Big thanks to Marc Sun and team who've been some of the fastest and most efficient code reviewers I've had the pleasure of working with There's a lot more we're planning next for GPU poors looking to make everything smaller: weights, gradients, optimizers and soon activations!

122 days is FAST

Even Ayn Rand would be amazed how the government is slowing down our most successful inventor - FAA delaying Starship launches - FCC refusing to work with Starlink, first claiming its incapable, then claiming its too good - President not commenting on the first private spacewalk

Overheard in Silicon Valley: "The transition from the Jobs Way to the Musk Way in US startup culture is fascinating to watch. The founders are switching from artistic UI hippies to weightlifter hard executors."

Serious: this AI thing truly is going to be a defining moment of humankind. Being part of it in any way is exhilarating.

Message from a friend: "I'm torn between not wanting to hire anyone who cheats in a coding interview and not wanting to hire anyone who doesn't cheat in a coding interview."

Excited for the 5090! What do you think? CES or Christmas?

Throw a lake in there because, why not.

Mechazilla has caught the Super Heavy booster!

What the SpaceX team grasped from the early days that basically no one else understood is that if price to launch changes by orders of magnitude entire markets are opened up. Starlink is just one example; it wasn’t even an idea when SpaceX was founded.

Quite extraordinary. Space is the new air. The rapid success with this launch system changes how we should think about space flight in the future.

The ROI of glory is infinite

What will a world look like with 100 billion digital human beings? Today we share our tech report on Project Sid – a glimpse at the first AI agent civilization (powered by our new PIANO architecture). github.com/altera-al/proj… 1/8

Now that we have seen what Full o1 can do, there is no one who can deny that we have not entered a new era. Especially with that 200k context window. Sam Altman is right: we are in the age of intelligence. o1-preview was already able to solve PhD tasks. Full o1 goes one better.

Starship hot stage separation

Recall President Reagan mourning the crew of the Challenger, honoring their legacy, and resolving to continue America’s space program Here is a leader after a tragedy; this is the bar

Gwern comments on "Implications of the inference scaling paradigm for AI safety": lesswrong.com/posts/HiTjDZyW…

Alibaba surged 8.4% after unveiling its powerful new open-source AI model, which outperformed DeepSeek's R1 with less data. finance.yahoo.com/news/alibaba-s…

The honeypot is a chef’s kiss. Well done.