Rokko

Jensen Huang, Nvidia CEO, on the future of graphics: "AI is now going to go back and revolutionize how computer graphics is done all together." He spent the whole keynote saying AI will generate everything. But watch the detail: even his showcase, neural rendering, starts from what he called "the ground truth, the structured data of virtual worlds." Human-built worlds first, AI on top. Same month, GTA 6 shipped. $1.5 billion, 8 years, zero generative AI, and it will be the most valuable entertainment ever made. AI didn't replace the craft. It amplified it. The ground truth is still built by humans.

Elon Musk said AI will soon build a game the size of GTA 6. Strauss Zelnick, the Take-Two CEO who actually spent $1.5 billion building it, called that "laughable." Then the knockout: "If AI were going to take anyone's job, why not the CEO's first?" GTA 6: 8 years, over 1,000 people, $1.5 billion, and zero generative AI. Pre-orders sold out on Amazon in about an hour. The richest studio on earth had an unlimited budget and every AI model on the planet. It chose humans. That's the tell.

How to lose $20 billion in two days. Bill Hwang turned $200 million into $20 billion, then vaporized all of it in a week. His own words: "I am just following God's word, and that is truly a fearless way to invest." No hedges. No doubt. When the margin calls came, he refused to sell. $30 billion wiped out, banks included. The market doesn't punish the dumb. It punishes the most certain. The edge was never conviction. It's doubt.

@shevaxgod 💰💰💰

@0xRokko 20b in 2 days😳😳😳

Marcos Lopez de Prado, Cornell professor, author of Advances in Financial Machine Learning, and the quant who ran up to $13B, just broke down the 7 ways machine learning lies to you in markets: 15:26 – how returns quietly erase the memory in price 30:33 – the labeling mistake every academic makes 45:33 – data leakage: why backtests die live 48:33 – backtest overfitting: any Sharpe is yours with enough trials 49:50 – a strategy on pure noise still expects a Sharpe of 3 This 1-hour talk will replace 100 articles on why your backtest is lying. Watch it today, then read how to tell a real edge from luck in the article below.

This is every named chess opening ever studied, ~3,700 of them, rendered as one force-directed graph. A map of where human theory has actually been. Your AI's memory should look exactly like this. Not a giant transcript of every session, but a graph of compressed lessons, each one linked, each one recalled only when it's relevant. That's why a graph beats a log: a log gets noisier as it grows, a graph gets smarter, because every new node connects into the web and makes the rest more useful. The model was never the moat. The graph of what it has learned is.

@caspr_exe he is young genius

Want to know how people earn $500k a year writing code? It starts exactly here. While everyone argues AI killed coding, this kid is silently grinding LeetCode in Python. 334 problems solved by hand, no AI, no talking. Data structures and algorithms, the exact thing he's drilling, is the gate every $500k engineer at Jane Street, Citadel and Google had to pass. He's just starting a decade early. The fundamentals aren't dead. Someone's quietly compounding them while the rest of us debate.

@cyrilXBT real important to know !!

A real-time globe that fuses 30+ free live feeds into one picture: war front lines, earthquakes, wildfires, markets, refugee flows, flights, satellites. A local AI reasons over all of it and forecasts what's likely next, from 24 hours out to a year, with a council of four personas (Strategist, Economist, Naturalist, Skeptic) that show where they agree and where they split. Runs entirely on one machine through Ollama. No cloud, no API keys, no cost. One local call returns the whole planet's live state, so any agent gets eyes on the world in a single request. And nobody's selling it. Open source, MIT, free. The model was never the magic. The grounding layer underneath it is.

The algorithm in this article helped build a $31 billion fortune. It's called Baum-Welch, and this thread teaches it to detect market regimes. The "Baum" is Leonard Baum, who used the exact same hidden Markov math to help build Renaissance's Medallion fund: about 66% a year for 30 years, the best track record markets have ever seen. It is the same algorithm that taught computers to recognize human speech. Pointed at price instead of sound. The math is 60 years old and public. The moat was never the equation. It was the nerve to trust it.

@caspr_exe own the track, not the runners.

NVIDIA made $215.9 billion last year without shipping a single AI model. It sells the compute every lab runs on, and $193.7 billion of that came from data centers alone, roughly 90% of the entire company. Blackwell is sold out. $500 billion in orders already booked through the end of 2026. Market cap around $4.9 trillion, the most valuable company on earth. Models come and go. The moat is the silicon they all rent. Own the compute, and you own the floor everyone else builds on.

This is a Rubik's Cube solver, built in minutes just by talking to Claude. Then it kept going. It broke on mobile, so he typed one lazy prompt, hit /yolo, and it fixed itself. Next he's adding webcam solving and 4x4, 5x5 cubes. Shipping a real, working tool used to take a weekend. Now it takes a coffee. The barrier didn't shrink, it collapsed.