CS + Econ @stanford @stanfordailab | Prev @generalatlantic @contextsuite @lux_capital
Stanford, CA
Joined February 2024
- Tweets598
- Following729
- Followers922
- Likes1.2K
Dario Soatto retweeted
Today, @JudgmentLabs opens for general access.
We believe long running agent systems will generate more exhaust data than the world can currently process.
Hidden within agent traces are rich records of how to make these systems better: what users ask for, when agents stumble, and how agents interact with the external world.
Judgment is the most performant monitoring and improvement layer for long-horizon agents.
We’re excited to see what you build!
Dario Soatto retweeted
Remember when Ilhan Omar deleted that 2013 tweet calling her father Nur Said?
That would make her name Ilhan Nur Said, and her ex-husband was Ahmed Nur Said, which means she did in fact marry her brother.
Fast forward to today, and the federal search warrant lists her as “Ilhan Nur Said Elmi, also known as Ilhan Abdullahi Omar.”
OOF! She is soooo COOKED! 😬
Dario Soatto retweeted
Introducing Catalyst.
What do you believe? How would you trade on it?
You notice data centers are popping-up everywhere, gold will skyrocket, or just know that Labubus will go out of fashion.
Catalyst researches your ideas, builds the optimal strategy, monitors events, and executes trades. Autonomously.
Dario Soatto retweeted
in my 2nd week at @cognition, i rebuilt devin's Command K (⌘K) palette from scratch with @ZreikJorge!
i designed it as a do-it-all menu. i strongly believe you should be able to type in any concept (whether it be a settings toggle, an old wayward session, or even a quick prompt idea) and have it in front of you in 1-2 keystrokes.
as software evolves, we shouldn't be thumbing through dashboards, menus, and buttons. the onus can & should be on the app to deliver you what you want. this will be especially important as we move to new semantic-focused input methods, like voice or brain-computer interfaces.
beautifully made ⌘K palettes exist in other apps like linear, notion, retool, claude, and more. many, like ours, are based on the cmdk library. developing the devin webapp's ⌘K palette has been one of my favorite design and product engineering projects - it's now the universal search surface for the app! ⭐️
Dario Soatto retweeted
This.
But here's some good news: As late as the early-2020s, if you even mentioned the race IQ gaps to a prominent lefty account on Twitter, they would deny that they exist. (And then of course they would typically block you).
But this doesn't happen very often here anymore because the subject has been openly discussed without censorship for several years now, and it's increasingly difficult for even lefties to avoid seeing (probably for the first time in their lives) the scientific information on the subject.
In 2023, Will Stancil started an online mob and attempted to dox me because I had tweeted about the US black-white IQ gap. I didn't render an opinion at all about whether genetics were involved. I merely stated that the disparity exists. That was enough for Stancil to call me a Nazi and a white supremacist and get his followers all riled up and in my replies.
But a few weeks later, Stancil was openly acknowledging the existence of the gap to his followers (while of course arguing that it was due to racism). I'm not sure what caused the change, but I suspect that someone he trusted had probably tapped him on the shoulder, and said, "Listen, man, you're embarrassing yourself."
Dario Soatto retweeted
I once heard @vkhosla say he instigated fusion in the U.S. If there were one person I would credit with instigating @AntaresNuclear it would be @pratyushbuddiga. A meeting in a blue bottle with him eventually led to a meeting in the Oval Office with the President!
ANTARES RAISES $470M SERIES C
Three and a half years ago, I sat down in the Blue Bottle on Fillmore St. I thought I was catching up with a friend to talk about fatherhood. We'd both just become dads. It prompted a deep question for me: how do you be a great dad while also being great at what you do?
History is littered with examples of great men who leave a strew of destroyed relationships and children in their wake. I'd written on my Substack about what I thought it may take to try to leave a different legacy. It resonated with a friend so he reached out to catch up.
That person was @jordanbramble.
Becoming a father had changed him in a deep way and he knew he wanted to build something deeply impactful, something that would change the trajectory of human civilization, but more importantly, something his son would be proud of.
He wanted to build nuclear reactors.
Coincidentally, I'd spent the previous year reading about nuclear energy and our lost Atomic Age. The consensus then was that we'd never build nuclear again in America.
But Jordan thought differently. As we jammed on the idea in the Blue Bottle, a possibility emerged. We could go to market through the government partners who needed Strategic Energy, the kind only nuclear can provide. We couldn't have foresaw the change in public sentiment, the executive orders after Trump was elected, or any of the other amazing tailwinds that have popped up since then. But we did see a small, critical path of possibility.
So we decided to partner up: first investor, inception round, no deck, nothing but a Google Doc that was put together after that meeting.
Sitting here a few years later, it's amazing to see all that the company has accomplished and the new Golden Age of Energy that has been unleashed. Criticality today, deployed electricity tomorrow.
Today, they announce a $470m Series C led by Paradigm and Caffeinated. It's been a wild ride, but there's so much more to go.
What makes me happiest is that when we caught up again recently in Idaho when the Secretary of Energy was visiting INL, we talked for a long time about our kids and families. It's hard, no doubt, but we're both still trying to be the best husbands, fathers, and men we can be.
We have a lot more work to do, but today is a good day. Abundant Energy from Earth to the Asteroid Belt.
Ad astra. LET's GOOOOOO
There's a new version of this post
Dario Soatto retweeted
⛳️ We built a $40,000 TrackMan golf shot tracker with a Raspberry Pi and a ~$100 ti iwr6843sk radar.
🏌️♂️More details below!
Dario Soatto retweeted
How to ragebait a VC
- Which podcast gave you that opinion?
- How would you add value if you found a way to be valuable?
- You look like you make all of the 2 but none of the 20.
- Sorry, we aren't raising from you right now.
- Which AI model did you use to vibe code your firm's website?
- What startup role would you take if they'd have you?
Dario Soatto retweeted
I broke my cofounder's arm to prove a point: you don't need a keyboard.
Meet @WillowVoiceAI Scribe, our voice AI assistant. It clears your email, Slack, and docs in seconds. We'll pay you $10k if you still need to type after trying it.
Don't believe? Try Scribe below and win.
Dario Soatto retweeted
We’re launching @JudgmentLabs today and announcing $32M in funding.
As AI agents take on more of the work that creates economic value, they generate massive amounts of production data: the clearest record of how they behave with users, software, and the real world.
Judgment builds infrastructure for improving AI agents from production data.
Dario Soatto retweeted
The AI Ecosystem Transaction site is live.
On it, you can comb through 265 deals totaling $2.6+ trillion in disclosed volume – spanning across hyperscalers, neoclouds, frontier labs, chip designers, server OEMs, and more.
Demo is below. Try it now at compute.jstwng.com.
Dario Soatto retweeted
Low-quality 4-bit attention has been blocking end-to-end FP4 serving, and we take a major step toward fixing that 🔥
In Attn-QAT, we made attention compatible with quantization-aware training.
This gives us FP4 attention with quality on par with BF16 attention for both language and video models while being 1.1x–1.5x faster than SageAttention3 on an RTX 5090, and up to 1.39x faster than FlashAttention-4 on a B200.
Blog: haoailab.com/blogs/attn-qat/
(1/5) FP4 hardware is here, but 4-bit attention still kills model quality, blocking true end-to-end FP4 serving.
To fix that, we propose Attn-QAT, the first systematic study of quantization-aware training for attention.
The result: FP4 attention quality is comparable to BF16 attention with 1.1x–1.5x higher throughput than SageAttention3 on an RTX 5090 and 1.39x speedup over FlashAttention-4 on a B200.
Blog: haoailab.com/blogs/attn-qat/
Code: github.com/hao-ai-lab/FastVi…
Checkpoints: huggingface.co/FastVideo/14B…
Dario Soatto retweeted
Dreamverse is our AI video engine that generates a video faster than you can watch it.
30s of 1080p video in 4.5 seconds. One GPU. Real-time editing.
This is vibe-directing.
dreamverse.fastvideo.org
(1/N) We're launching Dreamverse. Most AI video models take minutes to generate a 5 s 1080p clip. In 4.5 seconds, we can generate 30 s 1080p clips on a single GPU.
Our videos generate faster than you can watch them: stop waiting on prompts and start directing scenes live.
🕹️Demo: dreamverse.fastvideo.org
📑 Blog: haoailab.com/blogs/dreamvers…
Welcome to the era of vibe-directing 👇
Dario Soatto retweeted
⚡️We built a new real-time inference stack in FastVideo and have the fastest 1080p TI2AV (text + image to audio and video) pipeline ever.
Create a 5 s 1080p video with audio in ~4.55 s on a single GPU!
High-quality video generation must be fast to be truly interactive. The only limit in creative workflows should be your imagination.
If you have the need for speed (and quality), make video generation go blurrr (for free) at 1080p.fastvideo.org/ and create whatever you can imagine…
(1/N) Content creators have been stuck with costly and slow video generation APIs for far too long. We couldn’t take it anymore.😅😭
FastVideo’s new real-time inference stack has the fastest 1080p TI2AV pipeline ever.😍🚀🚀
Our optimized LTX-2.3 pipeline creates 5-second 1080p videos with audio in 4.55 s, on a single GPU!
3.9x faster than the next fastest option.
🕹️Live demo: 1080p.fastvideo.org/
📜Blog: haoailab.com/blogs/fastvideo…
Dario Soatto retweeted
wasian hall of fame:
keanu reeves
alexa chung
mitski
laufey
devon aoki
bruno mars
mike shinoda
olivia rodrigo
charli xcx
nancy kwan
eileen gu
alysa liu
ben thompson
Dario Soatto retweeted
In 2006, I was in my mid-20s, had wound down my company and was doing a postdoctoral scholarship to pay my bills. I had been told that I should raise venture capital for my next startup.
I googled “nanotechnology” and “venture capital,” and the first Google search result was Lux, based in NYC, founded by a couple guys my age, who had funded companies building next-gen lithography, batteries, displays, and longevity.
I was FLOORED. This is at a time when SoLoMo, ad-tech, and “cloud” were all the rage. I didn’t know VCs existed in NYC, or below the age of 40.
I reached out to Pete and Josh offering to help source and diligence asking for nothing but friendship/mentorship in return, as I didn’t even have an idea at the time, and they replied the same day inviting me to spend time with them. No other VC had given such a warm reply to my blind outreach.
Being flat broke, I submitted papers to land speaking slots at conferences in NYC to be able to go out to see them, and the rest is history.
20 years in, I’ve had the fortune of working with talent and being part of the inception and success of amazing companies in fields investors were shunning or simply not paying attention to. I’ve learned so much from working closely with incredible people who have brought different perspectives to the firm.
In 2006, deep tech felt lonely; today the talent is everywhere, but conviction still matters.
It’s a privilege to be part of this next chapter with our fresh $1.5B fund to continue working with founders as they are ideating to pulling together to launching to growing to establishing generational companies accelerating our future with feats of science and engineering.
We started Lux in 2000 with a simple conviction: the biggest opportunities lie at the frontier of science and technology that others find too hard, too early, or too confusing. Twenty-five years later, that conviction hasn't changed. What has changed is that the world has caught on.
Now compute, automation, and biology have matured into scalable engineering disciplines, and what was once fringe "deep tech" has become essential infrastructure. New talent, new tools, and new industrial capacity now exist to turn the most cutting-edge technology across the physical, computational, and life sciences into enduring businesses.
For these businesses and founders to become category-defining winners, they need meaningful capital and conviction. That is why we are especially excited to announce our largest fund to date, Lux Ventures IX, a $1.5B commitment to investing in the people turning sci-fi into sci-fact.
Today entire sectors of aerospace, biotech, defense, industrials, transportation and beyond are being reinvented by a new generation of brilliantly ambitious, often irreverent, scientists and engineers. And while markets have changed and capital reshapes around fewer companies and larger checks, our focus on the craft remains constant.
That focus matters because science doesn’t scale itself. The path from a result in the lab, only known to a few, to something durable the entire world will use requires a different kind of partnership. One that deploys the first $100K and the last $100M, that stays through the full arc of company-building, that mobilizes a powerful network of kinetic like-minded founders, and that treats capital as a tool rather than the product.
Forty-four people at Lux now manage $7B, still operating with the same discipline and purpose we had on day one. We find and fund founders others have overlooked, back conviction over consensus, and build alongside these rebels of science and tech at every stage.
We’re grateful to our valued LPs who've partnered with us through multiple cycles, and to the founders who trust us with their life's work. The mission continues, with even greater intensity and momentum.
luxcapital.com/news/announci…
Dario Soatto retweeted
2025: Train on SWE-bench (single tasks)
2026: Train on CodeClash (long horizon, open ended objectives)
Please check out our work!
