😳 Sam Altman's receipts just dropped

🔍 The New Yorker vs. Sam Altman: 100+ Interviews, Zero Chill

The New Yorker dropped a sweeping investigation based on 100+ interviews, unseen memos from ex-chief scientist Ilya Sutskever, and private notes from Dario Amodei (Anthropic's CEO). Sutskever's memos allege Altman misrepresented safety protocols to the board. Amodei's notes independently reach the same conclusion: "The problem with OpenAI is Sam himself." A Microsoft exec even floated a "small but real chance" Altman could be "remembered as a Bernie Madoff, Sam Bankman-Fried-level scammer."
Why it matters: No single smoking gun - but when your former chief scientist, the CEO of your biggest competitor, AND a Microsoft exec are all independently saying the same thing... that's not a coincidence, that's a pattern. This level of corroboration from the people who literally helped build OpenAI is unprecedented. Sir, the calls are coming from inside the house. (source)

📜 OpenAI Publishes a 13-Page Plan for When Its Own Tech Breaks Everything

OpenAI released a detailed policy document saying we're "beginning a transition toward superintelligence" and calling for a new "social contract." Proposals include taxing AI-driven profits, creating a sovereign wealth fund that pays dividends to every American, a 4-day workweek, taxes on robot labor, a "Right to AI" access for all citizens, and - my personal favorite - containment playbooks for rogue autonomous AI. Axios called it "the most detailed blueprint any tech titan has ever published."
Why it matters: The CEO of an $852 BILLION company is essentially asking the government to prepare for a future where his own technology breaks the economic system. You don't publish a 13-page "please regulate us before it's too late" document unless you genuinely believe it's coming. The slow gears of government may not move fast enough. No because WHY is the company building superintelligence also the one writing the emergency manual?? (source)

💰 Samsung's Profit Just Went 8x and Nobody's Ready for That Number

Samsung projected Q1 operating profit of 57.2 trillion won (~$37.9B) - an eightfold increase from a year ago and nearly triple its previous record. DRAM chip prices nearly doubled in Q1 as supply couldn't keep up with AI data center demand. Samsung is also closing the gap on SK Hynix in high bandwidth memory after shipping HBM4 chips to Nvidia in February.
Why it matters: This. Is. Huge. Samsung's earnings show how deeply the AI infrastructure buildout is reshaping the entire semiconductor industry. But headwinds are forming - rising energy costs from the Middle East conflict and Google's new memory-saving TurboQuant tech could cool the rally. For now though?? An eightfold profit jump is absolutely unhinged. (source)

🤝 Broadcom Locks In Google's Custom Chip Business Through 2031

Broadcom signed a multi-year deal with Google to develop and supply future generations of custom AI chips (TPUs) and components for next-gen AI racks. Separately, Broadcom also inked a deal giving Anthropic access to ~3.5 gigawatts of AI computing capacity via Google's processors starting in 2027. Anthropic's run-rate revenue has exploded to over $30 billion, up from ~$9B at the end of 2025, and the startup says this builds on its $50 billion commitment to U.S. computing infrastructure.
Why it matters: The custom chip movement is accelerating as companies look for alternatives to Nvidia's pricey GPUs. Google's TPUs are becoming a serious growth engine, and Anthropic going from $9B to $30B in run-rate revenue is the kind of growth that makes VCs cry tears of joy. The "Nvidia or nothing" era might actually be ending... (source)

⚠️ Nvidia Now Controls the Software That Runs 60% of the World's Supercomputers

Nvidia's December acquisition of SchedMD gives it control of Slurm, open-source software that schedules computing tasks on ~60% of the world's supercomputers - including systems used for AI training, weather forecasting, and nuclear weapons development. AI specialists fear Nvidia could subtly favor its own chips by writing software updates for its hardware before rivals like AMD and Intel. Critics point to Nvidia's 2022 acquisition of Bright Computing, which they say was optimized for Nvidia hardware after the deal.
Why it matters: This is lowkey terrifying. When the world's most valuable company controls the scheduling software for most of the planet's supercomputers, even small advantages could tilt the competitive landscape for years. It's giving monopoly vibes and it's a litmus test for Nvidia's commitment to open-source neutrality. Spoiler: history is not encouraging :/ (source)

🏛️ White House: AI Productivity Should Let the Fed Cut Rates

White House economic adviser Kevin Hassett told CNBC that a "supply shock" from capital spending and AI-boosted productivity is putting downward pressure on inflation, which should give the Federal Reserve room to lower interest rates. Hassett also said he expects rates to come down if Kevin Warsh, Trump's nominee for Fed chair, takes the position.
Why it matters: This is one of the first times the White House has explicitly linked AI productivity gains to monetary policy. "AI is making everything more efficient so please cut rates" is... certainly a take. If the argument holds, it could reshape how policymakers think about interest rates in an AI-driven economy. If it doesn't hold, well, at least it sounded smart on CNBC. (source)

🏢 Oracle Hires an Energy CFO Because Its AI Spending Is Getting Scary

Oracle named Hilary Maxson, CFO of Schneider Electric, as its new CFO - bringing in deep infrastructure and energy experience as the company faces scrutiny over ballooning AI investments. Oracle has forecast $50 billion in capex this fiscal year - more than double the prior year. Meanwhile, its free cash flow swung to a deficit of $394 million in fiscal 2025, and shares have dropped ~25% this year. The company is also laying off employees while redirecting investment into AI.
Why it matters: When your AI spending is so aggressive that you need to hire an energy infrastructure CFO to reassure investors someone's watching the checkbook... that's a sign. As AI infrastructure costs skyrocket across the industry, financial discipline could separate the winners from the cautionary tales. (source)

🔐 North Korean Hackers Catfished Their Way Into Axios (The npm Package, Not the News Site)

North Korean hackers spent weeks building trust with the maintainer of Axios, one of the web's most widely used open source projects, before hijacking it on March 31 to push malicious code to potentially thousands of systems. They posed as a real company, created a fake Slack workspace with fake employee profiles, and tricked the maintainer into downloading malware disguised as a meeting app update - a technique previously attributed to North Korean cyber operations.
Why it matters: North Korean hackers stole over $2 billion in crypto in 2025 alone, and this attack highlights the terrifying vulnerability of open source infrastructure that millions of apps depend on. They didn't hack anything - they just... made friends first. Social engineering is the real zero-day exploit. I'm screaming at the level of effort that went into this. (source)

📊 OpenAI & Google Rapid Fire: IPO Drama, Iran Threats, Healthcare AI, and More

The Information reported that Sam Altman and OpenAI CFO Sarah Friar disagree on IPO timing (OpenAI denied it). Iran's military singled out the $30B Stargate data center in Abu Dhabi as a target, publishing satellite footage. ChatGPT is now fielding 2 million insurance-related messages weekly. Google released AI Edge Eloquent, a free iOS dictation app that runs entirely on-device. And Legion Health won approval to let its AI app directly refill psychiatric medications without clinician oversight - a first.
Why it matters: AI's reach is expanding into geopolitics, healthcare, and everyday tools simultaneously. The Legion Health approval especially is a watershed moment - an AI app autonomously refilling psychiatric meds with no doctor in the loop. That's either the future of accessible healthcare or the scariest/most impressive thing I've read all week. Possibly both?? (source)

🚀 "Vibe McKinsey" Is Here and It Costs $250/Month

Indian startup Rocket launched a platform that generates consulting-style product strategy documents - pricing, unit economics, go-to-market plans - from simple text prompts. Plans range from $25 to $350/month. The company has grown to over 1.5 million users across 180 countries since its $15M seed round in September, drawing on 1,000+ data sources including Meta's ad libraries and Similarweb's API.
Why it matters: As vibe coding makes building easy, Rocket bets the real gap is knowing what to build. McKinsey charges thousands for strategy decks that may or may not be AI-generated anyway - Rocket's just cutting out the middleman in a Patagonia vest. If the outputs hold up, this could democratize strategic thinking for founders everywhere. We love to see it. (source)

🎨 Picsart Opens Creator Monetization to Everyone (Yes, Even You)

AI design platform Picsart launched a creator monetization program with no invite lists and no minimum audience size. Creators complete campaign-based challenges using Picsart tools, share content on social media, and earn revenue based on views, comments, shares, and reach. Earnings are tracked via a dashboard and paid out through Stripe. The platform has over 130 million users worldwide.
Why it matters: Picsart is evolving from a creative tool into a creator economy platform. By removing follower minimums, it's betting that AI-powered creativity can level the playing field for smaller creators who've been shut out of traditional monetization. No gatekeeping?? In this economy?? Refreshing. (source)

🛰️ Spanish Startup Raises $130M to Map Earth for AI (Before Even Launching Satellites)

Spanish startup Xoople closed a $130 million Series B to build a satellite constellation that collects precision data for deep learning models. They also signed a deal with L3Harris Technologies to build sensors for its spacecraft. The wild part?? Xoople's strategy is to embed its data directly into enterprise platforms like Microsoft and Esri before even launching its own satellites. Total raised: $225M.
Why it matters: AI models are only as good as their training data, and high-quality Earth observation data is becoming a critical input. Xoople's "distribution-first, satellites-second" approach is unconventional but honestly kind of genius - sell the data pipeline before you even have the satellites. It's giving "fake it till you orbit it" energy :) (source)