Dylan Patel

NGOs are NGMI

Being in SF is like being in Wuhan right before the pandemic Something is happening, it's gonna hit everywhere but so few people know it

> be me, 45th & 47th president of America > open the national Claude Code terminal > it’s gold-themed > “kill Mojtaba” > tools: palantir.exe, https://t.co/vPvJr0VRIs, netanyahu.friend > --fast-mode > --dangerously-skip-permissions > --spend-limit $1B > --credit-card powell_amex https://t.co/NzZ13iKmD2

This is crazy because it's like Claude Code for shoes https://t.co/BUdECooScE

We're in the Golden Era of AI recommending products to purchase. Soon we will be in the dark era of agentic commerce monetization slop.

Meanwhile Anthropic secondary sales are going poorly for the buyers. I'd be surprised if there's even 20% fill on that round. You're downright stupid to sell any Anthropic secondaries given the growth rate and how underpriced the last round was.

In early 2023 I wrote this about Google and TPUs Man I feel like such an Oracle because someone at Google told me they read this and started changing their strategy. From all 4 being facts to changing their strategy on a dime is so amazing https://t.co/apxaMC1nJE https://t.co/RoRDuwDYRY

Who t f wrote this cause it wasn't @marksaroufim lmfao

The number 1 reason Cursor and Thinking Machines should merge is not because one has amazing product and the other research. It's because they have strong culture fits, they both are no shoes in the office.

Debunking the false narratives around AI Datacenters. First it was that water usage is high, but it's not. This time it's about power prices. The narrative around powe pricing is silly and politically charged with no actually backing Manipulated statistics by people with agendas

That's our wife now

I don't give a fuck about War Claude I dont give a fuck about the Ayatollah Tell me about Tower 7 right now

I was literally just about to go to sleep... now I'ma be up for the next few hours https://t.co/141LIkrJq1

OAI right now: https://t.co/JBRUCr0KgT

Many Anthropic researchers like @TrentonBricken don't use sunscreen. The explanation they often give is "because of AGI" Today we finally understand why. War Claude will make sunscreen irrelevant when the nuclear radiation kills us all.

SemiAnalysis x Fluidstack are kicking off GTC with A Full-Stack AI Infra GPU Hackathon Power to Prefill, Dirt to Decode Build with the best, win prizes, and hear @marksaroufim GPU MODE, @cHHillee Thinking Machines, Thomas Raoux OpenAI, @garywu Apply below https://t.co/8ITqQLwo50

@firstadopter @roboco23 @fabknowledge Yes!

@roboco23 @fabknowledge That is gen z baby

@roboco23 @fabknowledge Millennial?!?

Went to my cousin's baby shower and got to see all my nephews incl my brother's baby. https://t.co/u9zCpWGnhx

Finally I can have some input on the skincare conversation Thank you @LOrealUSA and Nvidia @LOrealGroupe @lorealparisfr https://t.co/OPq4KVmes6

Crazy this 1 meal is 2,700,000,000 output tokens of Deepseek v3.2 Except I got grilled cheese so I'm saving the enviroment? https://t.co/PmNW1BmDLx

@Rhcarj @omercheema @imnotharsh @SemiAnalysis_ What's shallow about showing 18A node perf vs N3B and LBT saying later risk prod? Maybe he misspoke, we say this in the thread, but why should we ignore his speech?

@Deep_Burner @omercheema The article discloses everything. Just read it. We also spent a lot of time saying why it's not close to production yet and all the technical challenges to get there

Hynix and Micron objectively took density and power / perf lead in 2022 away from Samsung until at least 26. So we were right. The Substrate post said multiple times they had a lot to still prove, what exactly they hadn't executed on that they still had to, and that they were many years away from anything production. Not sure where you got ASML competitor from. That's the goal and we shared their progress towards it. Substrate paid me in 2022 for some work i did for them then, not in 2025.

@mayorhardin We put it out in a private report for clients. AMD brought it out publicly ofc I'ma defend myself

@hardiksd @EnerTuition @SemiAnalysis_ https://t.co/Pcqt8fCxKU

@EnerTuition @SemiAnalysis_ No thanks We have people who worked at ASML, Lam Research, TSMC, Intel, Nvidia. We understands the data. Our clients understand what we said.

@EnerTuition @SemiAnalysis_ That's there and detailed in a client note.

@tszzl Nah it's still worse for day to day use. Claude code is more than just for coding