frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Jeff Dean leaving Alphabet

https://www.nytimes.com/2026/08/05/technology/google-researchers-ai-startup.html
152•louiereederson•1h ago

Comments

drcongo•50m ago
Not through conscience.
Handy-Man•46m ago
Jeff and Sanjay both leaving!
tough•42m ago
Oh wow demis and Jeff, that Ethics engineer who left GDM, and reached to both and posted on lesswrong a couple weeks [1] ago did make a difference?

1. https://www.lesswrong.com/posts/iKm2FhpWkuuBojm82/why-i-left...

bkishan•41m ago
what was the post?
tough•40m ago
https://www.lesswrong.com/posts/iKm2FhpWkuuBojm82/why-i-left...
TulliusCicero•31m ago
...you think they've only been planning this for a couple weeks?
tough•29m ago
hmm not really, and it probably isnt the origin, but im glad to see him vindicated by having both C-level google leadership people who targeted to try and either stop google from doing the pentagon deal, both stepping down the same day.

At least it makes me feel like he reaching out might have made a difference on remembering this people killer robots are bad, and being part of it might not be on the best of their interests to go down on history for

cma•22m ago
Alex Turner discusses a bit here at this timestamp:

https://www.youtube.com/watch?v=pGlJQHVeKIA&t=14m27s

mitchbob•37m ago
https://archive.ph/Pogl7
ValentineC•35m ago
I wonder if Jeff Dean facts [1] (I hope people remember the reference) will carry over to the new startup.

[1] https://github.com/LRitzdorf/TheJeffDeanFacts

hoyd•8m ago
«Jeff Dean's PIN is the last 4 digits of pi.»

I had not read this before, but told many students the same about my PIN code and I a quiz about the last digits. Love it.

segmondy•33m ago
End of an era for Alphabet.
ExMachina73•29m ago
Just remember to make sure you turn in your devices.
calufa•26m ago
As LLM coding agents plateau— at least for the average engineer without tens of thousands of dollars or swarms of agents to run —I’d say that, from here on it’s going to be about ASICs, specialized LoRA/or-equivalent models, or a Ruby on Rails for LLM context engineering and orchestration, which LangChain and others seems well position, including Google as they own the entire stack. LLM free lunch has been over for a while, perhaps since the ReAct loop, and has been official since Ilya mentioned it at NeurIPS.

I feel the most exciting development these days is self-evolving agents. Especially if you have a way to verify their outputs with a formal system, or with a system developed since the 60s by armies of PhDs.

DeepMinds Gnome is a good example, where they use DFT to verify outputs. Approximating NP-problems is always fun for those who dare.

I am also building in this space. Its a mix between HPC, AI, and hard science. Pretty fun compared to waking everyday to LLM news that seem more like marketing stunts.

deeviant•20m ago
LLM coding isn't even close to plateauing. Right now, the major players are in a consolidation step, focusing more on economic efficiency but still not at the point where we're ready to start burning models to hardware and freezing the line.

They are straddling the line between pushing it forward, and justifying the business case. It's hard to do both at the same time.

calufa•5m ago
LLMs are hungry for tokens. Every day I see more startups claiming token usage at 50-100k per month. You can always brute-force your way in - just see HuggingFace's recent attacks.

If throwing more money at inference while accumulating compounding technical debt is the new norm, then we are not solving the problem, and the solution space is already covered.

Perhaps there are marginal gains at the expense of quadrillion-params LLM models with 10x the cost and energy. We are simply making inefficiency more expensive, camouflaged by VC money and great marketing.

If that is not plateauing, then I guess I will have to reconsider what plateauing means.

flakiness•26m ago
To be honest, this feels more like a lifestyle business (aka hobby) than a startup. They truly deserve it, but I don't expect a huge success as a business.

That said, I hope they write cool papers with various peers across the industry without worrying too much about the competing dynamics. That'd be a blessing for humanity, and good for their spirit.

ChrisArchitect•25m ago
Related:

The next chapter of our AI momentum

https://news.ycombinator.com/item?id=49184755

quaintdev•20m ago
I wonder if this has anything to do with it

https://www.lesswrong.com/posts/iKm2FhpWkuuBojm82/why-i-left...

next_xibalba•11m ago
Doubtful. His destination is both funded by Google and has a compute deal with Google.
montebicyclelo•13m ago
Did the ycombinator podcast which included giving advice to startup founders just a few days ago:

https://www.ycombinator.com/library/Vy-jeff-dean-the-1-rule-...

xnx•7m ago
As always, a very good presentation.
ndr•8m ago
Imagine being a VC and getting to this slide on their pitch deck, "involved in creating the following":

https://x.com/JeffDean/status/2085036253263921218

xnx•7m ago
I've seen tiny tiny hints from the outside that Jeff Dean was dealing with too much internal BS. Two examples that come to mind: Having to deal with Timnit Gebru fiasco, and even chips in the TPU series getting marketing names (Trillium and Ironwood) before switching back to more standard numbering.
compiler-guy•2m ago
I have no doubt that internal Google friction is one of the reasons they are moving. But the Gebru incident was almost seven years ago now. It is very unlikely to be a proximate cause.

I doubt numbering vs names on TPU releases even crosses Jeff's radar. It's not the kind of thing he cares about.

Balinares•7m ago
Well, damn. Jeff, Sanjay and Demis leaving all at once, it's hard not to read that as someone fucked up pretty bad.
sva_•6m ago
Hassabis isn't leaving
asimpletune•3m ago
They're structuring the new company as public benefit corporation.
zuzululu•13m ago
we definitely haven't hit plateau yet. I think a lot of people latch on to anti-LLM narratives without really thinking things through.
aoeusnth1•9m ago
Sounds like you get your news from 2024 when people thought things were plateauing after GPT4?

New 16T parameter model with a 4B token context window released

https://huggingface.co/tsfrm/vacuum-16t
2•jedberg•1m ago•0 comments

New platform to predict human medical outcomes [pdf]

https://oliolabs.com/RoseberryKrauszWilliamsTingley_2026.pdf
1•appliedbio•1m ago•0 comments

Coinage – every dollar invested until the second you spend it

https://www.coinage.app
2•nandhd•1m ago•0 comments

AI Security Leaderboard: Methodology, Results and Minimal Standard

https://arxiv.org/abs/2608.03070
1•sbulaev•1m ago•0 comments

Long Range Career Advice

https://twitter.com/spakhm/status/2085040887474667865
1•tosh•2m ago•0 comments

Proxmox officially supports Arm, with some caveats

https://www.jeffgeerling.com/blog/2026/proxmox-ve-arm-official/
1•mikece•2m ago•0 comments

iCloud Private Relay leaks real IP addresses

https://proton.me/blog/icloud-private-relay-ip-leak
1•mikece•2m ago•0 comments

Byline, a web app and CLI for finding and backing up article portfolios

https://www.npmjs.com/package/serpapi-byline
1•corbindavenport•3m ago•0 comments

Show HN: I was wrong about scrollytelling landing pages

1•vaitko•3m ago•0 comments

AI-Enhanced Photos of New Office, Faces Backlash

https://www.news18.com/viral/bengaluru-founder-shares-ai-enhanced-photos-of-new-office-faces-back...
1•bitlad•3m ago•0 comments

Show HN: Agents keep running with the lid closed

https://agentsneversleep.app/
1•alekseyrozh•4m ago•1 comments

Robinhood lets your invest in YC startups now

https://techcrunch.com/2026/08/05/robinhood-to-list-a-fund-that-lets-anyone-back-y-combinator-sta...
1•debarshri•6m ago•0 comments

The Download: NASA's new telescope and Chinese tech import curbs

https://www.technologyreview.com/2026/08/05/1141212/the-download-nasa-telescope-chinese-tech-impo...
2•joozio•6m ago•0 comments

Elon Musk's SpaceX unnerves investors with lavish AI spending plans

https://www.ft.com/content/41f7963b-dd50-4f27-a085-771ddec4a8ca
2•mapping365•7m ago•0 comments

Jeff Dean

https://en.wikipedia.org/wiki/Jeff_Dean
1•xnx•7m ago•0 comments

Grocery prices are Americans' top affordability challenge, report finds

https://www.cnbc.com/2026/08/05/affordability-grocery-prices.html
1•mapping365•8m ago•0 comments

Towards scalable AI-assisted pre-bunking of election misinformation

https://royalsocietypublishing.org/rsos/article/13/8/252226/482708/Towards-scalable-AI-assisted-p...
2•Anon84•8m ago•0 comments

Senators demand crackdown on wildfire "prediction market" bets

https://arstechnica.com/tech-policy/2026/08/senators-demand-crackdown-on-wildfire-prediction-mark...
2•cdrnsf•8m ago•0 comments

Show HN: ExANS – Lossless KV cache compression at 622 GB/s on H100

https://www.theopenlake.com/blog/exans-lossless-gpu-compression-for-bf16-kv-cache
8•arnav__1•9m ago•0 comments

Show HN: Webklip – ephemeral online clipboard with live sync and E2E encryption

https://webklip.com/
2•vsergione•9m ago•0 comments

Desktop Linux just cracked 10% market share – and Windows 11 is mostly to blame

https://www.zdnet.com/article/the-linux-desktop-finally-cracks-the-10-market-share-barrier/
2•CrankyBear•10m ago•1 comments

I'm Scared a Stranger Will Call My Novel AI, So I Built GitHub for Words

https://dylan.blog/2026/08/03/im-scared-a-stranger-will.html
1•harper•11m ago•0 comments

Your model already knows the answer: how benchmark answers leak into LLMs

https://elman.ai/news/your-model-already-knows-the-answer/
4•fran-mora•12m ago•0 comments

Reactivating TensorBoard: Why treat an oscilloscope like a photo frame?

https://github.com/PastToFuture-Whisperer/xprof-cubism-reducer/discussions/1
1•PTF-W•13m ago•0 comments

Sash: Keeping secrets away from software that acts on your behalf

https://zenodo.org/records/21797233
1•cmt-t•13m ago•0 comments

PBCareerOS – Multi-dimensional career intelligence and roadmap generator

https://customizecareermap.org/
2•Pballada80•15m ago•0 comments

Why everything breaks at 150 people

https://falconer.com/notes/cerebras-knowledge-base/
7•iamgioh•16m ago•1 comments

Show HN: T – Conductor, but for Your CLI

https://github.com/PiemonteF/t
2•Absonsonson•17m ago•0 comments

My NixOS Dev Machine

https://miccah.io/posts/2026-08-01-my-nixos-dev-machine/
2•speckx•17m ago•0 comments

Celld: Self-hosted, distributed Durable Objects

https://github.com/denoland/celld
3•calvinfo•19m ago•0 comments