|
Feb 19, 02:12
|
Should always be the default, but especially right now.
Unemployed Capital Allocator (@atelicinvest)
Since I love arguing / dunking so much, and there's dearth of actual real content on the non-bear side of argument, I've been writing a lot about the 'bull case' recently.
The truth is that if you're long, you should be spending the vast majority of the time on 'how does this new thing fuck me over?' - not 'holy fuck why are the bears so dumb'
|
N/A |
—
|
—
|
No analysis
|
|
Feb 18, 05:14
|
@TheLAPurchaser Yeah that ~MSD% delta is more consistent with what I've seen. Though the Rubin ICMS might change that if the attach rate is v high.
|
N/A |
—
|
✓
|
Batch analyzed (2 ideas in batch of 50)...
|
|
Feb 18, 05:04
|
Guess we're gonna find out soon just how much demand exceeds supply. https://t.co/nLquGMWmiN
|
N/A |
—
|
✓
|
Batch analyzed (2 ideas in batch of 50)...
|
|
Feb 17, 05:44
|
It's interesting watching how divergent AI vs robotics progress is in China.
AI is extremely resource constrained so Chinese labs are doing all sorts of groundbreaking research and pushing the field forward globally.
Robotics looks resource unconstrained so we get a bunch of
|
N/A |
—
|
✓
|
Batch analyzed (3 ideas in batch of 50)...
|
|
Feb 17, 00:33
|
@malinvested https://t.co/C6uaEPT2jN
|
N/A |
—
|
✓
|
Batch analyzed (2 ideas in batch of 50)...
|
|
Feb 14, 23:31
|
Can’t wait till LinkedIn finds out about this in 3 months https://t.co/b3Rj3l3RG5
|
N/A |
—
|
✓
|
The tweet expresses anticipation for a...
|
|
Feb 13, 02:40
|
$MU Wolfe conference
Q: do you still see your HBM share in-line with your commodity DRAM share this year?
A: "No, we've not given an update. As you mentioned, we said that when we launched initially in HBM for our high-volume products, we said that we would ramp our HBM3E 8-high
|
N/A |
—
|
—
|
No analysis
|
|
Feb 13, 02:25
|
An interesting point from the $MU CFO at Wolfe conference regarding LTAs and how this time is different. When $GOOG was spending $30bn in capex way back in *checks notes* 2023, it probably didn't care about LT visibility into memory supply. When GOOG is spending $180bn this year,
|
N/A |
—
|
—
|
No analysis
|
|
Feb 13, 01:17
|
Approaching village bicycle status. https://t.co/13vbvaf5jn
|
N/A |
—
|
—
|
No analysis
|
|
Feb 12, 11:25
|
https://t.co/JDFwE1s51r https://t.co/hBbShXwV64
|
N/A |
—
|
—
|
No analysis
|
|
Feb 12, 10:34
|
Yeah...Stripe is never coming public. https://t.co/CVZbUp77Bt
|
N/A |
—
|
—
|
No analysis
|
|
Feb 12, 04:17
|
$APP $U Listening to Adam and Matt answer questions, it is very clear who knows his business cold and who played D&D once when he was 12.
|
N/A |
—
|
—
|
No analysis
|
|
Feb 12, 04:01
|
$APP Adam basically confirming there is some hidden advantage of using Axon UA in MAX mediation that doesn't exist if using some other mediation. Not sure how grey this is, if APP is somehow giving itself an advantage in MAX auctions etc. But the upshot is that publishers can't https://t.co/tKomDI0GIT
|
N/A |
—
|
—
|
No analysis
|
|
Feb 12, 02:49
|
Feels like some important context was left off here... The HBM shortage/bottleneck is only half of the message. Below is the full response the SMIC CEO gave re the current memory shortage (translated using chatgpt). If the commodity DRAM and NAND shortage for PC and smartphones https://t.co/J4UbytNm6e https://t.co/GV6LTdQl4z
|
N/A |
—
|
—
|
No analysis
|
|
Feb 11, 22:53
|
What happens when $RDDT becomes increasingly filled with bots, AI slop and SEO slop? https://t.co/cAulQtXzrk
|
N/A |
—
|
—
|
No analysis
|
|
Feb 09, 07:20
|
And there we have it. $APP
https://t.co/Wd1Kwj5o9d https://t.co/JXOVhKLYPV
|
N/A |
—
|
—
|
No analysis
|
|
Feb 06, 20:56
|
https://t.co/mkhKbCCxdD
|
N/A |
—
|
—
|
No analysis
|
|
Feb 06, 20:56
|
Wrote a thought piece on $ADBE reflecting on why the stock moved from starter position into the too hard pile.
We turned our minds to 5 key questions re AI's impact on Adobe's core creative business. IMO any decision to own ADBE stock needs to be underpinned by the view that
|
N/A |
—
|
—
|
No analysis
|
|
Feb 06, 10:37
|
When you invest in global equities but live in Australia: https://t.co/efl4wU45fl
|
N/A |
—
|
—
|
No analysis
|
|
Feb 05, 06:18
|
"I'm pleased that we are collaborating with Apple as their preferred cloud provider and to develop the next generation of Apple Foundation Models based on Gemini technology."
So...how much is $GOOG fronting $AAPL's capex?
|
N/A |
—
|
—
|
No analysis
|
|
Feb 05, 05:26
|
$GOOG with users spending increasing amounts of time onsite in AIO and AI Mode, Google has an oppty to insert behavioural (targeted) ads into non-commercial queries, something it couldn't really do before because users would click off the SERP to get their information.
IMO this
|
N/A |
—
|
—
|
No analysis
|
|
Feb 05, 02:40
|
Does an OpenAI IPO increase or decrease the total amount of funding it can raise between now and whenever the business is self-funding? (assume when not if for this question)
|
N/A |
—
|
—
|
No analysis
|
|
Feb 04, 23:24
|
Everyone knows Zuck can be reckless but for the $GOOG to guide that level of capex...gives more credibility to the duration of the AI trade and the ROI that's on offer.
|
N/A |
—
|
—
|
No analysis
|
|
Feb 03, 22:47
|
Is $CSU cash flow real? If organic growth is LSD% and substantially all the growth (revs and EPS) is coming from acqn, then cash spent on acqn becomes capex.
On that basis, CSU had zero/negative FCF in 2022-24 and only this year has it done >$1bn real FCF. https://t.co/wXvlHrW7fz
|
N/A |
—
|
—
|
No analysis
|
|
Jan 29, 10:18
|
https://t.co/anrD7AMHwS
|
N/A |
—
|
—
|
No analysis
|
|
Jan 23, 05:07
|
Since @MasterPandaWu likes to amplify this CapitalWatch reporting, would love to hear their perspective on this, if they've looked into it:
$APP is the "forensic investigation" suggesting that Hao Tang and Huang Youlong are the same person?
Google Hao Tang: nothing
Google https://t.co/WsKUtLv3sw
|
N/A |
—
|
—
|
No analysis
|
|
Jan 20, 04:25
|
*Tang Hao. I'm embarrassing my ancestors.
|
N/A |
—
|
—
|
No analysis
|
|
Jan 20, 04:10
|
$APP is the "forensic investigation" suggesting that Hao Tang and Huang Youlong are the same person?
Google Hao Tang: nothing
Google Huang Youlong: 2021 red notice in France, "H." in Bordeaux extradition case, has a sister, implicated in Tuandai collapse, cousin of She Zhijiang
|
N/A |
—
|
—
|
No analysis
|
|
Jan 14, 07:55
|
Here's something I can't get my head around:
1) If incumbency and distribution matter so little in software because anyone can whip up [example application], how do these vibe-coded startups find a market and a biz model? You and your 3 mates can raise money to make something to https://t.co/qEAC3Vrx0R
|
N/A |
—
|
—
|
No analysis
|
|
Jan 08, 02:51
|
Oh. https://t.co/xsQKhcN9g5
|
N/A |
—
|
—
|
No analysis
|
|
Jan 08, 00:48
|
I understood almost 90% of that.
I don’t know what to do with it, but this has never happened before. https://t.co/Co9l84NZzn
|
N/A |
—
|
—
|
No analysis
|
|
Jan 07, 23:10
|
How will he do this if he loses the House? https://t.co/1AsXjZln1U
|
N/A |
—
|
—
|
No analysis
|
|
Jan 07, 06:16
|
Can anyone with access to Mac research verify the maths behind the 15GW constraint in this report?
400-500kwpm of HBM should be more than enough to produce 4,000 PB of HBM. 15 GW of Rubin NVL72 is likely only ~1,500 PB of HBM.
$MU $NVDA SK Hynix https://t.co/aeBQijOcqc
|
N/A |
—
|
—
|
No analysis
|
|
Jan 07, 01:36
|
Assuming all 6m Rubin packages are in the superPOD config which may not be the case.
|
N/A |
—
|
—
|
No analysis
|
|
Jan 07, 01:35
|
Is my maths right here?
9.6PB of flash in an ICMS for 8 NVL72 racks = each GPU package has 16TB
Say 12m Rubin die (6m packages) sold in a yr = 96EB of ICMS flash
Global NAND shipment is ~400EB and already supply constrained.
Where is $NVDA getting the bits from?
$SNDK $MU https://t.co/vG5la19NNl
|
N/A |
—
|
—
|
No analysis
|
|
Jan 03, 03:16
|
Surprised to not see more up 20% YTD posts today.
|
N/A |
—
|
—
|
No analysis
|
|
Dec 22, 00:39
|
"To understand just how profitable model inference can be, we ran our own back-of-the-envelope tokenomics analysis. Note there are a lot of assumptions that go into this – we are aiming for ballpark estimates here. In summary, depending on model size and API pricing or paying https://t.co/T67puyZmIW
|
N/A |
—
|
—
|
No analysis
|
|
Dec 18, 10:33
|
$MU "our contract structures that we are discussing are not like anything before. They are far stronger contract structures with specific commitments. And, of course, different from prior contracts is that those used to be like one-year contracts, and these are multi-year in
|
N/A |
—
|
—
|
No analysis
|
|
Dec 18, 01:17
|
For everyone still wiping clean their keyboards after the $MU result - here's a reminder of how memory trades:
The thing can derate 50% to offset a 100% guidance beat, and still wouldn't be at trough multiple. https://t.co/VQthvKssia https://t.co/GTEDzvsNwe
|
N/A |
—
|
—
|
No analysis
|
|
Dec 02, 21:51
|
Interesting. Autonomous delivery to fill unused capacity during ridehailing offpeak periods. https://t.co/xhGwpKtLj4
|
N/A |
—
|
—
|
No analysis
|
|
Dec 02, 20:53
|
Somehow we’ve gone from inflated Jensen Maths to hyperinflated IBM Maths. https://t.co/PSn6eICPQH
|
N/A |
—
|
—
|
No analysis
|
|
Nov 30, 22:59
|
Does anyone have a sense of what the financial arrangement for TPU procurement looks like?
Is $GOOG paying $AVGO for a completed TPU package? What components of that BOM are marked up vs passed through? Presumably GOOG isn't paying a 75% markup on HBM like NVDA customers are. https://t.co/fIjIq1NNjW
|
N/A |
—
|
—
|
No analysis
|
|
Nov 27, 06:27
|
I guess you'd argue those reserves ought to be set aside as restricted cash and therefore not "true" FCF.
|
N/A |
—
|
—
|
No analysis
|
|
Nov 27, 06:22
|
I'd like to understand this insurance angle more.
1) Insurance reserves are a creation of accrual accounting, not cash accounting. So the insurance expense in the P&L is non-cash and added back to OCF. But it is the net change in insurance reserves that is added back to OCF https://t.co/Yatzg4YYCc
|
N/A |
—
|
—
|
No analysis
|