the model doing the work has no logo

i sat in on a call this week with a fund manager who runs a group of ai focused index funds. he'd just launched two new ones built around the china side of the trade, and the reason wasn't china.

eighty percent of the new companies going up right now, he told me, are built on top of models most people in this business never say out loud. not the labs on every conference stage. the open ones, downloaded and fine tuned by whoever wants them.

the same week, moonshot's kimi k3, an open weight model out of china, kept showing up in coverage as the base for products built far outside china, and moonshot itself filed for a hong kong listing, unannounced, at a target near fifty billion dollars. deepseek is lined up for the same move.

none of that made the front page the way the big labs do. the fund manager's read: expect the model that built the thing you're using to go public within six months, quiet about it, while the argument over who's ahead keeps happening one layer up.

this is the part everyone building right now keeps getting backwards. you pitch investors on which frontier lab your product sits closest to, because that story raises money. the model underneath your product, the one doing the inference, the one you picked because it was free and good enough, never makes the slide.

it should. it's the piece deciding who gets paid.... a download that costs nothing today, filing to go public tomorrow, and eighty percent of a market built on top of it that nobody names.

a brand war is happening at the top, three or four labs, all closed, all raising at valuations that make headlines. underneath it a different market is forming, made of whoever downloaded the weights and shipped something real with them, and it has no logo anyone recognizes.

founders raising money right now feel this pressure already. a pitch deck leans on whichever lab's api sits inside the product, because that name sounds impressive in a meeting. the model doing the work almost never makes the slide, even when it's the one thing an engineer in the room could point to and defend.

those index funds are already forming around that second market. is your product?