Today, we enable AutoResearch in the physical world for the first time! Introducing ENPIRE: we give 8 Codex agents a fleet of robots, an allocation of GPUs, and generous token budget. We set them free with a simple goal: solve the task as quickly as possible, keep the robots busy but stay safe, don't waste precious compute. Make no mistake.
Then humans step aside and our watch begins. The robot fleet starts to come alive: they learn to look for visual clues, reset the scene, practice novel skills, tinker with control stack, read papers online, debate, reflect, get stuck, and try again directly on the hardware. All we did is to give Codex an API to the world of atoms, and the rest is emergence.
ENPIRE is able to solve high-precision tasks like tying zip-ties, organizing fine pins, and installing GPUs all by itself. We also discovered a new type of "physical scaling": 8 robots exploring in parallel improves significantly faster than fewer ones.
A part of our NVIDIA GEAR lab now self-improves tirelessly over night. We just read the reports in the morning.
/goal: we all take a holiday and Jensen wouldn't even notice ;)
We will be open-sourcing everything, so you can host your self-running robot lab at home too! Deep dive in the thread:
My favorite piece of career advice comes from Marc Andreessen
The first rule of career planning is to not plan it at all
The world is simply too dynamic to box yourself into a singular pre-determined path for your entire life
@ShanuMathew93 Hypothetically the GPU-hr costs are connected to utilization. So the idea that chips aren’t profitable because of utilization is the backwards way of thinking about it. Paybacks with 50% utilization and 5$/hr are still around 3 years no? So that’s the point.
I have been very impressed by @SemiAnalysis_ . I think of myself as a wide ranging systems engineer, looking for value at every level from the chip specs to the user interface, but SA exposes me to additional levels of "the system", both above (datacenters) and below (semiconductor fabrication). It probably puts me in "just knows enough to be dangerous" territory.
Neat things I learned today:
Some of the 800VDC datacenter design choices leverage parts commoditized by electric vehicles.
There is now a SiC MOSFET that can operate on 10kV electricity, opening up the possibility of working directly with medium (ha!) voltage AC power transmission lines without stepping down.
@TKavulla@Astoll15@ts_fisher@SimonMahan@Microsoft@NVEnergy The second order price impacts (gen moving further up the supply stack) are going to come up here too. That’s probably a bridge too far but this is really more of a PR issue than a fairness one.
@ts_fisher@TKavulla@SimonMahan@Microsoft@NVEnergy It seems to me like the only way to ensure it’s palatable would be to prove that other ratepayers get cost savings rather than adhering to but for principles.
Call it “but for” plus 15%
You have ONE LIFE. You have limited time. There is, for all intents and purposes, infinite content.
Every time you consume info, you are not consuming infinite other info, including literal cinema.
Your personal content strategy shapes your entire world! Choose the best!
@sdamico@mattyglesias Oh 100%. Just commenting that while that’s true the utilities and regulators have generally screwed up TOU rates, which is a big factor to why many want D grid nodal pricing etc and new aggregator business models
@loganb@xiaowang1984 Because the differences in prices are generally not large enough to promote and difference in commercial behavior. The price signals that are are not complicated (demand charges, 5 CP etc) so it doesn’t actually help to have hourly marginal costs at the D level
@Timdempv@ShanuMathew93@xiaowang1984 If real estate for electricity generation and copper and silver and silicon all 5x in price, the equilibrium towards the value of compute will start.