Moore's law
The observation that the number of transistors on a chip (and thus computing power) roughly doubles every two years, driving rapid technological progress.
As defined in cwt_patrick-collison
Other definitions
The observation that computing power roughly doubles at regular intervals, often invoked to predict rapid improvement in technology.
Mary Roach
Where it comes up · 95
- Semiconductors: Everything You Wanted to Know6 claims
- Cadence: Software Behind Semiconductor Design5 claims
- #8 The Intel Trinity: How Robert Noyce, Gordon Moore, and Andy Grove Built the World’s Most Important Company5 claims
- 325 | Alvy Ray Smith on Pixar, Pixels, and the Great Digital Convergence5 claims
- AMD: How Chips Are Changing4 claims
- ASML: Competing with Moore’s Law4 claims
- @Asianometry & Dylan Patel — How the semiconductor industry actually works4 claims
- Robin Hanson on the Technological Singularity 01/03/20114 claims
and 87 more
Most important claims · 10 of 162Search all claims for “Moore's law” →
- Bill Gates predicted that a pure software industry would arise, because software development costs were buried in hardware prices and software requires little capital, so selling it separately would spur innovation; this was his greatest contribution.#265 Becoming Steve Jobs: The Evolution of a Reckless Upstart into a Visionary Leader
- The split between fabless designers and foundries exists mainly because a fab costs about $20 billion, so the root cause is capital intensity rather than technology, letting each side specialize.Qualcomm: Making Smartphones Smart
- Semiconductor performance has doubled roughly every two years for about 40 years, but the cost of manufacturing has risen at nearly the same rate, which is what drove the separation of design and manufacturing.Qualcomm: Making Smartphones Smart
- Even ASML's top technologist, Martin van den Brink, has repeatedly underestimated how long Moore's Law would continue, first predicting at least 15 more years in 1999 and then about another 15 years 15 years later, showing that long-range forecasting in this industry is nearly impossible.ASML: Competing with Moore’s Law
- CPU clock speeds stopped rising around 2005 to 2008 because heat dissipation grows super-linearly with clock speed, so computing gains now come from parallelism and specialization, making throughput growth forecastable.Solana: Faster, Cheaper, More Scalable
- Extreme co-design is necessary because modern AI problems no longer fit inside one computer; to make 10,000 computers go a million times faster you must refactor algorithms, shard the pipeline, data, and model, which then forces optimization across CPU, GPU, networking, switching, power and cooling simultaneously rather than relying on Moore's Law.Lex Fridman Podcast #494 - Jensen Huang
- Phone chips have a one-to-two-year life cycle because foundries introduce new nodes roughly every two years and half-nodes yearly, so chip designers must redesign almost from scratch on a relentless treadmill.Cadence: Software Behind Semiconductor Design
- EUV machines were planned for the mid-2000s but did not come out until 2019, about 13 to 14 years late, even after Intel, Samsung, and TSMC co-invested in ASML in 2012, taking a 23% combined stake and funding EUR 1.4 billion of R&D.Five Handpicked Highlights
- Rising transistor counts, which were doubling in density roughly every two years under Moore's Law, made chip complexity grow too fast for vertically integrated firms to handle alone, pushing the industry toward disaggregation.Cadence: Software Behind Semiconductor Design
- Because our knowledge of the brain roughly doubles each decade and many published findings later turn out to be false, scientists should treat their conclusions not as discovered facts but as temporary assessments of how credible a current hypothesis is, to be given up without a fight if contradicted by evidence.134 | Robert Sapolsky on Why We Behave the Way We Do