Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

New SambaNova chip designed to handle 5 trillion parameter model | TechCrunch

Sep 19, 2023 - techcrunch.com
SambaNova, an AI solutions company, has introduced a new chip, the SN40L, which it claims will significantly reduce the cost of running large language models like OpenAI's ChatGPT. The company's CEO, Rodrigo Liang, stated that the new chips are 30 times more efficient, reducing the number of chips required to power these models. He claimed that while competitor chips would need 50-200 to run a trillion parameter model, SambaNova has reduced that to just eight.

SambaNova offers a full stack hardware and software solution for building AI applications, with the new SN40L chip being part of this offering. The company provides these solutions as a multi-year subscription, giving customers more cost certainty over their AI projects. The new chip is available starting today and is fully backward compatible with previous generation chips.

Key takeaways:

  • SambaNova has introduced a new chip, the SN40L, which is designed to significantly reduce the cost of running large language models, such as those used by OpenAI's ChatGPT.
  • The company's founder and CEO, Rodrigo Liang, claims that the new chips are 30x more efficient, reducing the number of chips required to power these models from 50-200 to just 8.
  • SambaNova provides a full stack hardware and software solution for building AI applications, with the company helping customers train models on their private data, which then remains under the customer's ownership.
  • The new SN40L chip is available starting today and is fully backward compatible with previous generation chips.
View Full Article

Comments (0)

Be the first to comment!