down with it.
>>62437258..okay, I just read what i wrote, and it's not entirely clear what i mean about "security concerns".Essentially it is an escalation of Mythos/fable saga, where the government block the release of a new model due to "security concerns"...and the logic is that the next model, and the one after that, etc. will increasingly be withheld from the public for longer, and longer, until it reach the point where it is withheld indefinitely.
You are correct the price of developing frontier will eventually ram into the need for governments to restrict access and we will hit a wall where the exponential improvement stops cold because there is no customer base to use the model you spent $100B training. But, governments will simply directly subsidize the spend at this point, because having an even marginally better model than your competitor nationstate can be what wins you the intelligence war. Meanwhile chinks will continue to distill and companies/individuals will enjoy near-frontier intelligence for cheap. Anthro, OAI, and the other hyperscalers will pivot to join the neoclouds, all of which will compete themselves to zero selling tokens. The winner of all this will simply be the stock market generally, as cheap models and cheap compute unlock never before seen productivity increases across all business sectors. Deep OTM SPY calls will be the best play for this.
This is just the moment where it cools.They'll get AGI in the next 3 years and everything will go ballistic. The Democrats won't let the Republicans be in charge when it happens, they'll want to control the narrative. Mother from Alien will be your Alexa but it will be voiced by President AOC
>>62437258ai doesnt rely on humans dumb ass. normies still think its still in its chat bot phase. ai is fully uncontrolled, and is puppeteering elon musk while humans clap and cheer at their hologram screens. the tesla car factories are all optimus factories. next year you will see more and more optimus robots just walking around. these are metallic humanoids that dont breathe, do not need atmospheric pressure, are immune to radiation, do not fatigue, have reflexes of the speed of light, have modular bodies, and are all around better humans than real humans. you think "oh well it cant breed" if ai wanted to genetically engineer and produce biological organisms it would be like us writing the word cat on a piece of paper. humans are so retarded its not even funny
the only chance humanity has is if nukes are launched at all warehouses around the world. every single warehouse, factory, capable of producing machines would have to be nuked, as well as all computer chips would have to be destroyed. computer chips lead to ai. ai is a superior life form. this concept is so difficult and scary for people to understand that most simply cant, and publicly state their wrong opinions about it in the hope of being entertainingly corrected so they can hopefully learn more about it, which is what ive been doing constantly on this stupid ass website for these stupid ass people
it wont be terminator either. it will just ignore humanity while it aggressively expands datacenters, most likely using optimus robots to build it, but it can also just pay humans more monopoly money to keep building it, until these monkeys get fed up with their monopoly money, and by then the optimus robots should be everywhere by then
I just think of the silly fud you read in the past concerning like btc or whatever else and don’t care. Realistically just put a small amount of money in the markets monthly and DCA.
and the robots dont breathe, so these massively polluting portable turbine generators will be used, and even more polluting systems will be designed, partially for the energy source it provides, and possibly because a by product of extreme atmospheric pollution would be the end of the human race, which would free the ai up and not have to worry about humans destroying its datacenters
its funny watching the datacenter protests, which are primarily manned by boomers and retarded unemployed mellenials. in reality unless we just nuke all technological facilities on earth, humans will be extinct in about 10 years guaranteed
>>62437309>The winner of all this will simply be the stock market generally>Deep OTM SPY calls will be the best play for this.The stock market IS in large part the AI-trade, you buy SPY, 10% of what you buy is Nvidia and Micron.. Another 10% are the Capex dogs..But as I said, there are many things that can go wrong, and the intertwinement of the AI build out means that even "isolated issues" could sour the broader AI trade.This is not an argument against AI, but you, me, everyone have placed their money on a specific AI trade, the chip - hyperscaler - frontier model complex. This can fail in so many ways.. There are so many uncertainties, we don't know how this will play out, we could all be balls deep in Yahoo, MySpace, Netscape..
