Sarah Eisen (in Jerusalem) sat down with Israeli PM Benjamin Netanyahu:

Iran “tends to lie and cheat” — deal requires Iran to dismantle nuclear infrastructure and remove nuclear material, not just pause enrichment

Trump used “FN crazy” language to Netanyahu; Netanyahu described it as a “tactical disagreement” — “we’ve always found a way”

Israel and US “ready to strike Iran again if necessary” if ceasefire doesn’t hold

Lebanon incursion ongoing: ceasefire conditioned on Hezbollah not firing; Israel will respond if they do

Netanyahu’s goal: nuclear disarmament, not regime change — framed as more realistic

Michael O’Hanlon (Brookings) analysis: three sticking points — nuclear enrichment limits, Hezbollah disarmament, Strait of Hormuz tolls; each has a viable compromise but none of the current proposals work

Iran may believe it has the upper hand as Trump faces midterm pressure

Iran running out of places to put its oil; economy “a shambles” — both sides are jostling for position

Leslie Picker reported: $135/share, 555.6 million shares, $75 billion total offering

Triple Alibaba’s US IPO record; fully diluted valuation $1.75 trillion

Includes valuation from EchoStar spectrum license deal and potential Cursor acquisition; excludes most of $1 billion in Mars colony performance shares

Fixed price chosen after extensive “testing the waters” with institutional investors — company feels demand is already well understood

Retail 30% of offering: harder to assess because so fragmented; roadshow begins likely tomorrow morning

Kelly Evans framing: “why pick a price now?” — Leslie Picker: “Retail is known to be price insensitive” at this valuation

Elon’s unconventional approach: not unusual for Musk; the JPMorgan event (90 locations, 26 states) is how they’re executing the retail component

Token maxing “never made sense” — any system where goal is to spend as much as possible “won’t last long”

“You would be crazy to be coding in 2026 and not using AI in some form” — Oracle CEO said “we literally work differently now than we did a year ago”

AI models are commoditizing: Anthropic = “ultra premium”; OpenAI = premium; but cost per token is a tenth or hundredth of competitors and falling

Investment framework: “further down the stack (copper, power, chips), the safer the bet; higher up (Anthropic, OpenAI), you need a very specific view on price defensibility”

On Meta CapEx (70–80% of revenue): rationale is that Meta has more human attention than any product in history; AI is the most straightforward way to monetize that at scale; the option to own infrastructure is valuable even if not everything gets used in-house

STORY_BULLETS

STORY_BULLETS

STORY_BULLETS