Tap a claim on the ring to put it at the centre.
← Falling back to older 7-nanometre chips would not rescue AI compute,…
17 connected korrents · 18 moments from 12 Jun 2017 to 5 Oct 2026.
Everything filed under semiconductors
semiconductors
Everything filed under China
China
Everything filed under Japan
Japan
Everything filed under permitting
permitting
Everything filed under data centers
data centers
Everything filed under benchmarks
benchmarks
Everything filed under LLMs
LLMs
Everything filed under AGI
AGI 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: 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. 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: Inference, not training, is now the constraint, and specialized low-energy hardware will beat general-purpose GPUs and TPUs on latency. — tap to centre the map on it Inference, not training, is now the constraint, and specialized low-energy hardware will beat general-purpose GPUs and TPUs on latency. Last stated 3 months ago
30 Jul 2026
JD
Jeff Dean — holds since 2026-07-30 — tap for who they are
Same subject: Semiconductor infrastructure has not kept up with AI's growth, and closing that gap needs capacity built in unconventional ways. — tap to centre the map on it Semiconductor infrastructure has not kept up with AI's growth, and closing that gap needs capacity built in unconventional ways. Last stated 4 months ago
18 Jun 2026
LT
Lip-Bu Tan — holds since 2026-06-18 — 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 faster AI progresses, the better NVIDIA does, so a shock like DeepSeek should expand its market rather than shrink it. — tap to centre the map on it The faster AI progresses, the better NVIDIA does, so a shock like DeepSeek should expand its market rather than shrink it. Last stated 2 years ago
3 Feb 2025
NL
Nathan Lambert — holds since 2025-02-03 — tap for who they are
Same subject: Criticism that a simpler, scaled-up architecture underperforms a heavily hand-optimized baseline actually demonstrates why that architecture-scaling research matters. — tap to centre the map on it Criticism that a simpler, scaled-up architecture underperforms a heavily hand-optimized baseline actually demonstrates why that architecture-scaling research matters. Last stated 5 years ago
11 Jun 2021
GW
Gwern — holds since 2021-06-11 — tap for who they are
Same subject: Traditional performance engineering methods are insufficient for the scale of modern AI datacenters and need new approaches. — tap to centre the map on it Traditional performance engineering methods are insufficient for the scale of modern AI datacenters and need new approaches. Last stated 8 months ago
6 Feb 2026
BG
Brendan Gregg — holds since 2026-02-06 — tap for who they are
Same subject: The local AI niche will end up dominated by whichever open-weight model runs practically fast on high-end consumer hardware, like a Mac or a 'GPU in a box' rig. — tap to centre the map on it The local AI niche will end up dominated by whichever open-weight model runs practically fast on high-end consumer hardware, like a Mac or a 'GPU in a box' rig. Last stated 5 months ago
14 May 2026
SS
Salvatore Sanfilippo — holds since 2026-05-14 — tap for who they are
Same subject: There will be no overnight intelligence explosion, because a model's best work takes so much test-time compute that time itself is the bottleneck. — tap to centre the map on it There will be no overnight intelligence explosion, because a model's best work takes so much test-time compute that time itself is the bottleneck. Last stated 4 months ago
26 Jun 2026
NB
Noam Brown — holds since 2026-06-26 — tap for who they are
Same subject: A country is primarily for its native people — tap to centre the map on it A country is primarily for its native people Last stated a year ago
15 Sept 2025
DH
David Heinemeier Hansson — holds since 2025-09-15 — tap for who they are
Same subject: American shipyards are unproductive rather than expensive: their workers are paid about what Korean and Japanese shipyard workers are paid. — tap to centre the map on it American shipyards are unproductive rather than expensive: their workers are paid about what Korean and Japanese shipyard workers are paid. Last stated 3 years ago
3 Jan 2024
AV
Austin Vernon — holds since 2024-01-03 — tap for who they are
Same subject: Chinese pop culture will not experience an absolute flowering like 90s Japan because CCP censorship makes it boring — tap to centre the map on it Chinese pop culture will not experience an absolute flowering like 90s Japan because CCP censorship makes it boring Last stated 2 weeks ago
5 Oct 2026
NS
Noah Smith — holds since 2026-10-05 — tap for who they are
Same subject: A Chinese crewed landing on the Moon before America returns would be a defeat for the free world. — tap to centre the map on it A Chinese crewed landing on the Moon before America returns would be a defeat for the free world. Last stated 2 months ago
18 Aug 2026
CH
Casey Handmer — holds since 2025-10-31 — tap for who they are
EB
Eric Berger — holds since 2026-08-18 — tap for who they are
Same subject: A continental army's first mission is to keep the regime in power, not to defend the border — which is what the People's Liberation Army is for. — tap to centre the map on it A continental army's first mission is to keep the regime in power, not to defend the border — which is what the People's Liberation Army is for. Last stated 4 months ago
9 Jun 2026
SP
Sarah Paine — holds since 2026-06-09 — tap for who they are
Same subject: A country cannot fund its own industrial strategy and call the same policy cheating when China does it. — tap to centre the map on it A country cannot fund its own industrial strategy and call the same policy cheating when China does it. Last stated a year ago
15 May 2025
HR
Hannah Ritchie — holds since 2025-05-15 — tap for who they are
Same subject: A healthy attitude may be to assume a rule had a reason but default to removing it, since situations might change. — tap to centre the map on it A healthy attitude may be to assume a rule had a reason but default to removing it, since situations might change. Last stated a month ago
20 Sept 2026
AR
Armin Ronacher — holds since 2026-09-20 — tap for who they are
Same subject: A neighbourhood redevelops itself when rising land values meet permission to build more intensely, and it stagnates when either half is missing. — tap to centre the map on it A neighbourhood redevelops itself when rising land values meet permission to build more intensely, and it stagnates when either half is missing. Last stated 9 years ago
12 Jun 2017
CM
Charles Marohn — holds since 2017-06-12 — tap for who they are
Same subject: Certification, not engineering, is what gates revenue for a new aircraft company. — tap to centre the map on it Certification, not engineering, is what gates revenue for a new aircraft company. Last stated 4 years ago
30 Jan 2023
ED
Eli Dourado — holds since 2023-01-30 — tap for who they are
At the centre
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 Read this korrent →
Inference, not training, is now the constraint, and specialized low-energy hardware will beat general-purpose GPUs and TPUs on latency.
Holds
Semiconductor infrastructure has not kept up with AI's growth, and closing that gap needs capacity built in unconventional ways.
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 faster AI progresses, the better NVIDIA does, so a shock like DeepSeek should expand its market rather than shrink it.
Holds
Criticism that a simpler, scaled-up architecture underperforms a heavily hand-optimized baseline actually demonstrates why that architecture-scaling research matters.
Holds
Traditional performance engineering methods are insufficient for the scale of modern AI datacenters and need new approaches.
Holds
The local AI niche will end up dominated by whichever open-weight model runs practically fast on high-end consumer hardware, like a Mac or a 'GPU in a box' rig.
Holds
There will be no overnight intelligence explosion, because a model's best work takes so much test-time compute that time itself is the bottleneck.
Holds
A country is primarily for its native people
Holds
American shipyards are unproductive rather than expensive: their workers are paid about what Korean and Japanese shipyard workers are paid.
Holds
Chinese pop culture will not experience an absolute flowering like 90s Japan because CCP censorship makes it boring
Holds
A Chinese crewed landing on the Moon before America returns would be a defeat for the free world.
Holds
A continental army's first mission is to keep the regime in power, not to defend the border — which is what the People's Liberation Army is for.
Holds
A country cannot fund its own industrial strategy and call the same policy cheating when China does it.
Holds
A healthy attitude may be to assume a rule had a reason but default to removing it, since situations might change.
Holds
A neighbourhood redevelops itself when rising land values meet permission to build more intensely, and it stagnates when either half is missing.
Holds
Certification, not engineering, is what gates revenue for a new aircraft company.
Holds