Even a 5090 has something like 50-60 flops per byte of bandwidth, you just can't saturate the compute without running large batches. (At least at inference, prefill is obviously more compute bound).

How many tokens can you realistically burn through in one chat session? Opus and many other frontier models do maybe 60tok/s, less 250k/hr out. In you can use more, but in most cases cache is 5-10:1 cheaper than new input. Say you average 500ktok in, 90% cache, per request. That amounts to 100-150ktok in new input-equivalent costs, which in most cases is ~20-30ktok in output-equivalent costs. Do a request every minute, that's a total of about 1.5-2Mtok/hr. At API prices that's $50/hr for Opus, but really it probably only costs Anthropic $10/hr to serve that.

That said, even if a developer is burning $50/hr, many, many employees at large companies cost more than $100k/yr to employ all costs considered, so making them say 20-30% more productive can easily make that worth it for most. If the labs shave their margins ultimately to more like 20-30%, you'd have ~$15/hr in costs to use the services, and nearly every white collar job is way over 30k/yr to employ. If your salary is 80k, you probably cost the company 200k all in, so making you 15% more productive offsets the $15/hr cost.

So first party providers are not in a horrifying position or anything from a subsidization standpoint. The people in bad shape are Cursor and Perplexity, who don't have frontier models and are dependent on the open source community, which is typicly 6-12 months behind the frontier. They have to pay full freight API costs at 80% margin for the big boys to serve their harnesses, which is indeed untenable, and they'll have to either force users to use open source models and/or in house models they can serve at-cost or they will have to charge vastly more.

Gemini, Claude, and ChatGPT first-party services like Antigravity, Codex, and Claude Code are not in serious trouble though.

So I'm not sure it matters whether a giant company uses it because random users can get the same thing for ~ free anyway.

Not ready to share just yet but nearly at the point that there are no Julia-syntax fallbacks in the entire base/stdlib and a femtolisp parser for the sexp syntax is able to build a complete Julia sysimage from the transpiled files. Already verified that I can transpile the .jl source of the Julia package for the syntax into the syntax, then use that transpiler to transpile again and load into the running sexp repl, then use that transpiler on the source again and get byte identical code, and along the way am testing to ensure that the entire Julia test suite passes in the sysimage being built.

So, with any luck here soon I'll have a sexp syntax for Julia that builds from raw transpiled sexp-syntax source and uses sexp syntax natively in the repl but can transpile & load any Julia code. Fingers crossed.

I'm aware of --lisp but it's not very good imo lol.

For instance, once you develop atomically precise manufacturing ala Drexler and have a complete model of biology, etc., drive solar panel efficiency to very near the upper theoretical bound for infinitely many junction cells for a raw panel of ~68%, then there isn't really anywhere to go that matters for humans. Material production would be solved, anything you could desire would be manufacturable in minutes to hours, a km^2 of solar panels could power 10-20k people's post-scarcity lives.

You eventually reach the upper bounds on compute efficiency and human upload model efficiency -- unknown but given estimates on upper bound for like rod logic (~1e-34Js/op), reasonably bounds on op speed (100MHz), and low estimates for functional uploading (1e16 flops), you get something in the zone of 0.1nW/upload, or several trillion individuals on 1m^2 of solar panel in space. When you put a simulated Banks Orbital around every star in the Milky Way in a grand sim running on a system of solar panels in space where the entire simulated galaxy has a 15ms ping to any other point in the simulated galaxy, what exactly is this infinite stream of learning? You've pushed technology to the the limits of physical law subject to the constraint of being made of atoms.

Are you envisioning that we'd eventually be doing computation using the entirety of a neutron star or (if they can exist) a quark star? Even then, you eventually hit a wall where physics constrains you from making significant further gains.

There is an ultimate end to the s-curve of technology.

It's possible they'll be even cheaper but that range is the cost according to the IRS of operating a typical vehicle all in, and that seems like a reasonable guess of the cost of an autonomous electric vehicle with far lower probability of crash than a human (all the savings basically going to profit margin).

At ~60¢/mi, there'd be a lot of people who would save money on balance using autonomous taxis to get everywhere vs owning a private vehicle (10k mi/yr would cost only ~$6k/yr, a pretty low cost of ownership/use for a private vehicle).

This one will have even more battery life, and gets 12 hours of use in 15 minutes, which I suspect will mean for me without the always on display I may well be able to charge it only while I'm actively in the shower (when I'd take it off anyway as I hate wet bands) and be good for the day.

I don't have much call for most of the camera system, and my battery life on my Pro is just fine. I have plenty of chargers typically, and for emergencies or times I know I'm going to be out I could potentially get the battery pack.

I basically never use cases on my iPhone, and at most will maybe use an ultra-thin one or some sort of structure adhered to the plateau just to make it flat across so as to not rock on a table.

If you don't want them to have a history only communicate via disappearing messages.