process node
A generation of semiconductor manufacturing technology, typically named by a nanometer figure (e.g., 7nm, 5nm), denoting transistor density and capability.
As defined in @Asianometry & Dylan Patel — How the semiconductor industry actually works
Other definitions
A named generation of semiconductor manufacturing technology, such as 7 nanometer or 10 nanometer, with smaller numbers generally meaning denser chips.
AMD: How Chips Are Changing
Where it comes up · 11
- Applied Materials: Sanding Atoms6 claims
- @Asianometry & Dylan Patel — How the semiconductor industry actually works5 claims
- Tech and Consumer Growth Investing3 claims
- AMD: How Chips Are Changing2 claims
- ASML: Competing with Moore’s Law2 claims
- Dwarkesh Patel Podcast - Dylan Patel — Deep dive on the 3 big bottlenecks to scaling AI compute2 claims
- Neil Movva - Making AI 10x Cheaper - [Invest Like the Best, EP.488]2 claims
- Inside the Trillion-Dollar AI Buildout1 claim
and 3 more
Most important claims · 10 of 26Search all claims for “process node” →
- Chipmaking involves thousands of sequential steps across hundreds of equipment types, so defect rates must be far better than Six Sigma, because a Six Sigma defect rate applied across tens of thousands of steps would make functional chips impossible.Applied Materials: Sanding Atoms
- Process node names such as 5-nanometer and 3-nanometer no longer correspond to the physical size of any feature on the chip and are largely marketing terms, though the industry still keeps squeezing more and smaller transistors onto silicon.ASML: Competing with Moore’s Law
- Performance per watt for bfloat16 multiply has not improved dramatically from Hopper to Blackwell to Rubin, and moving from TSMC 5nm to 4, 3, and 2nm has not changed performance per watt dramatically, so the bottleneck is not process node gains.Neil Movva - Making AI 10x Cheaper - [Invest Like the Best, EP.488]
- Shifts to gate-all-around and nanowire transistor shapes at 3, 2 and 1.4 nanometers create brand-new process steps that Applied Materials can compete to enter and co-engineer.Applied Materials: Sanding Atoms
- Intel lost its manufacturing lead in 2018 because TSMC's 7-nanometer node is equivalent to Intel's 10-nanometer node and TSMC got there first; since TSMC supplies most fabless chipmakers (including Apple's chips), this creates a roughly 20-30% competitiveness swing.Tech and Consumer Growth Investing
- Moving from 5-nanometer to 3-nanometer adds about 30 percent more process steps, and AI model complexity is why customers still want shrinks despite weakness in mobile and PC demand.Applied Materials: Sanding Atoms
- Around 2015 to 2016 Intel fell off the Moore's Law curve and is stuck at roughly 10 nanometer, about two process nodes or three to four years behind TSMC, which let AMD working with TSMC capture the full benefit of Moore's Law.AMD: How Chips Are Changing
- After a long period of acquisitions that built its portfolio, Applied Materials was blocked from merging with Tokyo Electron and then shifted from M&A toward organic growth, focusing on new equipment for future process nodes.Applied Materials: Sanding Atoms
- Semiconductor fabs run large R&D operations that produce zero direct economic value because they are learning the next process node, but that learning eventually enables volume production that makes the money.Inside the Trillion-Dollar AI Buildout
- If the West lost access to TSMC, supply would take a shock, but the performance gap to the best Western alternatives such as Intel is only about 2x worse performance per watt at worst, far smaller than chip-war discourse suggests.Neil Movva - Making AI 10x Cheaper - [Invest Like the Best, EP.488]