OpenAI's first custom silicon chips will be manufactured by TSMC, utilizing the A16 process node, which promises significant speed and power improvements compared to existing technologies.
The A16 process offers a speed improvement of 8-10% and a power reduction of 15-20% at the same speed, alongside improved chip density compared to TSMC's N2P.
Apple emerged as the first major customer to secure A16 production capacity, in alignment with its plans to integrate OpenAI's ChatGPT into its software ecosystem.
OpenAI has allegedly explored investing in its own semiconductor fabrication facilities, though current ambitions seem refocused on developing machine-learning ASICs in collaboration with Broadcom and Marvell.
Collection
[
|
...
]