Business Ideas/Semiconductors & chips/India
A flagship phone tops out near 19 tokens a second on a small model, and memory sets that limit, not the AI chip. App teams planning on-device AI features do not know which of theirs will fit.
Research Radar/Semiconductors & chips/United Arab Emirates · USA
A review of chiplet security says guards built into the base under the chips cost about 4 percent of speed. The paper runs no new experiment, so every number comes from work it cites.
Research Radar/Semiconductors & chips/India
A review of AI chips shows memory bandwidth, not arithmetic, sets the speed of language models. A flagship phone tops out near 19 tokens a second, and an H100 near 24 on a 70 billion parameter model.
Explainer/Semiconductors & chips
Moore's Law said the number of transistors on a chip doubles about every two years. Here is what Gordon Moore really claimed, which half of it has stopped, and what makes chips better now.
Explainer/Semiconductors & chips
An NPU is the part of a phone or laptop chip built to run neural networks on very little power. Here is what it does, how it differs from a GPU, and how to watch yours working.
Explainer/Semiconductors & chips
A semiconductor conducts electricity only when you tell it to, which is what makes a switch with no moving parts possible. Here is how silicon becomes a chip.
Explainer/Semiconductors & chips
A chip fab is the factory that prints computer chips onto slices of silicon. Here is how one works, why it costs billions, and why only a few places can build the most advanced kind.