most obvious overreaction from memory (ridiculous risk premium) but everyones too much a pussy to take the alpha. just buy and short neoclouds
once in a lifetime buying opportunity
most obvious overreaction from memory (ridiculous risk premium) but everyones too much a pussy to take the alpha. just buy and short neoclouds
once in a lifetime buying opportunity
extrapolating the short term price action across the entire ai trade is stupid.
> Ummmm… have you seen what has happened to all the bottleneck “bro” stocks? Maybe just maybe the market is pricing in a slowdown in the future?
the market is pricing in a higher risk premium. in the last week we had midterms and a black swan event
> Just look at memory stocks and their earnings reactions….
they have just been printing. microns going to retire 20% of their market cap in a year while markets are pricing in a worst case scenario (oversupply in H1 2028). personally i dont think the worst case is going to play out. either way at these expectations, the downside is relatively limited (still very unlikely, im happy to debate this) while the upside is enormous. risk to reward 101
fundamentally, ai cannot advance without resolving its infrastructure bottlenecks (memory, energy). i think you are looking at the dot com bubble and assuming the trade can collapse even while the technology continues to grow. the nature and economics of the ai buildout are structurally different.
the nature of the trade is highly capital intensive (incomparable to dot com). hence the insane earnings (and kinda the circular financing). however the economics hinges on the roi and while it is a concern right now, the solution isn't to stop building infrastructure. we have to continue building enough of it to bring down the cost of intelligence. intelligence has effectively unlimited demand and as the cost falls, entirely new use cases become economically viable (since intelligence is so fundamental)
as long as the neoclouds survive (which they should, see crwv and iren earnings), the ai trade is still on. none of the recent events change their near term operating outlook. by the time they do matter, the neoclouds would be in a significantly more stable financial position and micron probably would have bought back half their market cap
@treasury_bill@RJ9974896734346 so ai is going to forward but all the infrastructure behind it wont. are all the bottlenecks just going to magically resolve themselves
really no way? yes obviously they dont have that much cash on hand, but (as you said yourself) demand isnt collapsing, so asps arent going to drop tomorrow.
the longer you wait for prices to drop the more mc micron is buying back
looking at 2027 (already 75%+ sold out), the consensus is 275b rev and around 139b in fcf. they already mentioned 100% of fcf will be returned to shareholders via buybacks
current mc = 830b
(139 + 35) / 830 = 21% bought back
the entire point of investing is maximising risk/reward. you're focusing exclusively on the possibility of a cyclical downturn (again you havent mentioned any evidence other than historical precedent which ignores all structural differences) while disregarding all potential upside
@finding_value_@Lasa1985_@ThomasS98369110@james38051814 blud, theyre going to buyback their 1/2 marketcap and ur still worried about the stock being expensive
plus you're ignoring risk to reward. find a stock that is cheaper (risk to reward)
i fear we are nearing the end. while i'm bullish on memory, i'm increasingly concerned about neoclouds and their cost of capital.
although i feel much of the derisking and fud has been fairly irrelevant, the prevailing sentiment could make it increasingly difficult for neoclouds to access capital at a reasonable cost
fear is the mind killer. short crwv long micron
> Prices will collapse, not the demand
so demand continues to grow but supply outpaces demand
lets assume 0% growth in shipments and take microns most recent Q:
breakeven rev = cogs + opex = 7.18b + 3.30b = 10.48b
current rev = 54.23b
asp decline = 100% - 10.48b/54.23b = 80.7%
so are you expecting microns asps to decline by more than 80% within 4 years, with 0 shipment growth?
they have ltas extending till then...
@finding_value_@james38051814@Lasa1985_ honestly ur right.
lets ignore the structural changes (customer mix, hbm tradeoff) and pretend the market conditions are identical to previous cycles because it's historically cyclical?
im bullish on memory but this is pretty useless analysis. everyone knows the numbers look great. main concern is how sustainable these numbers are under a worst case scenario (e.g, neoclouds defaulting, disappointing ai roi -> lower demand -> oversupply). peg is extremely unreliable here
it's far more useful (especially when they are already discounted so much) to consider what the market is currently pricing in and assess the risk/reward
the market seems to be pricing in oversupply by H1 2028 which I think is overly pessimistic given the lack of concrete evidence supporting it beyond historical precedent (which ignores the structural differences from previous cycles with customer mix, hbm tradeoff, sca's)
demand growth is more evident to be extending beyond centralised compute (data centers) and into decentralised compute with soveriegn AI and potentially physical AI down the line
yep, thats just the nature of neoclouds. they're constantly needing to raise capital for infrastructure to remain competitive while their assets depreciate.
historically, crwv has repeatedly taken on more and more debt (and even atms) and i cant see why that changes in the short term. nbis is way better positioned but the fundamentals of neoclouds dont change