@jamonholmgren@DarioAmodei has mentioned 50% percent margin on inference. subscription and API combined. 50% margin after subscription tokens being 40x cheaper than the API pricing. inference costs would come down 10x a year, plus margin compression if there is enough competition
I've had ChatGPT-5.4 Pro working away at a project I always wondered about: how lucky are you to be alive right now?
Of all the ~117B humans who ever lived, only about 1.5% had a lifestyle roughly equivalently to a middle-class person in a middle-income country today, or better.
There are 3 elements to improving models:
1) Architecture
2) Compute
3) Data
No one is changing (1), (2) is actively being solved by the compute giants.
Now what’s left is (3), which has effectively become 2026’s “pickaxes in a gold rush.” Today, the choke point is fully human-created data.
We at @VibrantLabsAI believe AGI will not be achieved by human data alone, so we’re laser-focused on synthesizing as much as possible to advance models to the next frontier.
@levelsio@levelsio how do you test what you are building ? Automate it through playright? Isnt it very buggy and unreliable?
What about the look and feel of the app? How do you run and test it?
Thank god MCP is dead
Just as useless of an idea as LLMs.txt was
It's all dumb abstractions that AI doesn't need because AI's are as smart as humans so they can just use what was already there which is APIs
@naval Yes.
AI will drain moats built through years of engineering effort. Adobe, AutoCAD, Tableau, Jira ....
But network moats like WhatsApp, YouTube, GitHub, Uber, Amazon, Airbnb, Reddit .... AI alone is not enough here.