Skip to main content

Posts

Hitting and Breaking Ceilings and the Slope

What a 23% kernel speedup tells us about whether AI can improve itself and how to quantify the gains. In 2025, Google DeepMind’s AlphaEvolve, an AI system that discovers algorithms, found a way to make a FlashAttention kernel run 23% faster. This was a great result but it was not the beginning of a runaway ... Read more
Post a Comment
Read more
Recent posts

Understanding Different AI Data Center Leasing Prices and Categories

The high $50 Billion per GW lease rates for xAI/SpaceX Colossus and similar premium, immediately available AI clusters are for scarce of ready-to-use, large-scale coherent GPU capacity. Kevin Pathrath’s is wrong about a conspiracy. He is comparing against different types of data center. The $2.7M/MW ($2.7 billion per gigwatt build figure is shell-only and brownfield. ... Read more
Post a Comment
Read more

Self Improving Harness – Self-improving coding and research agent

Prime Agent is an open-source self-improving coding and research agent (harness) released by Prime Intellect on August 5, 2026. It is built for coding workflows and long-running autonomous tasks. It combines a Recursive Language Model (RLM) runtime with a Continual Harness, emphasizing token efficiency and expressiveness through programmatic tool calling, treating context as a variable, ... Read more
Post a Comment
Read more

Are Top AI Talent Leaving Google to Apply AGI and ASI for Giant Leaps in Science

Demis Hassabis is changing to a new role as Chair of Google DeepMind & Chief Scientist of Alphabet. This will allow him to focus on long-term strategy, and accelerating scientific breakthroughs, including leaning into my work at Isomorphic to help cure disease. koraykv will be stepping up to lead Google DeepMind as SVP, alongside @joshwoodward ... Read more
Post a Comment
Read more

LIVESTREAM – After Predicting the SpaceX Q2 Earnings- What Happens in the Next 18 Critical Months

I Predicted SpaceX Q2 Earning $100B ARR, 2+ GW 2026- NEXT 18 Critical Months. A 40% spot uplift + cheaper power/shell + wrappers can pull the crossover forward by a quarter or more versus a pure long-term $35–50B locked-rate case. Risks remain execution on power reliability/permitting, utilization, and any broader rate compression, but the combination ... Read more
Post a Comment
Read more

SpaceX Partners with Starmind AI1 Satellites for Rubin GPUs

SpaceX is partnering with @Nvidia to design the Starmind AI1 satellite compute payload. Each of the Starmind satellites will include NVIDIA Rubin GPUs and Vera CPUs for datacenter class space compute SpaceX is partnering with @Nvidia to design the Starmind AI1 satellite compute payload. Each of the Starmind satellites will include NVIDIA Rubin GPUs and ... Read more
Post a Comment
Read more