James Wang
@draecomino
Director of Product Marketing @CerebrasSystems Prev: Nvidia, ARK Invest, 21Shares
First success, then happiness.
"Happiness is being satisfied with what you have. Success comes from dis-satisfaction. Choose." @naval
After more than a year of getting burned with MoE gotchas, I finally sat down and wrote the guide I wish existed. Every paper skips the messy production details. This fills those gaps. No theory without implementation. cerebras.ai/moe-guide
Let's talk about MoE: 🔶 How many experts should you use? 🔶 How does dynamic routing actually behave in production? 🔶 How do you debug a model that won’t train? 🔶 What does 8x7B actually mean for memory and compute? 🔶 What hardware optimizations matter for sparse models?…
grok looked at 20 webpages and 41 elon posts to arrive at this answer. 'maximally truth seeking AI' 💀
$ARKK setting up for the highest weekly close since Feb 2022. Cathie redemption arc in full swing?
NY coffee shops are full of people using products from SF. No one in SF coffee shops uses anything from NY.
Implication: an NYC waiter will make more than a front-end dev in 2030.
Ben Todd has written the best thing on how to plan your career given AI/AGI. Will thread. A very plausible scenario is salaries for the right work go up 10x over a ~decade, before then falling to 0. So we might be heading for a brief golden age followed by crazy upheaval. 1/
"Invert, always invert" is a lame mental model. You can never invert your way to an iPhone, a Starship, or an AGI.
A single ChatGPT query uses about 3% of an iPhone’s battery in power and six drops of water in cooling. Source: Sam Altman
I'm looking forward to playing Skyrim where you can chat to an NPC for 3 hours and then they invite you home for a pot roast.
This dude create an AI video “interviewing” people from the 1500s and it’s hilarious 😂
These kind of charts are so silly – the base rate of smartphone penetration is totally different for each of these apps at launch.
NVDA a $10T stock under default conditions. NVDA a $100T stock if mania hits.
Nvidia smoked all the competition with Blackwell. Everyone except Cerebras.
Cerebras just beat NVIDIA Blackwell Last week: Blackwell hit 1,000 t/s on Llama 4. Today: Cerebras hit 2,500 t/s on the same model, same benchmarks by @ArtificialAnlys Blackwell smoked Groq, AMD, Google – everyone. Only Cerebras stands – and we smoked Blackwell.
I'm looking to hire a few growth engineer interns this summer. > No age requirements > Must be technical > In-person, relocation $$ covered > Unlimited Cerebras ramen > You'll work directly with me
Tim Sweeney is the bravest person in the tech industry in the past two decades and it's not even close.
many thought that Epic wouldn't achieve anything in its fight against Apple, but Fortnite is back on the iPhone in the US and Apple has been forced to not take fees from purchases made outside of apps. That seems like a win for all developers to me theverge.com/news/661290/fo…
Cerebras launched inference just 8 months ago. Today it is officially part of Llama API. Any developer can now click a button and get a wafer-scale chip to generate tokens at ~2,600 t/s. Insane progress.
ChatGPT is the greatest democratization of knowledge event since Google. And given how profound Google has been, there's just no way to over-hype AI.