korrents

semiconductors

The chips and the companies that make them: NVIDIA, TSMC, ASML, and what a fab can and cannot do.

FilterEveryone, all time111 from 35 people
Group All Technologists 11Scientists 1Economists & investors 1Founders & executives 4Writers & journalists 7
Who Everyone DPDylan Patel 22 JHJensen Huang 22 BPBrian Potter 9 DODoug O'Laughlin 7 LTLip-Bu Tan 7 JYJon Y 6 DHDavid Heinemeier Hansson 3 BGBrendan Gregg 2 CHCasey Handmer 2 DLDaniel Lemire 2 DRDax Raad 2 JDJeff Dean 2 NLNathan Lambert 2 SNSatya Nadella 2 ADAswath Damodaran 1 AVAustin Vernon 1 BRBarry Ritholtz 1 BDBrad DeLong 1 DWDan Wang 1 DCDavid Crawshaw 1 FVFederico Viticci 1 GHGeorge Hotz 1 GOGergely Orosz 1 HFHenry Farrell 1 JGJeff Geerling 1 KBKent Beck 1 KJKetan Joshi 1 LFLex Fridman 1 MBMiles Brundage 1 PLPieter Levels 1 SSSalvatore Sanfilippo 1 SLSergey Levine 1 TLTobias Lütke 1 TCTyler Cowen 1 ZMZvi Mowshowitz 1
When All time Past 1 yearPast 3 yearsPast 5 years
  1. HF

    Henry Farrell quoted

    Professor of international affairs at Johns Hopkins SAIS and co-author

    Our readingPower maneuvering over chokepoints increases the risk of broader confrontation.

    Their wordsPower maneuvering over chokepoints doesn’t simply corrode the global relationships that make miracles like the semiconductor supply chain possible. It increases the risk of broader confrontation.

    ↗The chokepoint view has a massive blind spotprogrammablemutter.com

  2. 4 weeks earlier
  3. GO

    Gergely Orosz quoted

    Writes The Pragmatic Engineer

    Our readingAI data centers have shifted from about 1 CPU per 8 GPUs to roughly 1:4, and the ratio could shrink to 1:1.

    data centers

    Their wordsAI data centers used to have a ratio of 1 CPU to 8 GPUs. Now the ratio is more 1:4, and it could shrink to 1:1.

    ↗The Pulse: a new trend of CPU shortagesblog.pragmaticengineer.com

  4. 2 days earlier
  5. BR

    Barry Ritholtz quoted

    Co-founder, chairman and chief investment officer…

    Our readingInvestors are underestimating Nvidia's ability to keep innovating and defend its dominant position as AI shifts toward inference.

    investing

    Their wordsBut I think people right now are, for instance, underestimating Nvidia's opportunity to innovate as well.

    ↗Transcript: Glen Kacher, CIO of Light Street Capitalritholtz.com 2nd of 3 in this piece

  7. TC

    Tyler Cowen quoted

    Economist at George Mason University

    Our readingA ban on self-recursive improvement in AI models would create a big incentive for some other parts of the world to invest a lot more in compute.

    Their wordsThe policy creates a big incentive for some other parts of the world to invest a lot more in compute.

    ↗Banning self-recursive improvement in AI models?marginalrevolution.com

  8. 1 day earlier
  9. JG

    Jeff Geerling quoted

    Developer, author and YouTuber covering Raspberry Pi

    Our readingGetting memory timings right for modern LPDDR RAM chips is technically difficult, and a marginal chip or a bad solder joint can cause unpredictable failure modes.

    Their wordsMemory timings, especially for modern LPDDR chips, are tricky to get right, and having a bad solder joint, or a cheap RAM chip that only barely passed validation, means you get weird failure modes.

    ↗Raspberry Pi locks down Pi 5 RAM upgrades in firmwarejeffgeerling.com 1st of 2 in this piece

  11. FV

    Federico Viticci quoted

    Founder and editor-in-chief of MacStories

    Our readingThe M5 Ultra Mac Studio is a massive leap in performance for local models powered by MLX.

    Their wordsIf you fall on the other end of the spectrum, though, and if you think this kind of stuff is neat…let me tell you: the M5 Ultra Mac Studio is a massive leap in performance for local models powered by MLX, and I have a few examples to prove it.

    ↗M5 Ultra Mac Studio Review: The Dream Mac for Local AI Agentsmacstories.net

  12. 2 days earlier
  13. DL

    Daniel Lemire quoted

    Computer science professor who works on fast data processing

    Our readingCPU performance keeps improving quickly rather than stagnating.

    Their wordsIn effect, what I am demonstrating is that CPUs are not boring. They are improving fast.

    ↗How did Apple Silicon get 50% faster in three years?lemire.me

  14. 3 days earlier
  15. KJ

    Ketan Joshi quoted

    Freelance climate and energy analyst in Oslo

    Our readingThe climate impact from NVIDIA's AI products in 2025 is comparable to that of a major state-owned coal company.

    Russia

    Their wordswe produced our own conservative estimate, finding that the climate impact from 2025's use of NVIDIA'S AI products is on par with the state-owned Russian Coal company.

    ↗NVIDIA’S AI Climate hoax – reportketanjoshi.co 1st of 2 in this piece

  17. BD

    Brad DeLong quoted

    Economics professor at the University of California

    Our readingIf AI hardware depreciates faster than assumed, the economics of the debt-financed AI capex boom invert and today's leverage turns reckless.

    Their wordsIf the useful life of an NVIDIA H100-vintage cluster turns out to really be three years rather than six, the capex cycle's economics invert, and the leverage that looked prudent looks reckless.

    ↗DataCenter Risk Is the Real Thing: CHART OF THE DAYbraddelong.substack.com 3rd of 3 in this piece

  19. ZM

    Zvi Mowshowitz quoted

    Writes Don't Worry About the Vase

    Our readingJensen Huang is a maximally cynical commercial and political actor whose one and only goal is to sell Nvidia chips to whoever he can, especially China.

    China

    Their wordsJensen is a maximally cynical commercial and political actor whose one and only goal is to sell Nvidia chips to whoever he can, especially China.

    ↗Trump Goes Full Hoax on AI Existential Riskthezvi.substack.com

  20. 1 day earlier
  21. DL

    Daniel Lemire quoted

    Our readingSubnormal floating-point numbers remain a real performance cost on Intel processors, but not on AMD or ARM processors.

    Their wordsThus it appears that on the latest AMD and ARM processors subnormals might not be a concern. But they remain very much a performance issue under Intel processors.

    ↗Subnormal floating-point numbers are expensive… on Intel processorslemire.me

  22. 3 weeks earlier
  23. DH

    David Heinemeier Hansson quoted

    Creator of Ruby on Rails, CTO of 37signals

    Our readingIntel's Panther Lake has finally matched Apple on power efficiency, so a new laptop should have one.

    Apple

    Their wordsWhat's really important is that if you get a new computer, a new laptop, you get one with an Intel Panther Lake chip. That is the revelation that um just leaped in the Intel side forward. Intel has finally matched Apple on efficiency in terms of uh power usage

    ↗Is Linux the perfect operating system? - Omarchy interview with DHHyoutube.com 6th of 6 in this recording

  24. 4 weeks earlier
  25. JD

    Jeff Dean quoted

    Chief Scientist at Google DeepMind and Google Research

    Our readingInference, not training, is now the constraint, and specialized low-energy hardware will beat general-purpose GPUs and TPUs on latency.

    energy

    Their wordsYeah, I mean it's a little different, but I think uh you're going to see more and more uh uh high performance and um low energy uh inference hardware systems because I think everyone is now realizing that inference is the key to making you know these agent-based systems be available to more and more people and that latency is really important and that specialization of the hardware is a really key way you can make uh things that are more energy efficient and lower latency than more general purpose uh computational devices like say GPUs or TPUs

    ↗Jeff Dean: The 1% Rule for Building in AIyoutube.com 2nd of 22 in this recording

  27. JD

    Jeff Dean quoted

    Our readingA chip that can do nothing but low-precision dense linear algebra is enough for machine learning, even though it cannot run Chrome.

    Their wordsAnd um if you build a specialized chip for low precision dense linear algebra and can't do anything else that turns out to be really useful for machine learning inference uh even though it can't run Chrome or Word or whatever.

    ↗Jeff Dean: The 1% Rule for Building in AIyoutube.com 4th of 22 in this recording

  28. 4 days earlier
  29. JH

    Jensen Huang quoted

    Co-founder and CEO of NVIDIA since 1993

    Our readingThe business is not building a great chip; it is accelerating an algorithm domain.

    Their wordswe realized early on uh that it's not about building a great chip, it's about accelerating an algorithm domain.

    ↗Jensen Huang: The Mindset That Built NVIDIAyoutube.com 2nd of 20 in this recording

  31. JH

    Jensen Huang quoted

    Our readingA computer company has to live five to ten years in the future, because a system takes three years to build, two to ramp, and is then used for a decade.

    Their wordsWe we kind of have to live in the future 5 to 10 years because it takes three or so years just to build a system, takes a couple years to ramp it up, and you're dealing and you would like them to be able to use the computer for 10 years after.

    ↗Jensen Huang: The Mindset That Built NVIDIAyoutube.com 11th of 20 in this recording

  33. JH

    Jensen Huang quoted

    Our readingPhysical AI will become one of the largest industries in the world and NVIDIA's next hundred-billion-dollar business, in more than three years but fewer than ten.

    Their wordslikely this will be one of the largest industries in the world and um, uh, it'll take longer than a couple two, three years. It'll take less than 10. And so this will this will be our next $100 billion business.

    ↗Jensen Huang: The Mindset That Built NVIDIAyoutube.com 16th of 20 in this recording

  34. 4 weeks earlier
  35. KB

    Kent Beck quoted

    Creator of extreme programming and test-driven development

    Our readingAn outside expert's understanding of somebody else's problem is bounded, because they do not have to live with the design after they leave.

    design

    Their wordsYour understanding of somebody else's problem is bounded because you're not in the middle of it. You don't have the same skin in the game. If you're not semiconductor test engineer, you don't have as much skin in the game as somebody who is because they're going to have to be using this interface for a long time after you're gone.

    ↗How Kent Beck shapes the software engineering industryyoutube.com 6th of 28 in this recording

  36. 13 days earlier
  37. LT

    Lip-Bu Tan quoted

    Chief executive of Intel since 2025

    Our readingA government taking a stake in its chipmaker is ordinary infrastructure policy — Taiwan, Japan and Singapore all did it.

    governmentinfrastructureJapan

    Their wordsJust I explained to President Trump, TSMC when they started they have the Taiwan government as a shareholder. If you look at Japan, you look at Singapore, this is the infrastructure US government get to provide the support.

    ↗Re-engineering the Semiconductor Supply Chain with Intel CEO Lip Bu Tanyoutube.com 1st of 16 in this recording

  39. LT

    Lip-Bu Tan quoted

    Our readingA foundry is a trust business before it is a technology business: a customer whose wafers yield badly misses its own revenue.

    Their wordsIt's a service business. It's a trust business. If people want to give you, you know, orders to have wafer to come, if the yield not good, they will be toast in term of revenue miss.

    ↗Re-engineering the Semiconductor Supply Chain with Intel CEO Lip Bu Tanyoutube.com 2nd of 16 in this recording

  41. LT

    Lip-Bu Tan quoted

    Our readingSelling silicon is no longer enough — customers now ask for the whole rack, so a chip company has to become a full-stack company.

    Their wordsAnd eventually, you have to really move into a full stack. So, not just a silicon, you need to have a software, and some of the customer asked me, "Give me the whole rack." So, there's a system that you have to build.

    ↗Re-engineering the Semiconductor Supply Chain with Intel CEO Lip Bu Tanyoutube.com 3rd of 16 in this recording

  43. LT

    Lip-Bu Tan quoted

    Our readingSemiconductor infrastructure has not kept up with AI's growth, and closing that gap needs capacity built in unconventional ways.

    infrastructure

    Their wordsHe and I, we share the same view that semiconductor infrastructure actually is not catch up with the AI growth.

    ↗Re-engineering the Semiconductor Supply Chain with Intel CEO Lip Bu Tanyoutube.com 4th of 16 in this recording

  45. LT

    Lip-Bu Tan quoted

    Our readingThe binding constraints on AI's growth are power, helium and memory — and almost nobody has noticed the helium one.

    Their wordsOne is of course everybody knows power constraint. Some country the power they just don't have that. They get impacted. And then secondly, a lot of people didn't realize the helium impact can be also very significant for semiconductor. And then the thirdly, is everybody know right now memory is a bigger shortage.

    ↗Re-engineering the Semiconductor Supply Chain with Intel CEO Lip Bu Tanyoutube.com 6th of 16 in this recording

  47. LT

    Lip-Bu Tan quoted

    Our readingShrinking below today's leading edge is still reachable, but it will be so expensive and difficult that no company can do it alone.

    Their wordsSo I think I can see you know, right now we have 18A and then now we do going to production of 14A. I can see 10 and 7. And so I think that path I think we can get there, but going to be more and more expensive and more difficult to do. And that's why we need partners. We cannot just do it ourselves alone.

    ↗Re-engineering the Semiconductor Supply Chain with Intel CEO Lip Bu Tanyoutube.com 8th of 16 in this recording

  49. LT

    Lip-Bu Tan quoted

    Our readingThe way to invest in semiconductors is to find where the bottleneck is, and right now the bottleneck is power and thermal.

    Their wordsSo I think power, thermal, those become the bottleneck. So I think I always look at from what is the problem we try to solve? Is it real? Is customer crying for it?

    ↗Re-engineering the Semiconductor Supply Chain with Intel CEO Lip Bu Tanyoutube.com 10th of 16 in this recording

  50. 3 weeks earlier
  51. DR

    Dax Raad quoted

    Builds developer tools — SST, and now OpenCode at Anomaly

    Our readingInference is a high-margin business even for a middleman: renting GPUs at scale, some models carry an eighty per cent margin over their sticker price.

    Their wordscuz because we rent GPUs at scale to run the models and we still use middleman by the way. So we're not like going all the way down to the down to the floor. Even for us there are some models the sticker price and the cost to us there's like an 80% margin in there.

    ↗Building OpenCode with Dax Raadyoutube.com 10th of 24 in this recording

  53. DR

    Dax Raad quoted

    Our readingDemand for inference may be growing exponentially while GPU production can only grow linearly, and the tightening happens where those two lines cross.

    Their wordsThe demand for inference is growing. So, like I don't think it's linearly growing. I think it might even be exponentially growing. But we haven't made our production of GPUs grow exponentially. That's like kind of a linear process. So as those lines intersect, there's going to be uh tightening.

    ↗Building OpenCode with Dax Raadyoutube.com 12th of 24 in this recording

  54. 2 days earlier
  55. SS

    Salvatore Sanfilippo quoted

    Programmer; wrote Redis and hping, and blogs about C

    Our readingNVIDIA hardware is unlikely to become cheaper, keeping local AI data centers out of reach for small companies.

    data centers

    Their wordsThere is almost zero hope that NVIDIA setups will get less expensive, and even a small company can't afford to easily purchase and handle a small data center for local inference.

    ↗Distributing LLM inference in DwarfStarantirez.com 1st of 2 in this piece

  56. 6 weeks earlier
  57. JH

    Jensen Huang quoted

    Our readingNvidia's business is the transformation of electrons into tokens, and its job is to do as much as necessary and as little as possible in between.

    Their wordsmy mental model of our company the input is electron the output is tokens that is in the middle Nvidia and our job is to to do as much as necessary as little as possible to enable that transformation to be done at incredible capabilities

    ↗Jensen Huang – Will Nvidia’s moat persist?youtube.com 1st of 24 in this recording

  59. JH

    Jensen Huang quoted

    Our readingNvidia's suppliers make their investments for Nvidia and nobody else because only Nvidia's downstream demand can absorb what they build.

    Their wordsNow why are they willing to make the investments for me and not someone else and the reason for that is because they know that I have the capacity to buy it buy their supply and sell it through my downstream.

    ↗Jensen Huang – Will Nvidia’s moat persist?youtube.com 3rd of 24 in this recording