ohong

TBPN was also reportedly on track to do $30M in sponsorship revenue this year, up from $5M last year. Audience size is modest but targeted & high value per viewer (mostly decision makers in tech).

exactly! haha you kinda predicted my next project: startups.rip

that's a great find! they were ~10 years too early...

oops, should be fixed

fixed!

good one! will add

could you try now?

I love HackerNews! This is more for builders who might otherwise be discouraged by rough early feedback.

I mean… that’s like making a list of best playwrights and not including Shakespeare

i’d say more projects that i wish existed + closing the taste gap btw the bar i have for products as a user vs. what i’m capable of creating as a builder. something like “token generated” is ofc only a proxy goal with no guarantees of quality. nevertheless it might be useful to some, like how writers set daily word count quotas for themselves

right now it just pulls from Claude Code's local data files via ccusage (https://ccusage.com/guide/#data-sources)

i think of straude as serving a similar role as habit trackers and github's contribution graph. some of us have a goal to use claude code daily, like how others might run or play the piano daily. it can be encouraging to do it with friends' support, see your progress visualised, etc. diff strokes for diff folks!

thx! good point. i once heard the cto of supabase say that they didn't hit pmf until they added the copy "open-source firebase alternative" to their website copy & overall product positioning. that was when it finally clicked for devs.

that said, if we're so fortunate to get enough users for this become a problem, i won't hesitate to take it off :)

thx for the great feedback!

great libraries already exist for usage tracking (like ccusage), think of this as just a social layer on top so you can follow other ppl's claude code sessions.

agree landing on global feed vs. "following" makes it more clear, will make that change now.

re: gaming the system, the cli tool pulls from local data usage data files to "verify". while this might be possible to spoof, it'd be a lot of work for not very much gain :)

It's on web! As long as your agent can hit API endpoints it should work.

Hey I rly like your project! And kinda amused by the comments here... Would love to chat more about it if you're down. My contact's in my profile :)

OpenAI committing to $10bn in orders of the new “XPUs”, to be used internally.

Hey, author of the post here. Thanks for the feedback! I should clarify that many of the stories came from the book Software Engineering at Google (I link to it at the end). Deep Research helped me glance through the bios of the people involved, which I then manually verified to the best of my ability before including in the published version. Please let me know if you find any errors and I’m glad to make corrections.

As a matter of style, I believe that one day soon, prefacing an online essay as being “AI-assisted” may read as anachronistic as something like “I wrote the following using a word processor, spell check, the internet, and Wikipedia.”

I totally grant you that day might not be today :)

That’s right, thanks for the clarification! Had to shorten to fit char limit. There were 2 best paper awards at NeurIPS.

https://archive.ph/BqEG5

Also announced at re:Invent that Apple is using their Trainium2 chips for AI inference workloads.

This is a very good list! Curious if all of them just came top-of-mind (if so, very impressive haha) or if you did any research / AI assist. Thank you!

Announced during Nvidia's presentation at this year's RISC-V Summit.

I haven't used it myself, but I've heard good things about https://www.usestable.com/ which sounds similar!

1/ Hsinchu, Taiwan for semiconductor manufacturing

2/ Denver, Colorado for fast casual restaurant chains: Chipotle, Smashburger, Quiznos et al.

3/ I'd also put forth Vancouver, Canada for fast fashion, specifically the "athleisure" look: Lululemon, Aritzia, Reigning Champ, Arc'teryx, Indochino

To start, I really liked the idea behind Muse, but they couldn't find PMF: https://adamwiggins.com/muse-retrospective/

And more recently, buildspace :(

Nothing but respect for all the hackers & founders who tried to put something new out there

Here's a good explainer on the need for specialised hardware for zk-proof & encryption algos: https://www.paradigm.xyz/2022/04/zk-hardware#hardware-matter...

I've met a few others in the same boat (SWE exploring hardware) and would love to chat sometime to bounce ideas if you're up for it. Feel free to dm me!

This article discusses the necessity & benefits of direct-to-chip cooling vs. current approaches (namely, cooling at the data centre level).

This is awesome! Super exciting to see more software-level composability & reuse making its way into the world of chip design. Long overdue.

I'm curious how you guys think about growing the corpus of open source data for training smarter chip design copilots, given the limited amount of publicly available Verilog code vs. what it took to train GH Copilot to work well for the most popular programming languages.