leo alterman

1/ There are two kinds of stock. Stock everybody wants. and stock nobody wants.

a distribution is permission.

the same deal at 2% of the fund and at 20% is two different investments.

in the dotcom mania, the winner's pitch (amazon) was selling books. "transform the economy" pitches died more often. winners often start by selling something boring.

AI capex. Not the size of the spend. Who pays for it: Phase one: retained earnings. Phase two: bonds and private credit. Phase three: suppliers financing their own customers so the orders keep coming. hyperscalers are in phase two.

AI didn’t create the gap between authenticity and the feeling of it. It just made the gap visible.

if the frontier model gets 90% cheaper tomorrow, you're either stronger or dead. stronger means you were buying it. dead means you were reselling it.

when everything goes wrong at a startup, it's rarely everything. unbundle it and it's a to-do list.

paper gains can't compound outside the fund. cash can.

sales is king. profit is queen. the queen only takes power once the king dies. in AI she's arriving early. every query costs money before anyone knows if the customer stays.

many AI agents solve a problem the demo created. the customer never had it.

if you can't choose to believe one company over another, you'll spend your career trying to sit one bottom on two chairs. an opinion is the price of ownership.

growing number of companies are worth more DEAD than alive. one word - data. models need to eat, and ‘new food’ is running scarce. (earlier this year Google outbid Mercor at Spirit Airlines’ bankruptcy auction paying $10M for their emails and teams chats)

can’t wait to pay $200 to get this approved as my service animal

every wave builds the rails before anyone knows the use. in 2020 it was IoT satellites. in 2026 it's gigawatts. the question is never whether the rails get built. it's who is paying, and what they think they bought.

an LP's venture allocation is the amount they can watch go to zero without changing anything. size from that number, not from a target IRR. if you can't lose 20%, you won't make 20%.

time for world to see the world

you don't find a pre-consensus deal. you get invited to one.

equity is the elephant. debt is the whale. bigger, quieter, less covered. the AI buildout is being financed by the whale while everyone's watching the elephant.

failure is not the opposite of excellence. mediocrity is.

buffett bought gillette because whatever happened to the internet, people would still need to shave. the AI version: whatever the next model does, will people still need this?