The 13 Semiconductors companies that produced the most founders (2026)

A founder factory is a company whose people leave to start companies of their own. This ranks every company in the graph by how many of its alumni — people who founded it or worked there — went on to found at least one company, counting only ventures started no earlier than the company itself.

Refreshed 2026-09-13 · updates nightly
Founder factories · SemiconductorsAlumni who went on to found companiesAMD: 5 alumni founded 7 companiesAMD5SambaNova Systems: 3 alumni founded 8 companiesSambaNova Systems3Marvell Technology: 3 alumni founded 6 companiesMarvell Technology3Intel: 3 alumni founded 4 companiesIntel3SeaMicro: 3 alumni founded 2 companiesSeaMicro3Ayar Labs: 2 alumni founded 2 companiesAyar Labs2Cognichip: 2 alumni founded 2 companiesCognichip2Nervana Systems: 2 alumni founded 2 companiesNervana Systems2Rivos: 2 alumni founded 2 companiesRivos2Graphcore: 2 alumni founded 1 companiesGraphcore2Icera: 2 alumni founded 1 companiesIcera2QuantumDiamonds: 2 alumni founded 1 companiesQuantumDiamonds2Rain AI: 2 alumni founded 1 companiesRain AI2
# Company SectorHQFounded Founders Companies started
1 AMD Semiconductors Santa Clara, USA 1969 5 7
2 SambaNova Systems Semiconductors Palo Alto, USA 2017 3 8
3 Marvell Technology Semiconductors United States 1995 3 6
4 Intel Semiconductors Santa Clara, USA 1968 3 4
5 SeaMicro Semiconductors 3 2
6 Ayar Labs Semiconductors 2 2
7 Cognichip Semiconductors 2 2
8 Nervana Systems Semiconductors 2 2
9 Rivos Semiconductors 2 2
10 Graphcore Semiconductors Bristol, UK 2016 2 1
11 Icera Semiconductors 2 1
12 QuantumDiamonds Semiconductors 2 1
13 Rain AI Semiconductors 2 1
How this is counted. A person counts as an alumnus of a company if they founded it or worked there. They count towards the company’s score if they went on to found at least one other company that was started no earlier than the company itself. Universities and other non-commercial organisations are pruned from the graph where we can identify them. Data is community-corrected — fix something.