Battle of name and battle of AI solutions.

3 innovation and digital news in 1 minute. Every Monday. Episode 378

An accelerator for the AI chat answer — Groq!

NVIDIA reigns of the processor sphere but there are some niches where not. Groq is one of them. A processor system for the usage of AI solutions, not the training. Big chips are made for training (inference = high computation), the Groq processor concentrate on the bare usage of the LLM. 10 times faster. Opinion: Very low latency, instantly — this is how an AI chat answer on Groq. The company has an own chat, Groq Chat as a demo. Based on open-source models like Mistral. Amazing fast answers. Never seen before.

Remember Groq, they are scaling production — fast!

Another Grok…

Grok is the name chosen by Elon Musk for the Large Languae model developped under the new X.ai umbrella. A new version of Grok has recently been launched, Grok 1.5. It seriously competes with the largest models like GPT, Claude or Gemini. And brand new, V — Grok V for vision. Opinion: It refers to a 1961 Robert Heinlein’s sci-fi novel where Grok is the Martian language. A cunning reference, as always with Elon Musk. And yes, we gathered that they want to gather 3 billion USD to grow the LLM (at the level of Anthropic’s valuation).

Never boring with Elon Musk!

Gaudi 3 is the name of the AI-chip competitor, from Intel

Intel announces its 3rd version of an enterprise AI chip called Gaudi 3. It competes with the market reference the high-powered NVIDIA H100. A so-called AI accelerator. This is how Intel position it: “delivers 4x AI compute for BF16, 1.5x increase in memory bandwidth, 2x networking bandwidth for massive system scale”. Opinion: This showcases the race in the AI chip business, for training above all and always in reference to the champion, NVIDIA. Intel is making a point here. It is worth apost. And since we are in namimg, Gaudi, it inspires as well.

What’s your immediate association with Gaudi?