/biz/ and rRddit experts say that there is an AI bubble, while AI experts say there isn't one. Who should I trust?https://youtu.be/26a2jf1AXf8?t=415
Everybody thinks there's an AI bubble, therefore there isn't one.
>>62581462Redditors are going to return all that money to the hedge funds.
>man who benefits from ai says theres no bubblerevolutionary
and blockchain experts kept telling everyone that blockchain is gonna revolutionize everything
AI bubble confirmedhttps://www.twitch.tv/xrdmain
>>62581462>trusting chinamen
>>62581462There is something worse than just an AI bubble.
>>62585773Oh shit that delayed draw loan piece makes a lot of sense actually, I was wondering why I wasn't seeing a lot of PIK's in the companies loans if they were supposedly hiding default. This is what would link this problem to the entire syndicated loan space. Good post actually.
>>62585815>>62585773Very interesting, unfunded delayed draw facilities do not count towards credit rating. Coreweave just closed on its 5th new delayed draw facility in the syndicated loan market. This is probably how you hide defaults until it is too late yes. Although one would think under representing a companies available credit might be bullish?
Now that I think about it though, I do wonder if much like 2008 there is a degree of credit risk here. These are essentially bilateral deals between the loan agents and each lender, assuming one lender defaults on their obligations to fund the draw, the loan would presumably have other lenders need to step in and cover their portion, which could theoretically put more stress on the balance sheets of the other lenders in the transaction?
>>62585941Or vice versa the funding is just cut for the defaulted lender putting pressure on the company which expected to be able to borrow more? Either way that does have elements of the 2008 style chain reaction of defaults. I did post about this in /smg/ like literally 2 days ago actually, I think the similarities between how the loan market operates with like 3-4 weeks for settlement and the OTC derivatives market before it got reworked with dodd frank are really interesting. Certainly in the case of a rush to sell with immense volumes it could stress the market where there's super long chains of ownership that aren't recorded as settled yet that need unwound
there is a big bubble for projects outside of the hyperscalers. the hyperscalers have a customer problem and need to convert a certain number of jobs into compute. the hyperscalers will be in bubble territory if they cannot convert something like 5-10% of the workforce into compute.
>>62585773>>62585815these are real estate assets with interconnections. even if they never break ground there is a tenant who can take over removing some of the downside.>>62585941real assets. the 2008 crash was caused by non real hypothetical assets being nowhere near worth what they said they were. If we start seeing "data centers" pop up showing revenue but not actually drawing any power or generating tokens we are in trouble.>>62585949Remember once again these are industrial sites. Even if they never break ground there is a tenant who can fill these spots. You need to take an estimation on how much energy is needed even without data centers to sustain the economy in five years and then compare that to where we are. Nvidia is trying to convert the GPUs into a lendable asset for this reason and secure energy for compute.
>>62586025>convert something like 5-10% of the workforce into computeImagine US government allows 5-10% unemployment to happen.
>>62586034would cause less issues than we think. automation is coming for something like 20-30% of the workforce by the end of 2039 or so. If it happens like it should then we get a period of deflation and as boomers die off most people can be nudged out of work and then be retired 10-15 years early. Decline in population then sort of makes conditions tight for some workers but for others it is basically like nothing happened.
If millennials and Gen Z pop boomer stocks by talking shit about AI, boomers should cut your lazy ungrateful asses out of their wills and just blow all their money in Pattaya rolling in hotel beds with tight little brown women glazed in cum.
>>62581462There isn't an AI bubble, there's a human bubble.
>>62586033>the 2008 crash was caused by non real hypothetical assets being nowhere near worth what they said they wereBut these are not "real assets", these are unsecured loans to corporations based on their credit rating, which may be in question if the loan structure is hiding defaulted payments through delayed draw facilities
>NgHow the fuck am I, a huwhite man, supposed to pronounce this? The closest I can imagine is like saying "Nig" without the IYellows explain thineselves
Upon further research apparently these loans are also superseded by bank bail in laws for European lenders. Maybe this could be a vector, rather than companies defaulting triggering cascading defaults it could be lenders failing to honor their funding obligations?
>>62585941Not sure I understand all that but I read that Caterpillar, who makes the earth movers and such, has made a killing in the last year selling generators to data centers. When I looked into it it turns out CAT’s biggest business is actually as a financial services company. This equipment is so expensive its needs to be financed so they provide their own in house financing, all debt they keep on their books as guaranteed income.
>>62585479>person who can't afford to invest says investing is dumbenlightening
>>62587886Basically unlike regular loans, these delayed draw facilities will have an outstanding amount that the borrower can "draw" on, in which case the various lenders making up this loan need to pay their share of the borrowing from the company. That reddit post is postulating that many borrowers are relying on these facilities to essentially pay their loan payments with if they don't have the cash. I'm wondering what happens if some of the lenders fail to produce the cash when the borrowers request it?
>>62581462I don't think there's an AI bubble, just a LLM bubble. LLMs are not profitable and use a lot of resources for trivial and low value work. AI as a big data processing tool, like palantir and Amazon, improves logistical efficiency and that translates into return on investment. we're going to see a lot of failed LLM firms, a few big consulting AI firms similar to Deloitte but with big data focus, and a lot of internal local AI architecture and operation positions
>>62581462>Well dressed asian man says AI is in a bubbleAHHHHHHHH SAVE ME NIGGERMAN
>>62587886is it just vendor financing or are they following the GE Capital playbook
