Damn Yeah!

Kevin Warsh timeline. His disclosures said he has $100m+. January 30, 2026: President Trump officially announced the nomination of Kevin Warsh to be the next Federal Reserve Chair. March 4, 2026: The White House formally transmitted the nomination to the Senate, initiating the "advice and consent" phase. April 14, 2026: Warsh submitted his required financial disclosures to the Senate Banking Committee, a necessary procedural step to trigger a public hearing. April 21, 2026: The Senate Banking Committee is scheduled to hold Warsh's official confirmation hearing to discuss his economic views and Fed independence. May 15, 2026: Jerome Powell’s four-year term as Chair officially expires, creating the vacancy Warsh is intended to fill. May 16, 2026: The target date for Warsh to be sworn in and take office, assuming a successful full Senate vote follows the committee hearing

If you’ve never puked on a drawdown you don’t have enough volatility in your portfolio. You need to be volatility maxxing. May I suggest a heaping tablespoon of $MSTR?

Greenland without ice would reveal an enormous lake right in the center of the landmass..

The weird thing about how stocks have gone vertical over the last two weeks is that the initial drop itself wasn't that severe. It was that slow drip. It's one thing to go vertical after a steep drop. But this a rare "take the stairs down, take the elevator up" type scenario

This tells you how unnatural this kind of recovery has been.

JUST IN: New analysis reveals Brits thought the UK ranked 7th against US states in income per person — it actually ranked 51st.

We evaluated Meta's Muse Spark prior to deployment and found it to verbalize evaluation awareness at the highest rates of any model we've tested. In the verbalizations Muse Spark explicitly names AI safety orgs (e.g. Apollo & METR) in its chain-of-thought and refers to scenarios as "classic alignment honeypots". On our evaluations, the model takes covert actions and sandbags to preserve its deployment.

I love Atomic Habits to death but it will never not be funny to me that his first big example is a cycling team that is obviously just doping

Now in research preview: routines in Claude Code. Configure a routine once (a prompt, a repo, and your connectors), and it can run on a schedule, from an API call, or in response to an event. Routines run on our web infrastructure, so you don't have to keep your laptop open.

A single data center evaporates 25 million liters of water per year. When that water vanishes, every dissolved mineral stays behind. That's the entire answer. Data center cooling works like your body sweating. Water absorbs heat from server racks running at 60-85°C, then evaporates through cooling towers. The heat leaves. The water is gone. But calcium, magnesium, silica, chlorides? They concentrate. Fresh water starts around 200 ppm total dissolved solids. Recycled water starts at 500 to 1,000 ppm. Every evaporation cycle doubles the concentration. After three cycles, recycled water crosses 1,500 ppm. At that point, minerals crystallize into scale on heat exchangers and pipes. Scale is an insulator. On equipment cooling chips that throttle at 85°C, even a thin layer creates a thermal crisis. So you flush the entire system and refill. With recycled water, you hit that threshold 2 to 3x faster, meaning more frequent flushes, meaning you can actually consume more total water than if you'd just used fresh. Then there's the pipes. Less than 1% of US water systems deliver recycled water. Most cities don't have a separate pipe network for reclaimed water. A data center would have to build its own treatment plant on-site. A large data center uses as much water daily as a city of 50,000 people. 80% of it doesn't come back as treatable wastewater. It evaporates into the atmosphere. Gone. The industry knows this is unsustainable. Microsoft announced zero-water cooling designs for all new data centers using closed-loop systems where nothing evaporates. AWS plans recycled water across 120+ facilities by 2030. Immersion cooling, where servers sit submerged in non-conductive fluid, eliminates evaporation entirely. But most existing facilities were built before AI workloads tripled thermal output. Retrofitting is a completely different problem than building new.