“I sat down with my CTO today”
Not with the engineers, not with the people actually building, not with the logging data itself - his CTO.
Go out of your way to block these people, and you will have a stronger grasp on reality.
Chamath reveals his company's AI token costs are doubling every 45 days but productivity is only up 5%
"I sat down with my CTO today, I said how are we doing on token spend. And he said the most incredible thing, he said right now, our token costs are doubling every 45 days. I said well what is the downstream productivity? And he said maybe 5% max."
"So my costs are doubling every 45 days, my upside is essentially flat. He said honestly, what we're finding out is that you need to use a lot more tokens to get to this next iteration of improvement because we've effectively already asymptoted."
"We're going to take a step back and try to figure out what to do. I don't know how many other companies will actually go through this reckoning now, but the point is everybody in the next three or four years will for sure go through it."
"I suspect that if you can get out now, you should get out now before all of that starts to seep into the water table. Because I think that's probably what allows you to get out at a huge price and raise a huge amount of money."
If benchmarks here are aligned with reality, pricing of $2 / $6 on input / output is exceptional.
Very easily leading on price / performance, if that’s the case (have not tested). For reference, Opus 4.8 is $5 / $25 (call it 2.5x cheaper, if being generous to @AnthropicAI).
@AnthropicAI wants a week to see how GPT-5.6 is, if it’s good enough they’re going to extend this. If it’s weak, they’re going to set Fable 5 out to API only.
@PythiaR No different than the Luddites of the early 1800s, where skilled laborers (textile workers, today journalists) protested the use of machines to accelerate production of finished goods.
As early as 1788 the UK had to pass laws, protecting against sabotage of knitting machines.
Confirmation from Anthropic that it was $AMZN who snitched to the US govt.
If I am Anthropic, I would work hard to shift my compute away from AWS. Your own compute provider and large investor running to the regulators without talking to you first.
@ajambrosino That’s because you’re actually taking great human care with the Codex app (the @skybysoftware team I would bet is involved).
The highest praise I can give is that Codex actually feels like software.
Until today, agents could buy about 2,000 tools through x402.
We just 10x'd that to 20,000+ 🚀
In partnership with @coinbase, we’re launching x402 support to give autonomous agents access to the largest marketplace of web automation tools.
No account, API keys, or human in the loop.
$WIX transitioning into becoming basically a frontend model development + deployment / implementation company might be a more viable path than whatever the present path is.
They currently employ 4500 people, orienting the entire company around this maybe most correct.
@Base44 has always been about big engineering bets.
when base44 was one person, we built our own backend as a service to power every app on the platform, instead of leaning on a 3rd party (like the rest of the industry).
most people thought it was a mistake. owning that part of the stack became one of our biggest advantages. it let us tune the backend to exactly what builders need.
today we're making a bigger bet of the same kind. we're training our own model, which makes base44 the only vibe coding platform that owns the entire stack, from the model down to the infrastructure your app runs on.
the early signals from the model looks good, but not great yet. it's still early, so we're putting it out now to gather feedback and iterate with you, making it better with more traffic from more builders.
Base 1 is live in the model picker today.
For those with us in this journey and for our fantastic community - please go give it a try and build something with it.
in the next few weeks we'll publish a technical blog on how we built it and how it performs on the Basemark benchmark.
more on the bet in the blog below 👇
https://t.co/o0IMYjaRKP
@sagharborcap Because they lack the focus needed to see it through, and are not even attempting to do fundamental investing - they tried competing with firms like RenTec, not you.
If the Trump administration thought they needed to sign a deal w/ Iran because of oil risk to markets - just wait til they wake up and realize the risk to markets from arbitrarily limiting deployment of US frontier models.
The uncertainty alone can wipe capital investment -30%.
@mclovin238@pmarca If it was, they would name the companies and the specific sectors.
And how the hell is eg JP Morgan going to get access as a “critical infrastructure” company (G-SIB), but not some regional or local bank? You can do this for literally any other sector / industry.
We believe in broad access and plan to make GPT-5.6 Sol, Terra, and Luna generally available in the coming weeks.
For now, at the request of the U.S. government, we’re starting with a limited preview among a small group of trusted partners in Codex and the API.
@RealJimChanos Because the operating income is artificially depressed, and it’s trading nowhere near 150x GAAP normalized.
You’ve done the actual math?
The @warriors have a new name on their jersey and it’s an AI company — Iren. I got to hang with @JimmyButler, Joe Lacob and Iren CEO @danroberts0101 to talk about the new partnership, the future of AI and #winning another NBA title.