O'Leary on AI warfare: "They will be tenderized constantly. Every data center built going forward in the US is going to have to give up at least 20% of its capacity to the US government."

NYDIG calls CLARITY Act "the most important forward catalyst" for digital assets.

OpenAI's compute chief on the bottleneck nobody talks about: a massive shortage of electricians, plumbers, and tradespeople to build these data centers. "They're very well-paying jobs that all of the hyperscalers and labs would actively hire for if you had the qualifications."

OpenAI's Head of Compute: "We believe the world of recursion is not that far where AI will design the systems it needs to train and run the next generation of AI, including chips."

GameStop CEO Ryan Cohen says "a lot of parties" have expressed interest in joining the eBay acquisition.

Belarus is seizing property and bank accounts of democracy activists living in Canada. Families still inside Belarus are being harassed as leverage. Photos from a rally in Alberta become evidence of "extremism." Apartments are confiscated without court dates or appeals. Since the fraudulent 2020 election and the mass protests that followed, over 1,100 political prisoners remain behind bars. More than 8,000 politically motivated convictions. An estimated 500,000 to 600,000 Belarusians have fled the country. But fleeing doesn't mean safety anymore. At least 207 participants in Belarusian Freedom Day events held in Canada, the US, UK, Poland, and Lithuania were identified by authorities and charged with "facilitating extremist activities." The playbook is straightforward: seize property back home, threaten family members who stayed behind, file charges in absentia with sentences up to seven years and fines up to $750,000. "Extremism" becomes whatever the regime decides it is. Facts and evidence are irrelevant. This is transnational repression. Authoritarian states projecting control beyond their borders to silence dissent in free countries. It's happening from Belarus, Russia, China, and Iran. The assumption that democratic nations provide a safe haven for political exiles is eroding in real time. Distance no longer protects you.

Joe Lonsdale on China going from a fraction to over a third of new therapeutics by stealing US IP. The stat that kills: pre-IND approval in China takes 90 days. In the US it takes over 700 days. "We have to accelerate this if we're going to compete."

"It's like a mafia where if you criticize it too much, they might just decide to make you wait a year. Unchecked, unaccountable power. We shouldn't have that in a free society." Lonsdale says the FDA needs competition and accountability.

Palantir co-founder Joe Lonsdale: "The primes have congressional districts locked down. They automatically get the contract even if their stuff sucks because it's a jobs program. It's not about deterring our enemies."

Me when someone says they're waiting for bitcoin to drop back to $30k.

Thinking Machines Lab released Inkling, a 975B parameter model with open weights available for download. That last part is the story. Most people interact with AI by renting access to a black box. The model can be updated, the price can rise, the acceptable-use policy can tighten, a government can pressure the provider, the service can disappear. That's fine for casual chat. It's a dangerous foundation for critical business processes, medical research, financial workflows, and national infrastructure. Inkling is an American counterexample to Chinese open-weight models like DeepSeek and Qwen that are spreading globally because developers can download, modify, and build on them. The country that supplies the foundation shapes the next layer of global products, standards, and technical dependencies. Winning the benchmark while losing the ecosystem would be a hollow victory. The argument is Hayekian. Much of the knowledge that makes an organization valuable is tacit, local, and private. A centralized lab cannot collect all of it. It cannot define one alignment layer that reflects the values of every person, company, community, and country. Customizable models let organizations encode their own specialized knowledge into the weights instead of renting the same generic model as everyone else. Open weights are not the end state of AI sovereignty. They are one necessary step toward it. The bitcoin parallel is obvious. Bitcoin gives people the ability to own money without renting permission from a central bank. Open AI points toward the ability to own intelligence without depending on a corporate API. The AI race is not only about intelligence anymore. It is about sovereignty.

Eli Lilly just dropped $2.8B to acquire AtaiBeckley and enter the psychedelic medicine space, treatment-resistant depression, DMT therapy, MDMA for social anxiety. But this isn't just a pharma company buying another drug pipeline. Jordi Visser has been making the case that LLY is fundamentally mispriced. The Street sees an obesity drug company. Visser sees "an AI company with an infinite balance sheet." Lilly has ~1,000 Nvidia Blackwell GPUs on campus running drug discovery with biotech partners. 55% revenue growth in Q1 2026, Nvidia-like numbers from a 150-year-old Indiana pharma company. The AtaiBeckley deal fits the thesis perfectly. Psychedelic-based treatments like BPL-003 (a nasal spray for treatment-resistant depression showing effects lasting months after a single dose) are exactly the kind of novel molecules that benefit from AI-driven drug discovery. You're not just buying a pipeline, you're feeding new compound classes into an AI engine that can accelerate development timelines and identify applications that traditional R&D would take decades to find. Lilly is building a healthcare AI platform disguised as a pharma company. GLP-1 drugs fund the machine. AI accelerates the pipeline. And now psychedelics give them a whole new frontier of neuroscience to run through it. If you own LLY for Mounjaro, you're underwriting the wrong thesis.

Treasury Secretary Scott Bessent: "38% of American households own no equities. There are financial service deserts where people haven't been able to get access to accounts like this."

Treasury Secretary Scott Bessent: "AI is going to let small business compete with big business. A business that used to take 10 people to start, you can start with one or two."

Bessent: 90% of precursor chemicals for our medicines come from China. "We worshiped efficiency and gave up resiliency, we wanted consumption but didn't focus on production."

.Jeffrey P. Snider says Marx's predictions map perfectly onto current lived experience, endstage capitalism, exploiting workers, housing unaffordable until 40. And why it's actually a cyclical pattern, not the endgame.

Senator Lummis pushing for a full accounting of the SVB collapse. Fed Chairman Warsh has committed to full staff cooperation. We'll see.

Kalshi now letting you bet on whether a drug passes clinical trials. We've gone from "prediction markets are a revolutionary information tool" to "place your bets on whether grandma's cancer drug gets approved."

$14.4 billion has left the largest US gold-backed ETF since March 1st. To put that in perspective, that's 50% more than the total outflows across all Bitcoin ETFs since their October peak. March alone saw $8.5 billion pulled from $GLD the largest single-month withdrawal in the fund's history. April added another $1.7 billion, May $872 million, June $3.2 billion. July has cooled off significantly, tracking just $46 million in outflows month-to-date.

Over 108 days, the war in Iran has cost U.S. taxpayers an estimated $113.3 billion and counting. That's $1.05 billion per day, more than the Small Business Administration gets in an entire year, burned every 24 hours. In less time than a single baseball season, the government spent more than the combined annual budgets of NASA, the EPA, FEMA, and the Department of Justice. More than every federal highway, bridge, and road project in America for a full year. Every taxpayer's share: ~$691. A family of four: ~$1,360. For a war most Americans couldn't find on a map three months before it started. And these are just the direct costs. Brown University's research on Iraq and Afghanistan shows that long-term costs like veteran care, equipment replacement, interest on the debt used to finance it, typically multiply the sticker price by 3-5x. The real tab for this 108-day war could land somewhere between $340 billion and $567 billion. Afghanistan's first 100 days cost ~$31.5 billion. Iran's 108 days cost 3.6x that. The money is always there for war. It's never there for anything else.