@TBU12345678 But also are things priced for a 7 yr cycle?
English
TheLAPurchaser
4K posts
TheLAPurchaser
@TheLAPurchaser
L/S Equity. Not financial advice.
Los Angeles, CA Katılım Nisan 2020
701 Takip Edilen6.7K Takipçiler
@TheLAPurchaser has there been a 7 year semi cycle…ever? will inflation not reaccel one time before then?
English
To the memory bears, since the convo is about the durability of the cycle, what happens to the demand function when agents and robotics take off in the ‘27-‘30 window? $MU $SNDK
TheLAPurchaser@TheLAPurchaser
$MU: We signed a LT 5-yr agreement that gives us confidence that the demand side function of the cycle is a lot bigger than anyone thought. we have more visibility into the biz than ever before. it only took (i) pricing going up 100% for a few qtrs in a row, (ii) every customer begging us for capacity, and (iii) visibility that things will continue to tighten into '27 for us to come to this conclusion. fast money:
English
@TBU12345678 I suspect a lot more supply will come online between now and then…
English
@PegasusFund SPX adds in the past 2 years have been so offensively stupid
English
Shoutout to the S&P, which is now down 75% on SMCI, which has always been obvious sanctions evasion scheme with prior instances of breaking the law
There is no way Jensen did not know about or at least suspect this
Pegasus@PegasusFund
PLTR up 500% YTD. MSTR up about 1000% YTD. AXON up about 100% YTD. Index selection clearly broken And TPL + SMCI earlier this year
English
@PythiaR Not so much this time is different or ends differently. It’s just lasts longer
English
Not all with 3-5 year contracts but all of them had various parties locking up into LTAs which either blew up or marked the top.
I’m also mindful of the capital cycle dynamic: cycles are born with a demand narrative and die because of supply additions. Every time the story is “you just don’t appreciate the demand strength” and the reason the cycle dies is supply is way more elastic than people think.
It’s never different but every time people say this time is different.
Maybe it’s different with AI but… the base rate on that bet is extremely poor.
English
@TheLAPurchaser Graphite electrodes
Memory
Container & dry bulk shipping
Lithium
Solar polysilicon
Offshore drilling
Asian Refiners post Russia invasion
There’s probably some others in individual commodities when stuff gets tight.
English
English
@TheLAPurchaser @chat_SBC UNH, the AI company?
English
this post getting lots of hate, but he is not wrong
implicit in high multiples for non-hypergrowth "compounder" businesses is durability of ROIC/CAP - based on five forces, non-cyclicality, asset-liteness - and smart inst investors realized that high PGR and low WACC mathematically meant these businesses were under-priced at almost any multiple as long as moat intact
that assumption will be likely be correct in the LT (for many of these) but there is a pullback in capital that thinks this way (some of it from flows, others from AI creating terminal worries) so the stocks trade same bc predicated on same meta
Mojo@MrMojoRisinX
Final answer: It's the same stock. MSFT/AMZN/V/MA/FICO/SPGI/JPM/BAC/HUM/UNH/TMO/DHR/DPZ/CMG all share nearly identical factor profiles of growth, momentum, quality, capital allocation discipline, institutional shareholder bases, et al.
English
English
@chat_SBC JPM is a capital light non cyclical business?
English
I rode stx from $70-100 to $280 in less than a year. Sold too soon ofc but fine bc 4 baggers are cool and no need to be greedy and have to manage risk.
Sold my ASML this month as well.
Always the possibility of leaving money on the table but I think the risk of overstaying on these things is generally higher than selling early (esp if using proceeds to rotate into better r/r stocks earlier in alpha curve).
To each their own though just not my style to try to squeeze every dollar out of a good chart.
English
@TheLAPurchaser @MRWahdy Yes but in deep cyclicals you usually have lost 20-30% before the datapoints ever hit
English
@evrgn11112231 @MRWahdy Yes agree it has wide r/r poles and one of the great things about public mkts is you can change your mind as thesis confirming/detracting datapoints come to mkt
English
I actually don’t have strong views about AI long term (kind of an impossible to answer S curve question) but I would consider myself bullish as I’ve owned semis for years. On one hand it’s obviously game changing and here to stay, on the other hand we have no idea what sustainable demand looks like bc again - none of these companies come close to making money. Most compute to date has actually been for training, who knows how sustainable that is. Will see on inference. This is like 2021 for VC backed loss making saas cos (and what happened to aws rev and capex from 21 to 22/23?)
My point on this is that for memory stocks in particular, your success is very path dependent. You either make a lot of money or suffer biggest losses of your career on an inherently unknowable future demand forecast. Everyone knows the next few Qs look great - it’s also not what matters for figuring out the final end point.
If you are actually AI pilled, far easier to just own meta imo bc they win either way but the value of this company with very high roic on this step change higher capex basis is enormous.
English
@evrgn11112231 @MRWahdy Not so much extrapolating as much saying the most memory intensive use cases are still on the come
Appreciate the disclosure
English
Not if it’s no sustainable and the demand evaporates one day bc it wasn’t real. Profits matter eventually.
Anyway, you asked the question initially about how memory bears think about demand and I answered.
I think the answer is time horizon and I am looking at a probability adjusted end point 5-10 years out while you seem to be infinite extrapolating recent datapoints. Different approaches.
English
@evrgn11112231 @MRWahdy And I think we do care about the LLM ROICs….if VC/PE is willing to fund it at high ARR multiples and eat the dilution in the formative years that’s a win for everyone
English
This all kind of comes back to what you think of AI, its utility, its benefits (productivity, etc), and what the tech delivers today and morphs into over time.
Sounds like you think like a lot of it today is fluff which i’d disagree with. And from there extrapolate may just be more hype in the future, which i again disagree with, but understand how you get to your conclusios from there….
I Think it becomes more obvious every qtr from
Here
English
Well, the demand is insatiable today, when tokens are heavily subsidized and people are using AI in large part due to fomo and its potential rather than what it is currently doing now. Also, capex takes awhile to flow through and very hard to know where real equilibrium is when early on the s curve (maybe much higher, but maybe not).
Roi on public cloud capex over the next couple of years is actually meaningless if the end users of that compute are themselves not making money and spending in unsustainable fashion.
English
@TheLAPurchaser The stock is dead money unless EP improves (it won't) or Champion is a blowout (it won't)
English
$BSX small long. Multiple sets you up to a much better r/r at 20x/18x ‘26/‘27 eps. CHAMPION probability of success is high — trial is overpowered. I’m not sure of the follow through b/c EP/Watchman #s don’t look great in Q1, and there will be (a decent amt?) of share normalization in EP (PFA) over the next 12-24mths. But most people are on the sidelines after being burned (myself included) by mgmt. And the R/R sets up well -10% by 30%. I think it’s an ok idea.
English
The demand is insatiable. everyone is brushing up against compute bottlenecks. The capex is front loaded. We haven’t seen the revenue acceleration in #s relative to the spend.
Point is you are saying you doubt the ROI but you haven’t seen the flow through yet which will be much more obvious to ppl in the coming qtrs
English
Not following. Capex leads revenue, revenue is mechanical output at least for public cloud and is a function of end demand (ie is oai and anthropic consumption and growth sustainable considering neither make money).
And then for meta we have no idea if their capex is sustainable or not and they will either earn great returns on it or they will cut capex by 50% or more in a few years and grow into it.
English
@evrgn11112231 @MRWahdy What about hyperscaler revenue acceleration?
English
Demand has negative convexity actually in memory stocks bc everyone expects the demand surprises (why else own memory stocks at huge multiples of book) but few are ready for hyperscaler capex reversal (very good chance they are likely all overbuilding just question of how much and when imo).
English
Prob a cover bid into the data readout next wknd…nobody wants to be short into that. From there you can chart your own destiny
English
@MRWahdy And I actually think there is huge question of demand since nobody has been able to predict the magnitude of the demand to date and the most productivity-boosting (and memory intensive) applications have yet to come
English
@MRWahdy Yes that is what I am asking
Have you mapped out the demand relative to the supply coming online…
English