Larry

They caught my boy @EventDrivenEddy in the wild

@HyperTechInvest Can we not make this sound like it happened yesterday. This happened literally months ago, started in the summer and was over by start of the year.

OpenAI is transitioning away from Datadog and into ClickHouse services, where $NBIS holds a 25% stake

@P_Remarks @akramsrazor @MadThunderdome Turns out it was the opposite. Lol

@P_Remarks @akramsrazor @MadThunderdome eBay is not acquiring GameStop lmfao

@akramsrazor @MadThunderdome I think someone will get desperate enough for a scaled user interface that pins gets viewed with a strategic lens too And the. eBay will acquire GME to expand its collectibles/pokemon trading card digital success into the real world

@S_curvecap Why does that feel wrong?

@bucketshopcap Dm

@LongOnlyLarry I have literally heard of multiple cases of TMT pod analysts being reallocated internally by their PMs. Maybe some have stayed but doubt in many cases their gross exposure is what it once was?

Will be interesting to see if all the mentally ill momo pod guys come back to software now after shamelessly dropping coverage/pivoting to semis. Prob need names to be up 200% first though, not 20%.

@Jespabe Okay so why are hyper scalers allocating (as a % of their capex) more absolute dollars incrementally to custom asics vs GPU’s?

@LongOnlyLarry Your argument would hold water if one chip ($NVDA) produced twice as many tokens as another, costing 3-4 times more, but that's not the case at all. Not to mention that ASICs aren't compatible with most uses because they were developed for a specific purpose.

Now that the day is over here’s my thoughts on the reaction: think the market is right NVDA is likely fucked here. Even if the exact number of chips isn’t correct and the commute cost is less than Deepseek is letting on there’s still enough proof there that you can get similar output using lagging edge chips so what’s the point of paying the high price tag of a hopper chip when you can get similar output using clusters of ASICS or even AMD chips. Don’t think the AI theme is over, but think the hyper scalers recalibrate where the proportion of their spend goes towards. Think Broadcom and Marvell end up big winners from this development which would make the sell of a prime opportunity to maybe long AVGO & MRVL and dare I say short NVDA?

@GoBirds555 @S_curvecap @TwannsWorld Because they talk to management intra q and when they move numbers people think it’s mgmt talking up or down numbers

@LongOnlyLarry @S_curvecap @TwannsWorld So then why do people make a big deal about sell side estimates moving one way or another intra quarter if nobody is using that info to buy the stock?

These post earnings moves have become so confusing to me. Expectations into the print change overnight Why do we have consensus estimates if they no longer matter? Is sell-side supposed to be an idea generation profession or is it just information ratification after the fact?

@GoBirds555 @S_curvecap @TwannsWorld Consensus estimates have never mattered because they’re not the ones buying the stock. They are just an information dissemination vehicle

Maybe I’m retarded (most likely) but can somebody smarter than me explain @LongOnlyLarry @S_curvecap @TwannsWorld

@S_curvecap Makes 0 sense

@ThainRos More spend has been repositioned to custom asics

@LongOnlyLarry Deepseek fears sent $nvda to $90s . People said deepseek would take big share, noone would need expensive chips. See sales of chips since

@marsrocc @TBU12345678 Tbu better to answer that q I don’t do any semis anymore

@LongOnlyLarry @TBU12345678 Crazy call on AVGO and MRVL. U think thats still where the trade is at?

biggest takeaway from tonight is that the $NVDA monopoly is over given how these hypescalers are talking about them so dismissively

@SorenCookie @TBU12345678 1.3 years is not not a year and a half and + it was a long duration call and I was at a long duration investment shop

@LongOnlyLarry @TBU12345678 Early is as bad as being wrong in this market. Almost too early by 1.5 year or perhaps 2 years

@ThainRos How did this age poorly?

@LongOnlyLarry Aged badly. And my position is that it will look even worse in a few years . They are the best at what they do , and there will always be opportunities for the those who master the frontier of their craft

@a16z Where Datadog?

Workday is arguably the most important and least loved product in enterprise software. More than 10,000 organizations run it, tens of millions of employees live inside it, it's approaching $10 billion in annual revenue, and it's incredibly sticky. HCM is the last large enterprise software category without a serious AI-native challenger, and that’s about to change. @joeschmidtiv on Workday's last workday: a16z.news/p/workdays-las…

@ragingbullcap DoorDash another really good example of this btw

You made 10x your money in two years if you bought this when FCF bottomed out and held to the highs. Good reminder that money is made not by lazily putting a multiple on backwards looking financials, but by having a variant view on the true earnings power of a business in the future. The consensus view at the bottom was that SPOT would never make money even though the inflection was rather obvious to anyone not looking in the rearview mirror.

@GergelyOrosz @Pragmatic_Eng Everything about this is so unbelievably wrong

Waiting when the CFO will notice the ~$170M Datadog bill after this... (yes, really! As shared in @Pragmatic_Eng back in October... though safe to assume OpenAI has reduced this since)

Yeah I’d shutter my fund too if I was long 200 bucks of dynatrace