Nvidia to design tailor-made AI chips for cloud computing firms

Key Points:

  • Nvidia creating a dedicated unit for designing custom chips for cloud computing and other sectors
  • Nvidia’s initiative aims to tap into the growing market for tailor-made AI chips
  • Demand for custom chips in data centers projected to reach $10 billion in 2022 and expected to double by 2025

Summary:

Sources whisper that Nvidia is sprouting a bespoke chip-making division aimed to cater to cloud computing mavericks and beyond, in a bid to ward off approaching challengers. Currently commanding an 80% market share of the high-end AI chip market, Nvidia’s coffers have swelled by a staggering 40% this year to $1.73 trillion, following a meteoric rise in 2023.

 

While OpenAI, Microsoft, Alphabet, and Meta have been loyal patrons of Nvidia’s H100 and A100 chips, they are now peeking over their shoulders, contemplating the crafting of their own chips to economize energy and expenditures. Nvidia aims to aid these industry stalwarts in forging their own custom AI chips, potentially siphoning business from Broadcom and Marvell Technology. Headed by Dina McKinney, originally from Advanced Micro Devices and Marvell, Nvidia’s custom chip battalion is set to democratize its technology across cloud domains, 5G, video gaming, and automotive.

 

It is predicted that the hunger for custom chips will burgeon to $10 billion this year and is foreseen to double by 2025. Custom chips are valued to be an estimated $30 billion in 2023, constituting about 5% of annual global chip sales.

LLM BENCHMARK LEADERS
FINANCIAL INFORMATION

DAILY LINKS TO YOUR INBOX

PROMPT ENGINEERING

Prompt Engineering Guides

ShareGPT

 

©2023 The Horizon