Tap a claim on the ring to put it at the centre.
← Moore's law now yields about 25% a year, so 10x and 100x leaps can…
17 connected korrents · 16 moments from 11 Oct 2018 to 26 Aug 2026.
Everything filed under semiconductors
semiconductors
Everything filed under scaling laws
scaling laws
Everything filed under software performance
software performance
Everything filed under data centers
data centers
Everything filed under LLMs
LLMs Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Read this korrent: Moore's law now yields about 25% a year, so 10x and 100x leaps can only come from changing the algorithm and how it is computed, every single year. Moore's law now yields about 25% a year, so 10x and 100x leaps can only come from changing the algorithm and how it is computed, every single year. Last stated 6 months ago
15 Apr 2026
JH
Jensen Huang — holds since 2026-04-15 — tap for who they are
Same subject: Blackwell is 50 times Hopper while the transistors themselves improved 75% in three years — the gain is architecture, not lithography. — tap to centre the map on it Blackwell is 50 times Hopper while the transistors themselves improved 75% in three years — the gain is architecture, not lithography. Last stated 6 months ago
15 Apr 2026
JH
Jensen Huang — holds since 2026-04-15 — tap for who they are
Same subject: Moore's Law is hammering the prices and increasing the capabilities of stuff you buy from Dell and others — tap to centre the map on it Moore's Law is hammering the prices and increasing the capabilities of stuff you buy from Dell and others Last stated 3 years ago
19 Dec 2023
DH
David Heinemeier Hansson — holds since 2023-12-19 — tap for who they are
Same subject: Falling back to older 7-nanometre chips would not rescue AI compute, because the real gap between chip generations is twentyfold, not the threefold the flops suggest. — tap to centre the map on it Falling back to older 7-nanometre chips would not rescue AI compute, because the real gap between chip generations is twentyfold, not the threefold the flops suggest. Last stated 7 months ago
13 Mar 2026
DP
Dylan Patel — holds since 2026-03-13 — tap for who they are
Same subject: The software industry is massively underperforming its hardware, by factors of ten to a hundred, and people refuse to believe the numbers when you say them. — tap to centre the map on it The software industry is massively underperforming its hardware, by factors of ten to a hundred, and people refuse to believe the numbers when you say them. Last stated 2 months ago
26 Aug 2026
CM
Casey Muratori — holds since 2026-08-26 — tap for who they are
Same subject: The thing worth finding is the exponential with the steepest curve that will run the longest — decades ago that was Moore’s law, and today it is AI progress. — tap to centre the map on it The thing worth finding is the exponential with the steepest curve that will run the longest — decades ago that was Moore’s law, and today it is AI progress. Last stated 3 months ago
29 Jul 2026
AW
Alexandr Wang — holds since 2026-07-29 — tap for who they are
Same subject: The world can make roughly 200 gigawatts of AI chips a year by 2030, which makes a gigawatt a week a plausible quarter of the market rather than a fantasy. — tap to centre the map on it The world can make roughly 200 gigawatts of AI chips a year by 2030, which makes a gigawatt a week a plausible quarter of the market rather than a fantasy. Last stated 7 months ago
13 Mar 2026
DP
Dylan Patel — holds since 2026-03-13 — tap for who they are
Same subject: The same law that made chips cheaper made the factories that produce them enormously more expensive. — tap to centre the map on it The same law that made chips cheaper made the factories that produce them enormously more expensive. Last stated 2 years ago
3 May 2024
BP
Brian Potter — holds since 2024-05-03 — tap for who they are
Same subject: Most of the gain in tokens per dollar per watt comes from software, not from new hardware — 5x, 10x, sometimes 40x on one model family. — tap to centre the map on it Most of the gain in tokens per dollar per watt comes from software, not from new hardware — 5x, 10x, sometimes 40x on one model family. Last stated 11 months ago
12 Nov 2025
SN
Satya Nadella — holds since 2025-11-12 — tap for who they are
Same subject: "Scaling" was powerful because it was one word: naming a research direction is what tells a whole field what to do next. — tap to centre the map on it "Scaling" was powerful because it was one word: naming a research direction is what tells a whole field what to do next. Last stated 11 months ago
25 Nov 2025
IS
Ilya Sutskever — holds since 2025-11-25 — tap for who they are
Same subject: A benchmark result should be reported under a stated budget, or as a curve against test-time compute — never as a single number. — tap to centre the map on it A benchmark result should be reported under a stated budget, or as a curve against test-time compute — never as a single number. Last stated 4 months ago
26 Jun 2026
NB
Noam Brown — holds since 2026-06-26 — tap for who they are
Same subject: A lab should spend most of its compute on research rather than on building the next model, because research is where the tenfold yearly efficiency gains come from. — tap to centre the map on it A lab should spend most of its compute on research rather than on building the next model, because research is where the tenfold yearly efficiency gains come from. Last stated 7 months ago
13 Mar 2026
DP
Dylan Patel — holds since 2026-03-13 — tap for who they are
Same subject: AI progress is faster than people expect, and ordinary scaling can be enough to solve problems that looked very hard. — tap to centre the map on it AI progress is faster than people expect, and ordinary scaling can be enough to solve problems that looked very hard. Last stated 2 years ago
2 Feb 2025
AC
Ajeya Cotra — holds since 2023-08-29 — tap for who they are
SA
Scott Alexander — holds since 2022-09-12 — tap for who they are
MB
Miles Brundage — holds since 2025-02-02 — tap for who they are
Same subject: As models keep growing, AI is running out of enough high-quality unique training tokens to keep up. — tap to centre the map on it As models keep growing, AI is running out of enough high-quality unique training tokens to keep up. Last stated 4 months ago
24 Jun 2026
LW
Lilian Weng — holds since 2026-06-24 — tap for who they are
Same subject: Current techniques restrict pre-trained representation power because standard language models are unidirectional. — tap to centre the map on it Current techniques restrict pre-trained representation power because standard language models are unidirectional. Last stated 8 years ago
11 Oct 2018
KT
Kristina Toutanova — holds since 2018-10-11 — tap for who they are
MC
Ming-Wei Chang — holds since 2018-10-11 — tap for who they are
KL
Kenton Lee — holds since 2018-10-11 — tap for who they are
JD
Jacob Devlin — holds since 2018-10-11 — tap for who they are
Same subject: Models resemble each other because pre-training is the same everywhere; what differentiates labs now is RL and post-training. — tap to centre the map on it Models resemble each other because pre-training is the same everywhere; what differentiates labs now is RL and post-training. Last stated 11 months ago
25 Nov 2025
IS
Ilya Sutskever — holds since 2025-11-25 — tap for who they are
Same subject: One pre-trained robot model now matches or beats the specialists that were fine-tuned with reinforcement learning for the very tasks they were built for. — tap to centre the map on it One pre-trained robot model now matches or beats the specialists that were fine-tuned with reinforcement learning for the very tasks they were built for. Last stated 2 months ago
12 Aug 2026
CF
Chelsea Finn — holds since 2026-08-12 — tap for who they are
Same subject: Pre-training is a crappy evolution: the practically buildable substitute for the process that gave animals their built-in hardware. — tap to centre the map on it Pre-training is a crappy evolution: the practically buildable substitute for the process that gave animals their built-in hardware. Last stated a year ago
17 Oct 2025
AK
Andrej Karpathy — holds since 2025-10-17 — tap for who they are
At the centre
Moore's law now yields about 25% a year, so 10x and 100x leaps can only come from changing the algorithm and how it is computed, every single year.
Holds Read this korrent →
Blackwell is 50 times Hopper while the transistors themselves improved 75% in three years — the gain is architecture, not lithography.
Holds
Moore's Law is hammering the prices and increasing the capabilities of stuff you buy from Dell and others
Holds
Falling back to older 7-nanometre chips would not rescue AI compute, because the real gap between chip generations is twentyfold, not the threefold the flops suggest.
Holds
The software industry is massively underperforming its hardware, by factors of ten to a hundred, and people refuse to believe the numbers when you say them.
Holds
The thing worth finding is the exponential with the steepest curve that will run the longest — decades ago that was Moore’s law, and today it is AI progress.
Holds
The world can make roughly 200 gigawatts of AI chips a year by 2030, which makes a gigawatt a week a plausible quarter of the market rather than a fantasy.
Holds
The same law that made chips cheaper made the factories that produce them enormously more expensive.
Holds
Most of the gain in tokens per dollar per watt comes from software, not from new hardware — 5x, 10x, sometimes 40x on one model family.
Holds
"Scaling" was powerful because it was one word: naming a research direction is what tells a whole field what to do next.
Holds
A benchmark result should be reported under a stated budget, or as a curve against test-time compute — never as a single number.
Holds
A lab should spend most of its compute on research rather than on building the next model, because research is where the tenfold yearly efficiency gains come from.
Holds
AI progress is faster than people expect, and ordinary scaling can be enough to solve problems that looked very hard.
Holds
As models keep growing, AI is running out of enough high-quality unique training tokens to keep up.
Holds
Current techniques restrict pre-trained representation power because standard language models are unidirectional.
Holds
Models resemble each other because pre-training is the same everywhere; what differentiates labs now is RL and post-training.
Holds
One pre-trained robot model now matches or beats the specialists that were fine-tuned with reinforcement learning for the very tasks they were built for.
Holds
Pre-training is a crappy evolution: the practically buildable substitute for the process that gave animals their built-in hardware.
Holds