Prerna Raghav 🌞

The British Government is a complicated beast. Dozens of departments, hundreds of public bodies, more corporations than one can count... Such is its complexity that there isn't an org chart for it. Well, there wasn't... Introducing ⚙️Machinery of Government⚙️

97% of the water on that map is saltwater. Data centers don't run on saltwater. The accessible freshwater humans can actually use, the rivers, lakes, and shallow aquifers that support 8 billion people, is 0.3% of Earth's total water supply. That tiny fraction is what data centers are pulling from. Google consumed 6.4 billion gallons across its data centers in 2023. Microsoft used 1.7 billion gallons, up 34% from the year before. Training GPT-4 alone consumed 13.4 million gallons in a single month at Microsoft's Iowa facility, equal to the monthly water usage of 130,000 Americans. Northern Virginia, the world's data center capital, used 2 billion gallons across its facilities in 2023, a 63% increase from 2019. Lawrence Berkeley National Lab projects U.S. data center water consumption will double or quadruple by 2028. The problem is where this water goes. Evaporative cooling doesn't return the water to the system. It's gone. And two-thirds of data centers built since 2022 are in regions already facing water stress. Arizona revoked new residential building permits in Maricopa County because groundwater was running out. Google's data center in the same county has a permit to draw 1.45 billion gallons per year. The state chose servers over homes. Posting a picture of the ocean to dismiss freshwater consumption is like pointing at the sun to argue a house fire isn't hot. Scale doesn't work when you're pulling from the wrong pool.

Huge W

Dario Amodei just raised $30 billion at a $380 billion valuation three weeks before going on camera to say his own technology terrifies him. Nobody is asking the obvious question: why does the CEO of the fastest-growing enterprise AI company on earth keep doing press tours about how dangerous his product is? Anthropic went from $1 billion in annualized revenue in December 2024 to $14 billion by February 2026. That is 14x growth in 14 months. Claude Code alone is doing $2.5 billion annualized. Eight of the Fortune 10 are paying customers. The company is preparing for an IPO that could be one of the largest tech listings in history. This is a man who has figured out that the fear IS the product. Every time Amodei goes on 60 Minutes and compares AI to a tsunami, three things happen simultaneously. Enterprise buyers hear "this technology is so powerful that even its creator thinks you need to take it seriously." Regulators hear "we're the responsible lab, write the rules around our safety standards." And investors hear "the market for AI safety infrastructure is so large that even the people building it can't see the ceiling." Anthropic's entire competitive moat is that it's the AI company that worries out loud. OpenAI plays offense, Google plays distribution, and Anthropic plays conscience. That positioning just attracted Sequoia, Microsoft, Nvidia, GIC, and Coatue into the same cap table. Sequoia is backing Anthropic AND OpenAI AND xAI simultaneously because the bet isn't on which lab wins. The bet is that the market is so large that backing all three still returns 10x. The post says "nobody voted for this." Correct. But $30 billion in new capital voted for it last month. The same man who published a 38-page warning about existential risk in January closed the second-largest private funding round in venture history in February. Those two events are not in tension. They are the same strategy. Watch what he does, not what he says. Anthropic committed $50 billion to US data centers. They signed a $30 billion compute deal with Microsoft Azure running on Nvidia chips. They tripled international hiring. They are sprinting, not braking. The "tsunami on the horizon" framing does real work. It makes the urgency feel external, like weather. Meanwhile, the company is growing at 14x annually because the product works and customers can't stop buying it. Amodei genuinely believes the technology is dangerous AND he's building it as fast as humanly possible AND that contradiction is the most bankable brand position in enterprise software right now. The part that should stop you cold: he said it, raised $30 billion the same month, and nobody found that strange.

Eric Schmidt's advice for making money in AI right now: "Found an agentic AI company. Not one designing agents, build an agent to do something."