IMO, OpenAI will be the next Google and possibly surpass it. The amount of new things that I see here in a week is nearly equal to what I saw in a year when I was at Google.

Elon Musk at Dubai's World Governments Summit (February 13, 2025): "Grok 3 has very, very powerful reasoning capabilities. In the tests that we've done thus far, Grok 3 is outperforming anything that's been released that we're aware of." "In fact, at times I think Grok 3 is kind of scary smart. And you're like, wow, this thing's smart. It's kind of scary. Grok 3 is scary. It's like, wow, this thing's, you know, comes up with solutions that you didn't even think were lucky. You wouldn't even anticipate, you know, not obvious solutions." "Grok 3 was trained with the most amount of compute and I think very efficiently trained. Also notably Grok 3 was trained on a lot of synthetic data." "It goes back and forth through the data and tries to achieve logical consistency. So when, if it's got data that is, wrong, it'll actually reflect upon that and remove the data that is, that is wrong." "We're really in the final stages of polishing Grok 3. Probably it gets released in about a week or two. So pretty, pretty soon." "And I think this might be, we think it'll be better than anything else. And then maybe this might be the last time that any AI is better than Grok."

"Probably his whole life is from a position of insecurity" OpenAI CEO Sam Altman said the company is "not for sale" and Elon Musk is probably trying to delay its progress with an unsolicited bid for the AI firm

if my ai company was going well i would not be trying to buy another one for one hundred billion dollars

Sam Altman was very charming and has a very bold vision that he has shared with me & other senior TSMC staff ! The chip order is of substantial size and we are on 3 nm due to accelerated timeline of the Stargate build out. We are building more 2 nm capacity to support the next phases of this project. I believe that OpenAI will do marvelous work and produce one of best chips in history! I have a feeling more will come, wait and see 😜

STAR GATE WILL GIVES US AGI. This is a fact… let me explain: Sam meeting with TSMC executives last year was to inform us of evidence of model scaling with new technique that can generalize beyond data and learn in new environments. TSMC was informed in order to build out capacity sooner for what would become the stargate project with assurance from US government that capacity would be purchased. $7 trillion was asked for with half a trillion allocation to start in funding… but where to get this funding ?

>company named Boom >look inside >no boom

Using a custom RDMA-based networking library, we've been able to achieve 3200 Gbps GPU memory transfers, bypassing NCCL limits for 97.1% theoretical bandwidth efficiency. Our latest blog shares our journey of building a custom high-performance networking solution on AWS.

no thank you but we will buy twitter for $9.74 billion if you want

Visual chain of thought. Looks like we get „thinking“- vision models

Periodic cooking of eggs (alternating 100C and 30C) allows a sugary yolk while totally firming the whites thanks to the different thermal conductivities of the albumin/yolk

DEEP RESEARCH WAS TRAINED WITH RL. YOUR REPLICATION IS NOT. DON'T CALL IT AN OPEN VERSION UNLESS YOU ALSO USE RL.

Same experience here. The llama based models are less capable in general somehow.