Founder & CEO @ Supertab | Economic Infrastructure for the AI Usage Economy | Pay-Per-Use, Pay-at-Call, Metering & Settlement @getsupertab
New York, USA
Joined April 2014
- Tweets2.1K
- Following662
- Followers715
- Likes3.5K
Remember when Steve Jobs put a thousand songs in your pocket? The real revolution was that for the first time you could buy the single song. As many songs as you wanted, but only the ones you wanted. (And it did more to kill piracy than any lawsuit ever did.)
Today’s agents - and tomorrow’s fleets of agents - want the same thing. Small units, only what they need, but fast, and at scale. And they'd want a way to pay for exactly that.
Amazing! @Microsoft365 is turning on pay-as-you-go by default for new Copilot Business licenses, on top of the seat. The company that perfected the seat license is metering its own AI, and that's a great call. A chat answer is about 5,000 tokens, but one agent workflow can run to a million... Seats price the person. Meters price the work.
We've run pay-per-use for readers for over 10 years, and now for agents and AI. SPIEGEL sold 5.7M articles one at a time on a tab. For people, pay-per-use solved subscription fatigue. With AI it's different: every answer has a real cost, so the unit economics force you to pass that cost on with a margin. Good to see the biggest software company in the world land in the same place.
I’ve been early before. Early is uncomfortable. It’s also the only time the position is still open.
Supertab Connect is now on Fastly - bot classification + content licensing, at the edge. The rails for paying AI go in now. Engage early, learn fast, compound. Or catch up later. #justconnect @fastly
To read one article on futurism.com, I was asked to let 1,220 “technology partners” store and access my data.
That's just absurd. It used to be a few hundred, and that was already ridiculous.
I'd rather pay a few cents for the article, but since that isn't even an option, I simply reject the offer.
I’m pro ads. But this is really the path to making money with them?
Cosmin Ene retweeted
Tired: Two-pizza teams
Wired: Two-slice teams
@danshipper on the best team structure in the AI-era
Cosmin Ene retweeted
.@chamath made a killer point yesterday on the @theallinpod.
@Apple and @Google Android store revenues are fxcked in an Agentic AI world. Because they will be able to easily bypass these GUIs to give you whatever you want.
Think about this when you're invested in these companies.
Also, think about how much more revenue will be able to go to creators that are currently being robbed for 30% of whatever they make.
Cosmin Ene retweeted
Your next users are machines
Agents burn nearly 5x the tokens people do, up 14x in six months
AI agents don't use the cloud like humans do. AWS CEO Matt Garman on rewiring AWS for its newest "users":
"Increasingly we're seeing teams that want a cloud that is great to work with agents and not just with people."
" We've doubled down on things like starting a database in three seconds and some capabilities that agents are excited about."
"Many agents wanna create a database, do a little bit of work, and then have the database go away. You don't really need that database to have five nines of durability."
"There are new building blocks that agents are gonna want that we just didn't really need before: compute sandboxes, gateways, agent permissions versus people or service role permissions."
" A lot of those are things that we have built and are building and thinking actively about because they are just brand-new building blocks. It's not using the existing building blocks differently."
@mattsgarman @RaghuRaghuram
Cloudflare shipped Pay Per Use this week. Right model, I've been building it for fifteen years.
One thing to read closely: the buyer defines what counts as a use, sets the price, and reports that it happened - a timestamp, a URL, an event ID. Not what it was for.
Unit, rate and meter, all on the buyer's side.
I think you need to keep the model, but move the pen.
(Biased, and I love it - we build this stuff.)
The unit economics crisis in AI is real.
AI companies optimise for two things: distribution, and unit economics.
Distribution is working but the economics aren't.
OpenAI's 2025 accounts, obtained by Ed Zitron and verified by the Financial Times, show about $34B in costs against $13B in revenue, and a loss bigger than the revenue.
You can't pull a large upfront check out of a company that ended the year like that. Pushing harder doesn't put the money there.
Which is why the form of the ask matters more than the size. A company that can't write a big check can still pay for what it uses, every time it uses it - that‘s monetization at grounding…
Two weeks of watching this argument: everyone agrees publishers should be paid. Nobody agrees on the unit. Per deal, per crawl, per citation, per grounding. The unit is the whole fight.
The New York Times at Digiday this week: is it worth breaking your content apart and selling it for fractions of a cent?
Right question, wrong unit. Nobody invoices a fraction of a cent. You collect them and settle once.
That's a running tab - for agents...
digiday.com/media/what-the-n…
And notice who gets to ask. The Times has the brand, the lawyers, the leverage. The publisher with 40,000 readers will never get the call. For them it isn't a good deal versus a bad one. It's a price on the page, or nothing.
We built one and patented it. It's being tested with AI companies and content providers now.
Biased, obviously.
The US government told the court in NYT v OpenAI that licensing for AI training is too cumbersome. It is, if every deal needs lawyers and a spreadsheet. But it isn't, if the price sits with the content and the machine pays when it reads or consumes.
Cumbersome is a rails problem, not a rights problem…
(I build those rails at Supertab.)
