←
Darrell Etherington15,478 posts
Darrell Etherington avatar

Darrell Etherington

@etherington

VP @SBS_Comms, ex-OMERS Venture, ex-Managing Editor at TechCrunch. Writer of @theangle, host of @originalcontentpodcast

◷ Joined July 2008▧ 453 Photos and videos
Darrell Etherington avatar
Darrell Etherington@etherington·Apr 10

New from me at @theangle: HumanX this year captured a theme that I'm seeing in my own experience of AI, and that I think reframes how we should be thinking about its impact on jobs and enterprise organization design:

Darrell Etherington @etherington Apr 10 Article The Great Reabsorption At HumanX this past week, I watched eight different tech leaders – including the CEO of AWS, the co-founder of Cursor, and the head of Anthropic Labs – describe essentially the same phenomenon without

Walden avatar
⟳ Darrell Etherington retweeted
Walden@walden_yan·20h

I’ll be joining MTSLive at 12:30pm PT today to deep dive Devin Bots! We’ve been seeing eng teams use Devin Bots to automate work to a completely new level

JENSEN BACK? | ISOMORPHIC RAISES 5B | NSA OUTAGE x.com/i/broadcasts/2rJQYgYay…

Mike Butcher, Editor, Pathfounders.com avatar
⟳ Darrell Etherington retweeted
Mike Butcher, Editor, Pathfounders.com@mikebutcher·Nov 16

*Interview: Plural confirms the rumours of a new €1bn fund to go deep into Europe's frontier tech* The fund was said to be raising "up to” €1 billion earlier this year. Themes for the new fund will include defence, energy and AI infrastructure. Partner Pierre-Dimitri Gore-Coty spoke to Parthfounders about the strategy. Read more: pathfounders.com/p/interview… P.S. We’re now days away from launching *Pathfounders Network*, the new ‘LinkedIn for European tech”. Apply here to get the Founding Member offer: pathfounders.com/network

will depue avatar
⟳ Darrell Etherington retweeted
will depue@willdepue·Nov 12

I have a major result to share. I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China. This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model. Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves. I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems. I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used. It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20. These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs. I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at unemploymentlabs.com/checkpo…. There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.

Post attachment
Panzer avatar
⟳ Darrell Etherington retweeted
Panzer@panzer·Nov 12

Happy to say that I've joined @usehercules as Head of Content Engineering. I love what @BrendanFalk and @gurgrant and the rest of the sweet team here are creating. The future of business software is very clearly a marriage of strong foundations and complete customizability. When I decided to go into tech journalism I didn't have a degree, I didn't have a background in writing and I had no connections to the tech world at all. What I did have was the Internet -- the great democratizer that let my words sit on the same network as the great writers I admired. I knew I could write, but I also knew no one would hire me without evidence, so I just started my own site. I wrote over a dozen articles a day for two years until what I was producing was finally readable. Then I got a job doing it and was able to work with and learn from incredible coworkers, mentors and subjects. at @thenextweb and @TechCrunch where I was writing about and talking to some of the best product people on the planet. Even though I've founded several small businesses over the years I grew to admire the people who had stepped off the ledge to try to build something genuinely new that could affect hundreds of thousands or even millions of people. I wanted to build. But I knew I faced the same problem - I was 'the content guy' and I didn't know how to code. But I also knew I came to it at the right time to have the same advantage as I had 15 years ago. It turns out that the skill sets that make you good at journalism like clear articulation, semantic precision, a systems-oriented mind, and above all, a genuine curiosity, also work well for coding agents. So I spent the last few years retooling myself. I now have code in production, tools in use and am filling PRs as we speak and I couldn't be happier. I'm hooked on the vision of Hercules - of bringing that same enabling power to business owners who know their world better than any SaaS provider or consultant, if they could only make it real. Let's go!

Post attachment
John Gedmark avatar
⟳ Darrell Etherington retweeted
John Gedmark@Gedmark·Nov 6

Excited to announce that @NASA has selected Astranis for the first phase of NEXUS, a new program to replace the aging GEO satellites that provide mission-critical comms links for some of NASA's most important space assets. That includes the International Space Station, Artemis, the Hubble Telescope, and NASA's Earth science fleet. NEXUS will soon require an entire constellation of highly capable spacecraft, as the existing fleet of large GEO relay satellites reaches end of life and are retired. If you didn't know NASA maintains its own fleet of massive GEO communications satellites I wouldn't blame you-- it's one of those pieces of high orbit infrastructure that has been quietly supporting everything NASA does for decades, but very much behind the scenes. These satellites are necessary because important space assets like the International Space Station are in LEO, and don't pass over ground stations nearly as often as their operators would like. So the ISS and these other important NASA assets end up relaying most of their data through a constellation of large, expensive GEO satellites called the Tracking and Data Relay Satellites, or TDRS, keeping them connected 24/7. NASA awarded Astranis a contract with a maximum potential value of $264.3 million, in a firm-fixed-price contract that gives NASA the option to provide additional funding up to the award limit to advance from initial concept development and risk reduction to an in-orbit flight demonstration. And that's just the beginning. We estimate NASA deployed the existing TDRS constellation at a cost of $13 billion-$15 billion in today's dollars. By working with a company like Astranis, they can replace that capability at a fraction of the cost, freeing up crucial resources for NASA's highest priority future missions, returning astronauts to the Moon and building the Moon Base. Special thanks to our partners at NASA for their hard work on this program and for making this happen in record time. We're super excited about this mission and providing a capability that NASA, and our nation, badly needs. Onward to the next phase. 🇺🇸🇺🇸🇺🇸

Post attachmentPost attachment
John Gedmark avatar
⟳ Darrell Etherington retweeted
John Gedmark@Gedmark·Nov 5

Incredible being hosted by NASA and @NASAAdmin at KSC for the UP Summit. Thank you for the opportunity to talk about the future of space. Next looking forward to MAX POWER 🇺🇸🇺🇸🇺🇸🇺🇸

Post attachmentPost attachmentPost attachment
Brydon Parker avatar
⟳ Darrell Etherington retweeted
Brydon Parker@parker_brydon·Nov 2

waymo in toronto

0:03
tobi lutke avatar
⟳ Darrell Etherington retweeted
tobi lutke@tobi·Nov 2

Yea. this is exactly as crazy as it seems.

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…

Post attachment
Alexander Whedon avatar
⟳ Darrell Etherington retweeted
Alexander Whedon@alex_whedon·Nov 3

We are excited to host you, alongside @GoogleDeepMind and @Meta, where you will be able to use early access to some of our unreleased models to compete in this hackathon!

Hey San Francisco.  Let's build! This Saturday, join us at the AI Engineer Code Summit Hackathon! Registration is open.

Sean O'Kane avatar
⟳ Darrell Etherington retweeted
Sean O'Kane@sokane1·Oct 28

Some personal news: in two weeks I'm joining The @WSJ to cover @Tesla, one of the most incredible (and I mean that in just about every way possible) companies of our time. It's a dream job, tbh, and I genuinely can't wait to get started. I'll also be covering the actually-really-happening-now rise and spread of autonomous vehicles, which I expect will be one of the most fascinating technology, business, and labor stories that I'm likely to live through. I can't wait to go deeper on companies like @Waymo and @Rivian, and see how this all shakes out in the coming years. If you work/have worked at Tesla or one of these companies, or work in the industry more broadly, *or* just have a fun story to share, I want to talk to you! you can always reach me on Signal: okane.01 🫡

Ethan Mollick avatar
⟳ Darrell Etherington retweeted
Ethan Mollick@emollick·Oct 27

Genie 4 is quite an insane experience, generating game-like enviornments as you go. It isn't flawless but I can type "Neogothic brutalist city inhabited by robots and horses" & it creates one in a minute. Then I asked it to add dragons. Then a friendly otter NPC. All in realtime

0:53
John Gedmark avatar
⟳ Darrell Etherington retweeted
John Gedmark@Gedmark·Oct 26

Today we're excited to announce we have upgraded our 4th launch with our partners at @SpaceX to a dedicated mission, launching in 2028. Given increased customer demand --in particular from the sovereign and air-gapped network verticals-- we decided to increase the number of satellites we're launching to 6. This is the entire production run of our spacecraft Block 4. Sovereign satellites, and specifically sovereign satellites in geosynchronous orbit, allow a country or enterprise to have their own dedicated comms satellite with a purpose built network deployed around it. In this new world of cyber threats and super-intelligent agents in the hands of malicious actors, we're seeing more and more the desire to pull infrastructure off the public internet entirely with dedicated networks. It's the most effective way to shrink the attack surface. And it's where the world is headed. As a result we've seen huge increase in demand for satellites that our customers can control and lock-down from a cybersecurity perspective. Something uniquely possible on a nation-state scale with a geosynchronous satellite. We are pumped about this launch. And the best part? We’re going straight to 22,000 miles up, with a direct-inject to GEO using an Impulse Helios upper stage. We are now seeing an insane amount of commercial and US government demand, and we’re only just getting started.

Post attachmentPost attachment
John Gedmark avatar
⟳ Darrell Etherington retweeted
John Gedmark@Gedmark·Oct 23

The cyber incidents of the last 48 hours have been pretty mind blowing. I’ve been trying to piece together the “why now” question, and I think what we’re seeing is the first of a new kind of super-agent driven vulnerability cascade-- first one vulnerability is found by a malicious actor, then other bad actors pop up to try and take advantage of it before companies patch their systems. But because each bad actor is using swarms of thousands of super-agents, those agents are not only exploiting the existing vulnerability, but finding new zero days as well, all in a matter of hours. And the cycle repeats. As of the writing of this post there have been vulnerabilities found across at least half-dozen independent systems including LiteLLM, PHP, eMudhra Certificates, NGINX, Bind 9, and NodeJS. The list continues to grow hour by hour. Basically everything connected to the internet is at risk. Not just company websites and the usual databases of sensitive personal information, but all our critical infrastructure-- banks, power grids, pipelines, government agencies, all of it. The good news is that cybersecurity and ops teams had a chance to see this coming. This is why from when Mythos first dropped and had our first warning, these teams have been hard at work shoring up systems across the board. This is also why at Astranis we’ve seen a massive uptick in demand for truly air-gapped, sovereign networks, deployed via dedicated satellite to span an entire country or connect a large enterprise. It is in fact possible to take an entire category of critical infrastructure off the internet entirely. With a fully self-contained network where all the endpoints are off grid. This is not a new concept-- anyone could do this for your home today with equipment you can buy from Best Buy. But to stand up such a network covering a large geographic area or an entire country, without touching the existing fiber or other network infrastructure, you can only do that with a dedicated satellite. And specifically a satellite in a special orbit called geosynchronous orbit. Over the past year we’ve begun working with some of our customers on what it will take to implement truly air-gapped, standalone networks in support of mission critical use cases like core government functions, emergency response, defense, core infrastructure, or all of the above. That might sound extreme, but that is the kind of measure we expect will be required to shrink the attack surface of some of the world’s most critical infra. That’s just the position we are now in. It's a brave new world.

Post attachment
Alexander Whedon avatar
⟳ Darrell Etherington retweeted
Alexander Whedon@alex_whedon·Oct 12

We hit 11.4B tokens in the last hour and nearly 7B in the last half hour, within the first three hours after launch! You guys are cooking! Thanks for the engagement. It is possible there may be some latency spikes. We are considering opening up some of our training clusters for inference. We will let you know if anything changes!

Post attachmentPost attachment
Alexander Whedon avatar
⟳ Darrell Etherington retweeted
Alexander Whedon@alex_whedon·Oct 12

Introducing SubQ Snap, the first long-context decision model for real enterprise work. Snap is the only decision model or LLM where compute per token stays constant as context grows. • 384K tokens of multimodal context, 12x larger than Jev. • 61.7 on Decision Index v0.2.1 vs Jev’s 57.9, outperforming on 13 of 16 enterprise benchmarks. •Processes 100K tokens and returns a decision in under one second. •$0.05/MM input tokens. Output is free. Available today to anyone. The first 10,000 sign-ups get 100M free input tokens. We’ll likely hit capacity fast and will add users as we add compute. Compute offers welcome 🙂. Latency will be variable as we tune the platform to this new model and category of users. Model speeds are 80ms for minimal context, 250ms for 32k, 660ms for 100k, and 2.5 seconds for 384k. Visit: subq.ai/

1:16