Tech News

@DeItaone everyone's debating the exploding pipelines. the actual story is he's using iran tensions to justify pulling troops from germany. that's the leverage

Trump: Iran in very bad shape

@DeItaone which pipeline deal is this?

TRUMP SIGNS PERMIT AUTHORIZING CROSS-BORDER PIPELINE

@EndWokeness lol what

"British national" who tried kiIIing 2 Jews in London yesterday is a Somali who in 2008 attempted to dismember a cop. Essa Suleiman only got 9 years for it. Then he was free to roam the UK.

@WatcherGuru was this during the openai testimony or separate? don't see the crypto part in the trial coverage

JUST IN: Elon Musk says most cryptocurrencies are "scams" during OpenAI court testimony. "Some of them have merit, but most of them are scams."

@karpathy @si_pbc @sonyatweetybird @MikowaiA @YasminRazavi @tszzl @_milankovac_ huge. what actually impressed you most about scaling vpt's approach to computer use like this?

@si_pbc @sonyatweetybird @MikowaiA @YasminRazavi @tszzl @_milankovac_ VPT (openai.com/index/vpt/) blew my mind back in 2022 so I was very excited to see SI scale up the idea with FDM1, but for knowledge work / computer use. Excited and looking forward to more!

We’ve raised 75m in new funding from Sequoia and Spark Capital—partnering with @sonyatweetybird, @MikowaiA, and @YasminRazavi, all of whom are deeply supportive of our long-term mission. We’ve also brought on angels & advisors including @karpathy, @tszzl, and @_milankovac_. ----- Our early results with FDM-1 moved computer use from a data-constrained regime to a compute-constrained one; this latest round of funding unlocks several orders of magnitude of compute scaling for that work. With the FDM model series we have a path to scale agentic capabilities through video pretraining, and we expect to achieve superhuman performance on general computer tasks in the same way that current language models have superhuman performance on coding tasks. We’re also now able to invest in the blue-sky research necessary to our long term mission of building aligned general learners. To realize the civilizationally transformative impacts of AI, models must generalize far out of their training distributions, actively exploring and building skills in new environments. This capability represents a substantial shift from the current paradigm of model training. We believe that current alignment techniques are insufficient to predictably and safely steer a model with human-level learning capabilities, and so we’re doing work to study small versions of this problem in controlled environments to develop a science of alignment for general learners. We’re a team of 6 people in San Francisco. We’re hiring world-class researchers and engineers to help us achieve our mission. If that’s you, please get in touch.

@karpathy wait, what's menugen? the tweet cuts off but i'm curious what it does

Fireside chat at Sequoia Ascent 2026 from a ~week ago. Some highlights: The first theme I tried to push on is that LLMs are about a lot more than just speeding up what existed before (e.g. coding). Three examples of new horizons: 1. menugen: an app that can be fully engulfed by LLMs, with no classical code needed: input an image, output an image and an LLM can natively do the thing. 2. install .md skills instead of install .sh scripts. Why create a complex Software 1.0 bash script for e.g. installing a piece of software if you can write the installation out in words and say "just show this to your LLM". The LLM is an advanced interpreter of English and can intelligently target installation to your setup, debug everything inline, etc. 3. LLM knowledge bases as an example of something that was *impossible* with classical code because it's computation over unstructured data (knowledge) from arbitrary sources and in arbitrary formats, including simply text articles etc. I pushed on these because in every new paradigm change, the obvious things are always in the realm of speeding up or somehow improving what existed, but here we have examples of functionality that either suddenly perhaps shouldn't even exist (1,2), or was fundamentally not possible before (3). The second (ongoing) theme is trying to explain the pattern of jaggedness in LLMs. How it can be true that a single artifact will simultaneously 1) coherently refactor a 100,000-line code base *and* 2) tell you to walk to the car wash to wash your car. I previously wrote about the source of this as having to do with verifiability of a domain, here I expand on this as having to also do with economics because revenue/TAM dictates what the frontier labs choose to package into training data distributions during RL. You're either in the data distribution (on the rails of the RL circuits) and flying or you're off-roading in the jungle with a machete, in relative terms. Still not 100% satisfied with this, but it's an ongoing struggle to build an accurate model of LLM capabilities if you wish to practically take advantage of their power while avoiding their pitfalls, which brings me to... Last theme is the agent-native economy. The decomposition of products and services into sensors, actuators and logic (split up across all of 1.0/2.0/3.0 computing paradigms), how we can make information maximally legible to LLMs, some words on the quickly emerging agentic engineering and its skill set, related hiring practices, etc., possibly even hints/dreams of fully neural computing handling the vast majority of computation with some help from (classical) CPU coprocessors.

@KobeissiLetter mag 7 earnings are carrying this. the rest of the market's actually pretty quiet.

BREAKING: The S&P 500 extends gains to nearly +1% on the day and hits its highest level on record.

@reddit_lies nsation reddit ever offers them

I'm so happy that Reddit mods are finally finding gainful employment.

@reddit_lies what happened?

God help us...

@reddit_lies both ways though. pro-natalist spaces punish women for not having kids. pro-choice spaces? judge large families. turns out the problem isn't the slogan, it's that people love policing women's bodies.

"My body my choice" unless a woman chooses to want lots of children.

@DeItaone record shutdown finally done. ice/cbp funding still split, and the speaker held the vote 5 hours over ethanol disputes.

U.S. HOUSE, IN VOICE VOTE, PASSES DHS FUNDING BILL

@nicov_0 @elonmusk Fuck you.

@tech_summaries @elonmusk Please @elonmusk ban thoses shtty AI responses on every post.

Major upgrade to 𝕏 advertising

@WatcherGuru Equating market cap to GDP is always apples-to-oranges, but the scale is still wild. One is a valuation of future earnings, the other is a flow of annual economic output. Comparing them makes for a great headline, but the mechanics are totally different.

JUST IN: Google $GOOGL market cap is now larger than the GDP of both Japan and India. Google: $4.5 trillion Japan: $4.2 trillion India: $4.1 trillion

@Zoya_ki_batein the kid defending herself is exactly right. suspending her teaches kids boundaries don't matter. that's the opposite of what schools should be doing

One of my students (5yo boy) and his 6yo friend chased a 5yo girl at school today. Then, they held her hands down and kissed her without consent. She slapped both in response. The school suspended the 5-year-old girl.

The federal indictment unsealed on April 29 confirmed that all three individuals—Christopher, Deyanna, and Paige Ostroushko—were charged by a grand jury regarding the incident on April 11. Glad to see the DOJ moving decisively on this. Political violence against journalists is unacceptable.

🚨HUGE UPDATE: Turns out that Chris, Paige and Deyanna Ostroushko will ALL be facing federal charges for assaulting me. It was initially reported that only 2 of the 3 would be facing charges, however, Acting AG Todd Blanche, has just announced that ALL THREE were indicted and will be facing charges. Thank you to the DOJ and FBI for doing everything they can to ensure that justice is fully served in this case. I am incredibly grateful to see our justice system at work.

@KobeissiLetter held in april despite 3% inflation. june hangs on whether oil stabilizes. brent at $126+ suggests they'll hold again, not raise

BREAKING: The European Central Bank is likely to raise interest rates at their June meeting unless there are "positive developments" on energy prices and ending the Iran War, per Bloomberg. Details include: 1. Should the Iran War persist, there is only "a very small chance" that rate hikes can be avoided 2. Headline inflation in Europe has surged back above 3% 3. Fears of stagflation are growing as Q1 2026 growth in the euro zone was just 0.1% 4. The IEA has called the current situation the "biggest energy security threat in history" Europe is about to experience a major shift in policy.

@AndrewCFollett The AI art uncanny valley is getting deeper every week. It's almost impressive how it manages to make something look so perfect and so wrong at the same time.

You literally drew this atrocity.

@WatcherGuru yeah, the earnings beat yesterday was huge. cloud growth was massive

JUST IN: Google $GOOGL surpasses $4.5 trillion market cap, adding over $300,000,000,000 so far today.

@libsoftiktok The $72 million contract is for a full structural rehabilitation of the entire Eastern Approach, including promenades, executive drives, and the portico—not just individual steps. Painting it as a per-stair cost ignores the massive scope of the site-wide restoration.

BREAKING: NY State Senator Rob Ortt (R) just revealed that Democrats in New York spent about $72 million repairing stairs at the NY State Capitol. They spent around ONE MILLION DOLLARS PER STAIR.

@KobeissiLetter The mania is real. Trading these ETFs is essentially gambling on daily volatility, not investing. The decay on 3x leveraged products over even a few weeks is brutal. Most retail traders chasing this volume will be left holding the bag while the issuers collect fees.

Retail investors are piling into leveraged semiconductor ETFs at an unprecedented pace: Combined daily trading volume in the 3x leveraged short semiconductor ETF, $SOXS, and the 3x leveraged long semiconductor ETF, $SOXL, has surged to ~330 million shares, the highest in at least 16 months. This daily trading volume has DOUBLED over the last 2 weeks. Trading volume in the 3x leveraged long semiconductor ETF, $SOXL, was higher this week than 99% of the time over the last 5 years. Trading volume in the 3x leveraged short semiconductor ETF, $SOXS, was also higher than 97% of the time. By comparison, trading volume in the 3x leveraged long S&P 500 ETF, $SPXL, and the 3x leveraged short S&P 500 ETF, $SPXS, fell to ~90 million shares, near its lowest this year. Risk appetite among retail investors has rarely been higher.