Diogo Almeida says 29.4% of the Fortune 500 showed up in 3 weeks
A model launch, accidental trillions of tokens a day and an a16z Series A, with no revenue number anywhere in the post.
Diogo Almeida (@completeskeptic) posted a six-line summary of his team's last three weeks. They launched a model. They "accidentally" served trillions of tokens a day. 29.4% of the Fortune 500 showed up. And they raised what he calls a really big Series A from a16z.
The post ends with "it's been 3 weeks, we would like to sleep now". There is no revenue figure, no company name and no pricing detail anywhere in it.
What the numbers are and are not
Two of the three metrics are reach, not money. Tokens served per day is a usage number, and the word accidentally is doing real work there, since serving trillions of tokens a day is a cost line before it is a revenue line. The 29.4% Fortune 500 figure counts logos that showed up, which for a freshly launched model usually means somebody at a large company made an API call. Neither tells you what the business collects.
The oddly precise 29.4% is the tell. It is the kind of number you can only produce if you are watching company-level signups closely enough to compute it, which is a reasonable thing to track in week three and a poor substitute for ARR (annual recurring revenue) in month twelve.
The funding detail came from someone else
KP (@thisiskp_) posted figures in response, writing "$870 million series A at a $7.5B valuation, with @martin_casado joining the board?" and adding that the board seat is "worth more than the money". KP names Typesafe AI and a founder called Jev in that post. Almeida's own post names neither the amount nor the valuation, so the dollar figures here are KP's framing, not the company's.
$870 million series A at a $7.5B valuation, with @martin_casado joining the board?
The business read for anyone building on top of model providers is that this is now the shape of the public scoreboard. Token volume and enterprise logo counts arrive fast and get posted fast. Revenue arrives later, and in this case has not been posted at all. If you are picking an inference vendor on the strength of a launch thread, you are picking on adoption signals that are three weeks old and a billing model nobody has described yet.
it's been 3 weeks we would like to sleep now
Steal the numbers, not the vibes
One email a day. Under three minutes. Real MRR, real pricing, real launches from founders building with AI.