takeaways from MS TMT conf: - AI still being used more for IR and software development than workflow automation - winners need to own context, workflow, and distribution, moats exist around data and orchestration - pricing model still very much in flux and experimental - no immediate shift from straight seats->tokens, will be some combo of seats + usage + agents + credit - rev ramp will be gradual and play out over FY 26, not straight forward to go from copilots to production agents - huge emphasis on security, trust + verifiable outcomes, citations / data lineage in responses - proprietary data, proprietary data, proprietary data

Citrini’s Dash and Uber examples were intended to be more provocative than others. no one’s denying that network effects, brand, and distribution are real moats. But for these ride-sharing/delivery, and many other consumer-focused companies at the center of the winner/loser debate, the moats are: 1. reliability of the service/good provider 2. bundled economics/subscription lock-in 3. regulatory hurdles for new entrants For Uber/Dash, buyers don’t care where the food or driver comes from (esp when it comes to food delivery). They want the lowest price, provided a timely ride and food that doesn’t arrive messed up. So if agents can shop around looking for the lowest price between providers, the buyer becomes more provider agnostic. Most people don’t have the time or care to open anything more than Uber or Lyft when at the airport. but if there are 10 more providers in 5 years bc of robotaxis and the shopping is more efficient, then they won’t have strong preferences for uber vs ___. People care even less about provider on the delivery side. One of the challenges there is not knowing if a given restaurant is carried by Dash or Grub or Toast or UberEats. so you go to the restaurant website and select delivery and it gives you 3-5 options to choose from. No one cares which one they choose there unless they have an affiliated /pro sub to remove fees. And again, the driver/supply side should get solved in the not too distant future.

a big reason why there’s such a visceral reaction across the country to that game is because it introduced the non-fan to hockey guys. 24 yr old with the game winner, no teeth, just talking about how much he loves his buddies and wanting to do it for the guys before them. was on the fence about whether I’d try to get my kids playing hockey, but watching the game with my 1 yr old son, everything about that team made me want him to have a childhood with hockey in it. No other sport has what it does

instead of "not interested in this post" / blocking as the only real ways to curate the timeline, it'd be great to have a "see more of this type of post" that gives the user some discretion over the timeline, beyond simply clicking on a post and getting fed more @nikitabier

The quality of our thinking is not determined by the volume we read, but by the rigor of our engagement and the questions we dare to chase. @Evti on @phokarlsson "How I Read" henrikkarlsson.xyz/p/how-i-r…

does any stock more consistently bounce between a range than $snap and $8-$12? at $8 it becomes a takeover candidate, at $12 they prove time and time again that the narrative hasn't changed, DAUs are capped and give terrible guidance

very interesting from MS on retailers most/least exposed to resumption of student loan repayments this Fall

didn't even realize it but my follower count dropped from 800 to 122, i lost all my anime bots and would like them back @elonmusk

500k jobs in the regional banking system...going to test that "but the labor market is still resilient" narrative

% of portfolio holdings for Sweden's largest pension fund (Alecta): 5.1% in $FRC 3.6% in $SIVB 1.9% in $SBNY 🤔

two outcomes here - Fed removes FDIC limit, guarantees all deposits at SVB and puts an end to this thing OR hundreds of thousands of layoffs next week, SPX off 15-20% within a month, years of damage to private markets

Flattered by the dramatic growth in my follower count - will have a thread on audience building soon

props to @andrewrsorkin - not a fluff interview. pointed questions, not letting him pivot, trying to force him off script