You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
For power electronics that have to operate at far higher voltages with lower switching losses than Silicon, sure. But I'm betting on second (or third coming) of silicon wafer scale integration (a la Cerebras), chiplets & 3D packaging, plus possibly the second coming superconducting semiconductors as more of the data center future. Just saw this new semi company today !
The Next Chapter of Compute Begins Below Zero
Introducing the world’s first commercial Superconducting Computing Platform—redefining performance, power, and possibility at 4.5 Kelvin.
Discover Snowcap, the world’s first commercial superconducting computing platform. Unleashing breakthrough efficiency and power at 4.5 Kelvin, Snowcap is redefining what’s possible in AI and data center performance.
While certainly having some technology advantages over silicon, my understanding is SiliconCarbide has some physical properties, which make it way harder to process. This causes commercial disadvantages, limiting the overall use of SiC.
SiC will definitely remain a strong technology in future, but I do not expect it to replace Si.
Aside from voltage, modern silicon switches hava managed to close on SiC for most mainstream applications.
Hundred kHz switching, fantastically small RDSon, temperature and current ratings that only "exotic" switches were reaching 20-15 years ago, are now mainstream.
My opinion: an average type-c power supply today is designed by someone with "YouTube videos" level of knowledge about power electronics. A certain portion of them are designed by someone with "copy and pasted from Chinese FTP" level of knowledge.
Around 2010, big Western retail chains were some times unknowingly buying chargers designed as capacitive droppers, with shocking results.