Cerebras

[Matt]

Been impressed with these guys after doing a PoC with Total. Buy thru hive

https://cerebras.ai/press-release/totalenergies-research-technology-usa-selects-cerebras-systems-cs-2-to-accelerate-multi-energy-research/

https://www.barchart.com/story/news/28443972/new-ai-chip-beats-nvidia-amd-and-intel-by-a-mile-with-20x-faster-speeds-and-over-4-trillion-transistors?utm_source=rss&utm_medium=feed&utm_campaign=nasdaq

Can't wait to talk with them again at SC.

[Matt]

IPO mv'd to October.

[Landlocked]

From what I’ve read SLB is looking to get into this stuff too.

[Matt]

O and g is massively invested in all of what we are talking about, especially the nvda thread.

[Matt]

From what I’ve read SLB is looking to get into this stuff too.

Petrel, Eclipse and the other SLB software suites...visit etc, TOTAL's Sismage... all about using AI now too.

[Matt]

Put the whole model on the wafer.

https://youtu.be/8i1_Ru5siXc?si=5-IJvhhN4Q7JE8Yc

[Matt]

Interesting info on the blackwell issues re interposer, bridges and heat expansion issues

https://youtu.be/7GV_OdqzmIU?si=7bZeyLXSHoOSpqCZ

[Matt]

https://www.nextplatform.com/2024/10/25/cerebras-trains-llama-models-to-leap-over-gpus/

Still no ipo date

[Matt]

Waiting on deck

https://www.forbes.com/sites/johnwerner/2024/11/04/nvidia-cerebras-race-to-supply-big-chips-for-ai-inference-activities/

[98aggie77566]

Are you hearing much on customer use cases with Cerebras?

Ive read of the one whale in EMEA…but very little beyond that.

Still haven’t seen whether they built a chip that solved one big customer’s specific challenge…or a whole lotta customers?

[Matt]

It's a unique option. I will speak with them (next) Monday

[Matt]

https://www.forbes.com/sites/karlfreund/2024/11/18/cerebras-now-the-fastest-llm-inference-processor--its-not-even-close/

Saw them today, good stuff. Watching

[Matt]

https://www.businesswire.com/news/home/20250130028856/en/Cerebras-Launches-Worlds-Fastest-DeepSeek-R1-Distill-Llama-70B-Inference

[Matt]

https://venturebeat.com/ai/cerebras-becomes-the-worlds-fastest-host-for-deepseek-r1-outpacing-nvidia-gpus-by-57x/