🎉 Shoutout to @flipsidecrypto for putting together this awesome mapping of blockchain builders! 🔥 🙏 Grateful to @APompliano @cburniske @hosseeb @jony_levin @lawmaster @mitchliu @QWQiao @RyanSAdams @twobitidiot & others for leading the way! 🙌 flipsidecrypto.com/blog/best…

Been thinking about Sutton’s Bitter Lesson in a context nobody’s really applying it to: IP law. His whole thesis is that general methods running on scale and compute always beat human-crafted, hand-tuned approaches. Every time. Chess, Go, vision, language, doesn’t matter the domain, brute-force search and learning wins over people trying to encode their own cleverness into the system. Now apply that to intellectual property. If agents can run discovery and refinement loops 24/7, continuously optimizing toward whatever the “best” version of an idea, a drug, a design, a piece of code actually is, at some point the thing patent law was built to protect (a discrete, ownable, human-originated insight) stops looking like how the best IP actually gets made. So you’re left with a fork. Either patent and IP protection in the US stays basically as is, and gets quietly run around by whoever’s compute-rich enough to just out-discover it. Or the law itself has to change to deal with AI-native discovery, which means redefining who owns what came out of a machine loop, retroactively, for IP that already exists. That second path is the one people aren’t pricing in. Because if you change the rules on what counts as ownable after the fact, you’re not just updating patent law. You’re greenlighting the state’s ability to unwind property rights that were granted in good faith.

The @RobinhoodApp Chain mainnet went live yesterday and @Alchemy is one of the core infra integrations underneath it. Proud of this one, it’s the kind of partnership we’ve been building toward for years. The part I’d actually pay attention to isn’t the chain itself. It’s the shape of the thing. Stock tokens in 120+ countries trading 24/7. A 7% stablecoin yield built into a self-custody wallet. Agentic accounts where you plug your own AI model into their trading stack. Gas covered for 90 days to pull builders in. It’s more than a brokerage adding crypto. Everyone spent five years asking when TradFi would come onchain. It’s not coming. It’s here, and it brought 26 million funded accounts with it.

couldn't agree more

Everyone's reading today the same way: Warsh came in hawkish, dots show a hike, cuts are dead for the year. I'll take the other side. I think they cut, and the way he structured this announcement is the tell. Warsh has said for years he wants the Fed to talk less, vaguer guidance, stop pre-committing to a path. So when the dots printed hawkish, everyone read conviction. I read room to move. Once you stop telegraphing, you can go any direction next and call it data-dependent instead of a U-turn. That's not an accident. That's the whole point of the new comms regime. Now put oil on top of it. It's falling fast, Iran deal, Hormuz back open, the energy spike that drove this entire inflation scare is unwinding while we watch. That spike is the only reason he got to sound tough today. And it's already leaving. So the hawkish print isn't a forecast to me. It's air cover. He looks like an inflation hawk on day one, oil does the easing for him over the next few months, and when he finally cuts he points at the data and never has to say he changed his mind.

Second order thing that most people aren’t connecting. The Iran deal everyone’s trading as geopolitics is also a Fed story. Oil came down hard this week. That pulls some inflation pressure off, which is exactly the cover the Fed needs to sit still instead of leaning hawkish. So the peace deal doesn’t just move energy names, it buys Warsh some air cover on his first day in the chair.

So SpaceX bought Cursor for 60 billion. They signed this as an option back in April. Either do the deal or pay 10 billion to walk away. Then they just… waited. Went through the whole IPO, let the stock list and run up, and only after that flipped the option into a real merger. So Cursor gets paid in SpaceX shares, priced off wherever the stock’s trading the week before it closes. I keep coming back to the order of operations. They took the company public, and then four days later used the new stock to pay for an acquisition they’d already lined up months ago. The IPO basically created the currency they’re spending here. Maybe that’s just good treasury work. But it’s a lot of moving pieces in a very specific sequence, and I don’t think that’s an accident.

The French wine tariff threat is a cleaner lesson in leverage than most M&A deals I've seen. Trump doesn't actually care about wine. He cares about France's 3% digital tax on US tech. So he reaches for the thing France can't afford to lose, its luxury exports, and squeezes there. You don't attack where the fight is. You attack where it hurts. Macron said no anyway, which is its own kind of tell. Either he's confident, or he's decided looking weak costs more than the tariff does.

bought some solana

Everyone's transfixed by the $100B data center headline. Wrong thing to stare at. Stare at the depreciation schedule instead. The compute inside these buildings is obsolete in roughly three years. So the real number isn't $100B once, it's $100B on a loop, forever, or you fall behind. The shell, the power, the cooling, that lasts. The expensive part doesn't. We're building cathedrals around hardware with the shelf life of a phone. Whoever solves "how do we stop rebuilding the inside every three years" wins the entire decade.

Three weeks ago the US was blockading the Strait of Hormuz. Tonight Trump posted “let the oil flow” and the war is over. Signing in Switzerland Friday. Same country that in the last 72 hours hosted the biggest IPO in history and export-controlled the best AI on earth. I keep saying it and it keeps being true: nobody could explain this era to someone from 2019.

SpaceX goes public at $2 trillion. US scores four goals on opening night of their World Cup. Government shuts down the most advanced AI in the world with a 5pm Friday letter. Yesterday was a lot.

I’ve been in Web3 infra for years and I used to roll my eyes a little when people said “decentralization or bust.” Tonight Anthropic, the best AI lab in the world, had its two flagship models switched off globally because one company claims they found a jailbreak. No court. No appeals process. No timeline. Letter at 5pm, lights off by 6. This is the counterparty risk that’s been theoretical for five years. It’s not theoretical anymore. Web3 projects like Venice $VVV has been building the open-source alternative the whole time. Timing is everything.

SpaceX closed its first day at $161, up 19%, and the funniest part of my feed is people calling it a disappointment. A $2.1 trillion company. Day one. “Disappointing.” But honestly the bears have half a point and nobody’s articulating it right. Betting markets had this opening way hotter. It touched $176 intraday and faded into the close. On the biggest, most hyped IPO in history, with retail handed a huge allocation, the marginal buyer showed up and then… stopped. Day one tells you about demand. Day five tells you about conviction. I’m watching where this sits next Friday, not today.

pump it

SpaceX trades tomorrow and I keep seeing people debate whether 94x sales is insane. Wrong debate imo. Of course it's insane on the numbers. They lost $4B+ last quarter. Musk keeps 82% voting control. You're buying a story, not a cash flow. But I traded the SpaceX-adjacent names for months before this and the lesson was the same every time: this thing doesn't trade on fundamentals, it trades on access. The most wanted asset in private markets just became buyable by anyone with a brokerage account. First time in years. That demand doesn't care what the multiple is. Doesn't mean buy it. Means if you're shorting it day one because "the valuation makes no sense," you're bringing a spreadsheet to a sentiment fight