|
Baddog posted:Lol, you all are gonna laugh - jesus christ, I need to get some more caffeine in me. I was going off some dumbass feed that made it look like earnings was *tonight*! Actually my *entire* feed was "NVDA earnings are here!!!1!" But earnings are actually still a week out, and next weeks puts are uhhh, quite a bit more realistic for the anticipated movement. Oof, if you keep making these mistakes maybe think about a broker that shows you were it is for your options plays (This is Tasty Trade, purple line with bell for earnings arrow points to after market)  They also shade the expected move which can also be useful for sanity checks and "this doesn't look right" before coffee. 
|
#
?
Feb 14, 2024 21:08
|
|
|
|
| # ? Jun 6, 2024 14:25 |
|
|
pixaal posted:Oof, if you keep making these mistakes maybe think about a broker that shows you were it is for your options plays (This is Tasty Trade, purple line with bell for earnings arrow points to after market) That's pretty slick. Yah, I expect as I get older gonna have more senior moments, have to get better about double-checking. Click my mouse, lose my house. Only a small bet here tho, eye balling just taking the loss early. This thing is way overbought on exuberant optimism, could very likely have a pullback before the weekend. Much more likely for the hype train to keep chugging for another week though. Which is why the puts were cheap!
|
|
#
?
Feb 14, 2024 21:18
|
|
|
UnfurledSails posted:SMCI is going to hit $1000. Nuts can it do so in the next 35 minutes please?
|
|
#
?
Feb 14, 2024 21:21
|
|
|
I bought A LOT of NVDA puts yesterday for September so expect that update in a couple months!
|
|
#
?
Feb 14, 2024 21:21
|
|
|
I get that NVDA it incredibly overvalued but are you guys sure about long term puts on them? AI feels like a gold rush that's only just getting started, and NVDA is currently the top pickaxe seller on the market.
|
|
#
?
Feb 14, 2024 21:32
|
|
|
Umbreon posted:I get that NVDA it incredibly overvalued but are you guys sure about long term puts on them? AI feels like a gold rush that's only just getting started, and NVDA is currently the top pickaxe seller on the market. The only difference between this thread and WSB is that we use punctuation.  (Yeah I'm not being totally fair, but really this is sort of a gambling thread. Nobody's sure about poo poo here.)
|
|
#
?
Feb 14, 2024 21:38
|
|
|
Umbreon posted:I get that NVDA it incredibly overvalued but are you guys sure about long term puts on them? AI feels like a gold rush that's only just getting started, and NVDA is currently the top pickaxe seller on the market. No I'm not sure about anything I'm just gambling efb edit again: also kinda hedging my AMD LEAPS that printed. Sold all of my 2024s today, keeping my 2025s. Assuming AMD go down if NVDA go down, there's still plenty of time to recover (and I guess I already took enough off the 2025 to be fine mentally if they crater and don't come back)
|
|
#
?
Feb 14, 2024 21:47
|
|
|
Garfu posted:I bought A LOT of NVDA puts yesterday for September so expect that update in a couple months! So drat expensive. Was trying to set up a reverse iron condor, because of course it seems ~80% that it'll be either 1500 or 500 in september. But when you punch in the numbers, it is hard to make it feel plausible at all. My inclination is always to try to sell IV when it is too high/hyped, and you're buying it here.  Losing money from 572 to 1227!! Garfu posted:also kinda hedging my AMD LEAPS that printed. This is good.
|
|
#
?
Feb 14, 2024 21:56
|
|
|
Baddog posted:So drat expensive. Was trying to set up a reverse iron condor, because of course it seems ~80% that it'll be either 1500 or 500 in september. I normally like my butterflies like 7-20 days out so this is probably not great but if you want to cover basically 700+  I tried to get a short butterfly working to get you a credit but I'm getting like 6,000 credit max win 10k, max loss 22k which is way outside budget I feel. Playing with it I'm not sure I like a position for less than 20k and that's too rich for me and like 10x the price you were going for in your example. The broken wing is a touch less than your example but as I said I'm generally doing single options that far out which is completely different. My Butterflies are generally open less than a week. I think the bet you want is like long 2 ATM short a 500 short a 1200 of whatever is skewed better that day puts or calls. The budget play is pick a side break a wing.
|
|
#
?
Feb 14, 2024 22:21
|
|
|
SMCI and ARM are making me look like I know what I'm doing.
|
|
#
?
Feb 14, 2024 22:32
|
|
|
Baddog posted:Lol, you all are gonna laugh - jesus christ, I need to get some more caffeine in me. I was going off some dumbass feed that made it look like earnings was *tonight*! Actually my *entire* feed was "NVDA earnings are here!!!1!" But earnings are actually still a week out, and next weeks puts are uhhh, quite a bit more realistic for the anticipated movement. lol. well, i guess you've got a chance. i'm waiting until after nvda earnings to buy puts. they probably still have another quarter of insane guidance growth before comps start to get tough. plus i bought a 4080 super two weeks ago and do AI stuff on their hardware on the regular. gently caress! but that p/s, it's silly, even with high projected 2024 growth in other news this is a great quote-tweet https://x.com/choffstein/status/1757894178648850681?s=20
|
|
#
?
Feb 15, 2024 01:24
|
|
|
The bubble can't pop until every potential bear has lost all their money trying and failing to short the market.
|
|
#
?
Feb 15, 2024 01:35
|
|
|
Femtosecond posted:Maybe anything gravel and concrete? Like the basics. Floods wash out highways and you need poo poo tons of gravel and concrete to rebuild that. Cement might be worth adding to that. It’s the critical ingredient in concrete and might be more likely to be made by a large company with a centralised kiln.
|
|
#
?
Feb 15, 2024 01:43
|
|
|
pmchem posted:
lol perpetual doomerism never ends
|
|
#
?
Feb 15, 2024 01:44
|
|
|
UnfurledSails posted:SMCI is going to hit $1000. Nuts 
|
|
#
?
Feb 15, 2024 21:44
|
|
|
lol Overlays remind me of this beauty https://clips.twitch.tv/CovertFancyPorpoiseFloof-v9xN1FhNAepuy5hw
|
|
#
?
Feb 15, 2024 22:57
|
|
|
There are a few times in history I've made big bets on something I felt I had 'insider' information on. My personal MVPs were AMD's incredible rise when they started releasing technical details of Zen, and calling the speculation on solar going into 2021. I'm making another pretty big bet here: NVDA is going to fall back to 300. The challenging part here is 'when'. There are a couple big reasons for this. First and foremost, AI is solving no real problems. What people call AI today is pattern matching against a big data set. It's put together in a fun way, but it's actual utility to solve real problems is kinda dogshit. The useful machine-learning adjacent applications don't need massive GPUs to power their capabilities, and the most promising spaces for model based solutions is probably image generation. This horseshit that is Bing and Windows 11's umpteenth assistant, now powered by AI -- that's all dead in the water in 3 years time. That is just a very compute heavy solution to pressing the 'I'm Feeling Lucky' button on google, with arguably worse results and no context behind said results. Furthermore, the way we generate AI Models today is a dead end. It's not going to go from AI today to actual AGI. AGI is a totally different path in a totally different solution space. A good enough LLM does not graduate to AGI - it just is more robust pattern matching. This poo poo is just not useful at the end of the day except in some relatively narrow applications. It's definitely not the future of 'interacting with everything'. Now, lets say AI actually lives a longer life. That also doesn't really matter because the actual valuable resource at play here for these model based solutions isn't necessarily compute - it's memory (and bandwidth, and some other things and there is nuance here, but...). There are physical limits to how much memory you can effectively attach to a chip, and while GDDR7 promises a 50% increase in density, it's not like GDDR8 is just around the corner. The utility of replacing your fleet of 4090s or H200s is going to be substantially less then buying into them in the first place. There is an asterisk here specifically with the datacenter chips, which are on HBM3e memory. It's more likely that Nvidia can produce a more interesting upgrade path for the purpose of AI in that space then they can in the consumer GDDR, PCIe based market. I don't have Nvidia puts yet, but I'm waiting for some kind of sign that this poo poo is all going to spin down real fast and result in a market with a massive oversupply of hardware. It'll happen and I'm going to be betting on it.
|
|
#
?
Feb 16, 2024 09:25
|
|
|
I can see a fair bit of commercial use for the current versions of AI, it's not great at what it does but it's good enough to replace copywriters, graphic design etc at least at the bottom end of the market. There will still be people willing to pay for an actual person to do that sort of stuff too but I am certain there's money to be made getting AI to churn out crappy marketing material cheaply. It wouldn't surprise me if eventually the entire services sector undergoes a bifurcation where the rich get stuff done by humans to a high standard and the poor make do with cheap AI equivalents of dubious quality. It also wouldn't surprise me if some niche uses for AI emerge where it really excels - I'm thinking dedicated models for data analysis and pattern recognition in scientific research or business intelligence. All of that is probably going to drive a certain level of demand for things like better chips, although maybe not as many as what is currently priced into companies like NVDA. I do think it's generally overhyped and much like the dot com bubble a lot of the stocks riding high on AI mania will come crashing down before too long.
|
|
#
?
Feb 16, 2024 12:13
|
|
|
i mean with nvidia puts the timing is the hard part, yeah. this article is one in a long line but it's fun if you also look at the dates/content of the 'further reading' links at the bottom: FT Alphaville Nvidia Nvidia is nuts, when’s the crash? https://www.ft.com/content/0cffec6e-ede8-4513-93b1-bb45f03e6595
|
|
#
?
Feb 16, 2024 13:28
|
|
|
pmchem posted:i mean with nvidia puts the timing is the hard part, yeah. this article is one in a long line but it's fun if you also look at the dates/content of the 'further reading' links at the bottom: Your going to have to quote what you want it needs a subscription.
|
|
#
?
Feb 16, 2024 14:04
|
|
|
Canine Blues Arooo posted:There are a few times in history I've made big bets on something I felt I had 'insider' information on. My personal MVPs were AMD's incredible rise when they started releasing technical details of Zen, and calling the speculation on solar going into 2021.  I have 110 late Sep puts let's do this! My wife's car payment depends on it!
|
|
#
?
Feb 16, 2024 14:40
|
|
|
pixaal posted:Your going to have to quote what you want it needs a subscription. Before the AI fever breaks few will bet against Nvidia’s market value rising further as stock market investors swoon at thoughts of the bot-overlord future. This week Nvidia’s market cap passed the $1.8tn mark, leapfrogging Alphabet — whose 2023 net income was greater than Nvidia’s 2023 revenues — to become the third most valuable US company after Microsoft and Apple. Nvidia options have gone wild — Tesla wild. Nvidia’s chips are essential to the current generation of machine learning models and their associated services, and so will earn the company extraordinary profits for the foreseeable. That seems to be the general idea at least. Pour cold water over it at your peril. Allow us instead to merely pass on a few back-of-the-envelope calculations about what that valuation implies from the tech-focused curmudgeons over at Chameleon Capital. Toby Clothier has pulled a dusty discounted cash flow model from a drawer and plugged in Nvidia’s numbers. To get to a $740 share price simply requires that the company maintain a monopolist-like operating profit margin of 55 per cent for the next decade, while also growing sales tenfold, from $60bn a year to more than $600bn. For context, the entire industry sold $527bn worth of chips last year, according to the the Semiconductor Industry Association. Over the past decade Nvidia did admittedly achieve a similar level of growth: in 2014 its sales were a mere $4bn. However, Clothier points out that Nvidia’s unusual profitability is a recent phenomenon related to the very high prices pushed through in response to overwhelming demand: The EBIT Margins were all over the place from 2014-2023 (range of 12-37 per cent) and certainly nowhere near a steady 55 per cent. So Nvidia shareholders are making a bet that the law of large numbers does not apply, and that competition, innovation, and pricing pressure will not come to bear until at least the mid-2030s. Good to know. Clothier’s discount rate is 10 per cent by the way. At a 15 per cent growth rate and 30 per cent sustainable margins his antiquated model cranks out a share price of $176 — purely as an example for discussion, not a target. Those who have made a fortune from Nvidia are very welcome to share their own assumptions in the comments. Further reading: This is nuts, when’s the crash? Nikola is nuts, when’s the crash? Tesla is nuts, will it ever crash?
|
|
#
?
Feb 16, 2024 15:04
|
|
|
From seeing some of the things the technical artists that were keen about AI at my old work were playing around with there's def some value in AI taking away some of the most dull and arduous work from artists (think: creating normal map textures for the surface of a space ship). What this means in terms of value is hard to say. If I had to guess at this early, early stage maybe one could get away with hiring one less artist at a medium sized games team if the whole team is super powered and more effective? Would scale as games get bigger, so those 700 person Assassin Creed teams probably benefit more. So there's def some value in an Enterprise solution for Adobe and other tool makers and they can charge quite a bit more on the art side. On the programming side I'm less sure. I think there's probably some value in systems writing boilerplate code, but programmers are untrustworthy and are gonna want to check that stuff anyway. But in the same way, maybe programmers get a bit more effective, but I see the potential being less than the art side. Perhaps there's an ability for completely mundane sorts of apps to be have huge amounts of them filled in by AI but that sort of thing would be a low value business area in the first place. So again another way for Microsoft to charge even more for Visual Studio. Is "Enterprise tooling gets another price level" enough to justify the NVDA price? I dunno. One of the most amazing things about AI art is its ability to create some amazing concept art, but you know I don't think companies were really spending too much effort on concept art to begin with so in many ways nothing is really going to change here except that some super early pitch powerpoint is going to have crazy good art instead of sketches. Nice for folks in the room, but doesn't really move the needle one way or the other in terms of the business. I agree with the person earlier who said that copywriters are going to have a bad time as the low value space surely gets taken over by AI. Long been annoying to see Realtors writing some bare minimum, typo ridden 200 word blurb into their home listings and still get a big percentage of the home sale despite their lack of marketing effort. I'm shocked if they're not all using AI to do this now and put even less effort into things. So yea to me it seems like the primary disruption is in spaces that are low value to begin with. In aggregate is it something or does a higher value space need to be disrupted for this stock price to be worth it?
|
|
#
?
Feb 16, 2024 18:11
|
|
|
Femtosecond posted:So yea to me it seems like the primary disruption is in spaces that are low value to begin with. In aggregate is it something or does a higher value space need to be disrupted for this stock price to be worth it? Don't forget many of these newer AI are more efficient and produce not just better results but with less power. They are buying more videocards because demand is going up for their product not that their product demands more power.
|
|
#
?
Feb 16, 2024 18:26
|
|
|
<taps thread title> as little as I know about options I dont think i'd call his SMCI position "risk-averse" (SMCI, for those not following, is up like 180% YTD, but down 20% today)
|
|
#
?
Feb 16, 2024 22:25
|
|
|
drk posted:<taps thread title> Me: "Risk averse" Also me: A 4-position portfolio consisting of 3 options and 1 penny stock.
|
|
#
?
Feb 16, 2024 22:50
|
|
|
Shroud posted:Me: "Risk averse" I like/own RKLB tho  They are still a >2B company, and have been launching consistently lately with a ton of good clients and more than a solid backlog. But drat, that stock just is not recovering.
|
|
#
?
Feb 16, 2024 22:58
|
|
|
The problem with AI as a commercial growth area is that the AI needed to do basic tasks like writing emails and marketing copy, can be trained on public data sets in a day or two, and can already run on consumer hardware, I have the orca 2 LLM running on my work MacBook pro and it will write you a haiku about any topic you can think of and write c++ to sort a binary tree. The Delta between free research models and pay stuff from Google and open ai is not very much until you get into the very nuanced stuffFemtosecond posted:I agree with the person earlier who said that copywriters are going to have a bad time as the low value space surely gets taken over by AI. A friend of a friends wife works for, if not grammerly.com then some just like it. Their whole business model is targeting both native born people with bad grammar, and probably more lucratively foreign born professionals that need to send out coherent communications in English That company has leaned in heavily into AI as ChatGPT does their services for free, but with some extra copy/paste involved My brother in law was born out of county and he pays like $30/mo for an AI service to write emails to, using his example, reschedule a meeting at the last minute, and division wide communications We have an internal tool for the analysts to upload a PDF and then query it for summaries or whatever. An analyst can do ~10 documents in a day vs ~3. We don't have fewer analysts, we just have vastly better data now gay picnic defence posted:It wouldn't surprise me if eventually the entire services sector undergoes a bifurcation where the rich get stuff done by humans to a high standard and the poor make do with cheap AI equivalents of dubious quality. Do they not have Walmart and Dollar General where you live Hadlock fucked around with this message at 23:14 on Feb 16, 2024 |
|
#
?
Feb 16, 2024 23:08
|
|
|
in the immediate future nvidia likely has two or three years of production already sold to the big cloud companies who are desperately trying to get more gpus live in data centers for cloud ai infrastructure in the longer term I have no idea if they have a technology moat that will last a decade, or if ai solutions that lean heavily on GPU will remain the big thing, or if some other big thing or innovation will come along but I put it to you goons that you probably have no idea either and betting for or against the movement of the entire software industry's whims is foolhardy at best
|
|
#
?
Feb 16, 2024 23:12
|
|
|
I don't think they have a lasting moat at all, but long dated puts are just way too drat expensive. I don't want to just short the stock because of uncapped risk and borrow fees. And something like NVD decays probably even faster than the put premium. Doesn't seem to be a good way to bet on "lower next year than this year". Trying to make money on stocks going down is *hard*. Probably also a good point that they have quite a backlog built up.
|
|
#
?
Feb 16, 2024 23:20
|
|
|
SMCI is the better short if you are going to short something. IMO. Also: Short Semiconductor Stocks With These Two ETFs: https://finance.yahoo.com/news/short-semiconductor-stocks-two-etfs-140002113.html
|
|
#
?
Feb 16, 2024 23:27
|
|
|
Nvidia in my opinion is the next meme stock similar to Tesla from years back. It will take a long time to correct itself. Yeah it's a real company with billions and physical products but the irrational mindset retail investors have will propel it to $1000/share this year.
|
|
#
?
Feb 16, 2024 23:32
|
|
|
Hadlock posted:We have an internal tool for the analysts to upload a PDF and then query it for summaries or whatever. An analyst can do ~10 documents in a day vs ~3. We don't have fewer analysts, we just have vastly better data now I just had an idea for a business that develops contract language under adversarial AI conditions so that the language of the contract is heavily favorable to you, but AI agents will ignore it in summaries.
|
|
#
?
Feb 17, 2024 00:13
|
|
|
I just don't see the long term market for high end hardware to train LLM. You can train a GPT3 (not 4, but 3) in two days on three A100, and then you can distribute the 2gb binary blob indefinitely for pennies, and it'll run on commodity hardware today It might* cost you $600 to train a custom bespoke model today but it's a one time fixed cost Maybe generating flawless 4K video in real time will still need beefy hardware but my prediction is that in 5 years LLM will be a fully solved problem until (if) we develop AGI *Based on public data on training times and current hourly rental rates for A100 Hadlock fucked around with this message at 00:22 on Feb 17, 2024 |
|
#
?
Feb 17, 2024 00:19
|
|
|
Baddog posted:I like/own RKLB tho They've seemingly have never had a profitable quarter? Per their last quarterly report, they are hoping to close the gap to "adjusted EBITDA break-even", which is better than where they are at but also yikes. Ex-SPAC also big red flag. What they do is admittedly pretty cool, but it feels uninvestable on anything but a speculative basis
|
|
#
?
Feb 17, 2024 00:26
|
|
|
Hadlock posted:I just don't see the long term market for high end hardware to train LLM. You can train a GPT3 (not 4, but 3) in two days on three A100, and then you can distribute the 2gb binary blob indefinitely for pennies, and it'll run on commodity hardware today Conflating "AI" generally and its huge range of applications with "LLMs" specifically (or with Generative AI, which I've also seen thrown in) is part of the muddy thinking going on in this thread that makes me very wary that folks actually know what the gently caress they're talking about. Siri and Alexa are AIs. The image processing in a high-end digital camera is an AI. The people doing cutting edge climate modeling are using AIs. They're using AIs to find cancer on MRI images. AIs are being used to design new chips. There are AI applications in quantum computing (which, lol, there's another money-pit buzzword tech to bet against, if we're looking for more). AIs are being used in automobile traffic monitoring & management, pathfinding, investment algorithms, political science, social sciences. Big Data and data analysis is a huge area for business, that does include or touch on some natural-language processing but that's not the whole thing. AIs in the realm of machine learning are driving thousands of different areas of study and development in basically every industry. Again I have no idea if any of this translates into sustained absurdly-high profits for NVidia, specifically, but if your first and main thought when discussing AIs is ChatGPT and Midjourney, you're looking at just the tiiiniest tip of the ice that has captured the public imagination and mistaking it for the whole iceberg. AWS thinks this is worth spending on https://nvidianews.nvidia.com/news/aws-nvidia-strategic-collaboration-for-generative-ai so does google https://cloud.google.com/nvidia so does microsoft (azure) cloud: https://www.nvidia.com/en-us/data-center/gpu-cloud-computing/microsoft-azure/ oracle https://www.nvidia.com/en-us/data-center/gpu-cloud-computing/oracle-cloud-infrastructure/ NVidia's biggest customers: https://qz.com/nvidia-generative-ai-google-microsoft-meta-1851206854  this ain't just chatgpt and its wannabe followers driving this level of investment in one year Leperflesh fucked around with this message at 00:46 on Feb 17, 2024 |
|
#
?
Feb 17, 2024 00:39
|
|
|
drk posted:They've seemingly have never had a profitable quarter? Per their last quarterly report, they are hoping to close the gap to "adjusted EBITDA break-even", which is better than where they are at but also yikes. Ex-SPAC also big red flag. They've been heavily investing in bigger and bigger rockets, so sort of expected to lose money for awhile. I think they are pretty close now though, yep.
|
|
#
?
Feb 17, 2024 00:41
|
|
|
Leperflesh posted:
All of these existed before GPT technology, they were just extremely expensive to develop and at best have barely acceptable results and intolerant of all but the highest quality data sets I guess my point is, the silicon used to consume a trained AI is already adequate and exists at best buy, and LLM can already absorb data sets like financial/economic data and give you pretty accurate outputs. Just because your LLM is only trained on all economic papers written through Q3 2023 doesn't mean it can't analyze financial disclosures from FY24 or 25, training a new LLM or buying an updated one might only happen annually. OpenAI is already only releasing updates to their GPT product every six months or so
|
|
#
?
Feb 17, 2024 01:02
|
|
|
Joink posted:Nvidia in my opinion is the next meme stock similar to Tesla from years back. It will take a long time to correct itself. Yeah it's a real company with billions and physical products but the irrational mindset retail investors have will propel it to $1000/share this year.
|
|
#
?
Feb 17, 2024 02:01
|
|
|
|
| # ? Jun 6, 2024 14:25 |
|
|
Subvisual Haze posted:Tesla showed the way to exploit the enormous passive investing system though. Invest in and then sell your crypto assets as necessary to obtain 4 successive quarters of slightly positive earnings to force inclusion into the SP500 index. Then the flood of passive SP500 investing will backstop your shareprice. Index effect is more or less dead: 
|
|
#
?
Feb 17, 2024 03:35
|
|