DatologyAI generated 12 trillion synthetic tokens for pre-training across web, math, and code domains.
“we have hit a "data wall." You need to spend exponentially more computing power and data to get models that only get linearly better.”
scaling
19 tracked signals on scaling.
Clay runs over 350 million go-to-market AI agents monthly, processing trillions of tokens per week.
“We run this over 350 million times a month. It processes trillions of tokens every week.”
Coding agents grew 1400% in 2026, overwhelming GitHub's infrastructure built for human-speed development.
Bitter Lesson means scale computation via search and learning, not human knowledge injection
“don't be distracted by human knowledge as AI traditionally has been many times. Instead focus on learning methods that will scale with computation like search and like learning.”
OpenAI CFO frames full-stack chip-to-product compounding as path to cheaper, scalable intelligence
Chai Discovery is treating drug design as a scaling problem, prioritizing model simplicity over complexity
“one of the exciting things we're trying to do at Chai is to make the drug discovery process like a little bit more like engineering”
NVIDIA GB300 NVL72 sets world record pre-training DeepSeek-V3 671B at 1,648 TFLOPs per GPU
“As compute per token falls, communication increasingly determines how efficiently models scale across thousands of GPUs.”
Frontier lab xAI may be running at sub-10% Model FLOPs Utilization, far below best-in-class 60-70%.
“The AI scaling debate always focuses on the question of "how do we get more GPUs?" but the better question may be: how do we make the most of ones we already have.”
Jeff Dean argues there is still plenty of training data left despite fears LLMs are running out.
“it's true we've like used quite a lot of of the public text data in the world”
Scaling agentic applications requires governance to prevent uncontrolled AI sprawl
“Building an agent is getting easier.”
AI models and biological brains share geometric structure, suggesting intelligence is a physical law
“if you apply enough compute to matter, you get this thing that looks like intelligence”
Hugging Face now hosts 3 million public models, a 150x increase in a few years.
“More than 30% of Fortune 500 use hugging face as a part of AI workflows.”
Specializing a coding model doesn't violate the bitter lesson; it scales data to saturate finite model capacity.
“we need to free up the weights from distractions the model may have”
Enterprise CDOs and Chief Scientists share frameworks for scaling AI agents organization-wide
General Reasoning is building RL systems to scale agents toward long-horizon tasks
Ramp's Forward Deployed Engineering team scales enterprise agentic features using two principles: always be scoping and scale with tokens.
“Always be scoping and scale with tokens.”
Companies lose their fearless engineering culture and stop taking risks as they scale to thousands of employees.
“we would much rather fail in pursuit of the extraordinary than succeed in the ordinary”
Hugging Face claims knowledge distillation can now be run cheaply at scale
PostgreSQL performance on large tables degrades silently through bloat, vacuum storms, and WAL explosion without proactive maintenance.
“performance doesn't break suddenly, it degrades silently over time”