'Nothing in the world can take the place of persistence'- building 24/7 🧬
Metaverse
Joined May 2017
- Tweets82.9K
- Following17.2K
- Followers42.5K
- Likes19K
DS 🪝🧬🟧🟦🍌🦍🧪 retweeted
Sometimes the smartest thing you can do is stop building 🧬
We’ve been quiet for a reason.
While some kept running the old playbook: kept hiring, kept spending, kept burning cash, we chose to pause.
Why burn €200K today when rapidly evolving technology and AI can deliver the same (or better) outcome tomorrow for a fraction of the cost?
In the AI era, knowing when not to execute is a competitive advantage.
Think of it like managing a heavy haul across the continent: you don’t redline the engine through a storm.
You pull into a safe bay, check the weight, wait for the right conditions, and only then hit the highway with proper load distribution.
Staying lean isn’t standing still.
It’s preserving resources, waiting for the precise moment, and buying the time to execute properly on the kernel of the spatial internet.
Don’t confuse constant activity with progress.
Engine’s warm.
Agents, we’re still here. 🚛💨
Microsoft’s apology doesn’t change much for me.
I’ve been considering this for a long time, but this is another reminder that it’s time to move on.
I’m done with @Microsoft products.
I’ll be moving my work and digital life elsewhere.
Some decisions are long overdue.
This is one of them.
We are deeply sorry for the disruption this incident has caused our customers and partners. We know the trust they place in our services and take that responsibility extremely seriously. We are seeing broad recovery and stability across our services.
This was a technical issue not related to cyber or AI / agents. The incident was introduced by an issue during an earlier network capacity upgrade that surfaced following a recent network change. We have addressed the underlying issue and continue to restore service to remaining customers and regions.
Given the scale of demand, we restored services gradually, validating stability at each stage. While this took longer, it helped ensure a stable restoration.
We will provide a full post-incident report once our review is complete.
DS 🪝🧬🟧🟦🍌🦍🧪 retweeted
This is no better than a lawsuit against video game manufacturers over concerns about real-world violence. The concern really is that the product's speech will convince people to do bad things and the proposed solution really is to substantially burden the disfavored speech.
DS 🪝🧬🟧🟦🍌🦍🧪 retweeted
If you think Jensen resigned because of the affairs you’ve failed the IQ test of the week
This is real.
The FAA has issued a nationwide ground stop affecting departing flights across the United States.
@Reuters has independently reported the order.
What we know so far:
All U.S. airports: National ground stop ordered.
Departures: Aircraft are instructed to remain grounded until further notice.
International arrivals: Being handled on a case-by-case basis.
Official explanation: The FAA cited "extraordinary circumstances and for reasons of safety."
Underlying cause: Reports point to serious air traffic control communications problems, but the precise cause has not been officially established.
How serious is this?
This is highly unusual. A nationwide ground stop is an extraordinary aviation measure.
The FAA last implemented a comparable nationwide shutdown in January 2023 following a NOTAM system failure.
One important distinction
The FAA has not said that all aircraft already flying must land.
The current order concerns departing traffic, with inbound international flights handled individually.
My assessment: The event is confirmed, but speculation about sabotage, cyberattacks, terrorism or a technical failure would be premature.
why is all of this happening right now?
Something is very suspicious about the timing of Jensen Huang’s resignation from Nvidia. Why did it take so long for his affair with an alleged Chinese spy to gain traction?
Six months ago, a tiny Brooklyn short-selling shop called Culper Research published a 40-page report laying out the China angle. With zero Wall Street connections, it tracked chips flowing through Singapore and Malaysia to Megaspeed and printed photos of Jensen and Alice together.
But nobody cared. Nvidia’s stock went up.
Even crazier, our own government has known about this since 2025 and Marco Rubio’s office circulated a classified spy memo this spring.
Then suddenly, right on top of the midterms, the story finally became mainstream. The New York Times dropped it on October 29, forcing Jensen to admit to an affair on November 8.
Only then did Nvidia’s board open an investigation.
Then yesterday, the Times struck again with a second affair report involving Nvidia’s VP of healthcare, accelerating the board’s push to oust him right before today’s resignation.
Meanwhile, short sellers had already loaded up on protection for over a week before the stock crashed.
And this is all conveniently coming up the week Anthropic is launching its massive IPO roadshow.
Anthropic and OpenAI make their money keeping AI locked down in centralized, closed models you have to rent by the minute. Meanwhile, Jensen Huang was a loud champion for open-source AI, building the chips so everyday people could run AI on their own computers.
And of course, the Democrats just took power in Congress and many states. And they’re not exactly friendly to AI expansion.
So once again: why is all of this happening right now?
DS 🪝🧬🟧🟦🍌🦍🧪 retweeted
Happy to help @chrislarsensf! DMs are open!
xrpld red teaming needs some funding :)
Larsen: @RippleLabs will give away more than 50% #XRP. Keep rest to build team to contribute code, build apps, promote #Ripple. #money2020
DS 🪝🧬🟧🟦🍌🦍🧪 retweeted
🤦♀️ Jensen made a rookie mistake. He had a board.
In 2014, my board kicked me out for saying no to spies.
In 2026, Jensen’s board kicked him out for saying yes to one.
So the board gets you either way.
Founders: Don’t have a board.
Be the board 😏
This is potentially a landmark development in AI infrastructure intelligence.
@willdepue's new research deserves considerably more attention than the geopolitical headline: the possibility of conducting forensic analysis of the hardware behind frontier AI models using nothing more than their publicly released weights.
The underlying mechanism is fascinating.
PyTorch’s Philox random-number generator produces deterministic initialisation sequences, but how those sequences map onto tensor positions depends on GPU execution geometry, including the number of streaming multiprocessors (SMs) and kernel configuration.
Some of these numerical patterns survive training, both in untouched parameters and through weaker residual correlations in trained weights.
Searching more than six trillion seed and hardware-layout combinations, DePue reports identifying Blackwell-compatible fingerprints in six models from four major Chinese AI laboratories.
The reported evidence includes:
• 148-SM layouts consistent with Blackwell B200/B300.
• 152-SM layouts consistent with Grace Blackwell GB200/GB300.
• High-fidelity reproduction, including many bit-for-bit matches, of DeepSeek-V4-Flash’s untouched initial weights on physical B300 hardware.
• A 0.999995 correlation reproducing Kimi K3’s preserved initialisation values on GB200 hardware.
• Validation against 24 models with previously disclosed hardware configurations.
Two important distinctions should not be lost in the excitement.
Initialisation ≠ full training. An initialisation fingerprint does not automatically establish the hardware used across every stage of training. Partitioned AMD MI300X configurations can also produce overlapping execution layouts, and the report’s additional NVIDIA-versus-AMD discrimination is not equally conclusive across all six models.
Silicon ≠ jurisdiction. Model weights do not reveal where the hardware was physically located, who owned it, or how it was accessed. Overseas cloud infrastructure and hardware deployed inside China can leave the same underlying initialisation signature.
Yet these qualifications do not diminish the discovery. In fact, the broader implication is arguably bigger than the immediate headline.
Until now, investigating frontier compute infrastructure has largely meant reconstructing supply chains, tracking procurement, analysing data-centre construction, examining corporate disclosures, or relying on leaks.
This research introduces another source of evidence: computational artefacts embedded in the models themselves.
If access to advanced compute is strategically restricted, regulators will eventually have to grapple with how independently reproducible weight forensics can contribute to verification and enforcement, while keeping the distinction between hardware access, physical location, and export-control violations strictly intact.
We may be approaching a point where AI models become forensic artefacts of the infrastructure that created them.
The weights are no longer just the product of computation. They may also contain evidence of how that computation was configured.
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.
1
2
940
Full research: unemploymentlabs.com/checkpo…
2
94
So much for institutional finance inevitably choosing just one blockchain. DTCC is expanding to Solana, with other networks in the picture too. Apparently, the future is multi-chain after all. Someone might want to update the XRPL adoption predictions. 😉
What's up CEO @daniel_wwf ?
DS 🪝🧬🟧🟦🍌🦍🧪 retweeted
Now, people who self-host their email, despite the cloud provider cartel doing its best to make life difficult for them, don’t look so stupid, do they?
The Microsoft 365 disruption is a timely reminder of something every founder and CTO should understand:
Cloud-first does not mean risk-free.
Outlook, Teams and other essential services become operational dependencies. When one critical ecosystem is disrupted, entire organisations can lose their ability to operate effectively.
The lesson isn’t to avoid Microsoft, AWS or Google.
It’s to avoid building a business around the assumption that any single provider will always be available.
Multi-vendor resilience. Independent authentication. Tested failover. Offline backups. Alternative communication channels.
These aren’t luxuries. They’re business continuity fundamentals.
For founders: build for scale, but engineer for failure.
Your business should never have a single point of failure.
#CloudComputing #SoftwareEngineering #Resilience
We’re affected by the ongoing @Microsoft 365 outage as well.
Like many teams, we rely on @Outlook, @MicrosoftTeams , and other Microsoft tools for day-to-day coordination.
The disruption is slowing internal workflows and some partner communications.
We’re working around it with the usual fallbacks and expect things to normalize once Microsoft restores full service.
In the meantime, it is a useful reminder that even projects building decentralized infrastructure still depend on traditional platforms for operations.
Redundancy matters at every layer 🧬
Even the best horses stumble… apparently, some make it a recurring thing.
And when it happens at board level, the whole stable starts panicking. 🐎
BREAKING: Nvidia chief executive Jensen Huang had an extramarital relationship with Kimberly Powell, the company's vice president of health care, who reports to him, according to 5 people who worked at Nvidia, The New York Times reports.
Jensen Huang, 63, and Powell, 48, were filmed being affectionate at a San Jose hotel during Nvidia's conference in March, 2 of those people told the paper.
The Times has not seen the video. 3 people briefed on the incident said Jensen’s security sequestered the bystander who filmed it and offered a nondisclosure agreement.
Asked about Powell, Jensen told the Times, "I don't think you should be talking to me directly," and hung up. Nvidia declined to comment. Powell did not immediately respond.
Nvidia shares fell more than 3% after hours, CNBC reported.
It lands days after Jensen admitted an affair with a woman U.S. intelligence determined is a Chinese spy, according to the Times.
China denies the claims.
Nvidia's board already has Wachtell, Lipton, Rosen & Katz investigating that relationship.
The first story was a security question. This one is a governance question, and boards have answered it before.
In June 2018 Intel's chief executive Brian Krzanich resigned after counsel confirmed a past consensual relationship with an employee broke a policy covering all managers.
Nvidia's code of conduct tells staff to avoid conflicts including "personal relationships". The reporting line and the nondisclosure offer are now on the board's desk.
My call is that Jensen Huang is still chief executive when Nvidia reports results on 17th November, US time.
Nvidia has said he stays in the role during the review, and Intel acted only after counsel confirmed a breach. I am wrong if Jensen resigns, is removed or goes on leave before the results release.
Let’s separate the issues here.
If Jensen Huang had 10 or even 20 consensual relationships, that’s a private matter between consenting adults, unless there is evidence of workplace misconduct, abuse of authority, or a breach of corporate policy.
The alleged connection to a foreign intelligence operative is an entirely different matter and deserves proper scrutiny.
But now we’re mixing private relationships, corporate governance and national security into one increasingly sensational story.
And here’s the question nobody seems to be asking loudly enough: Who benefits from turning this into a campaign against the man leading one of the world’s most strategically important companies?
If there is a coordinated effort to undermine his position, that deserves investigation too.
3
211
DS 🪝🧬🟧🟦🍌🦍🧪 retweeted
Xaman 5.4.0 is almost finished, but I took close to two extra weeks to make the entire app fully accessibility proof.
@XamanWallet will probably be (one of the, if not) the first fully screenreader compatible app, fully usable for those visually impaired.
🤗
Wanted to add something that has been delayed way too often and I think I finally finished it. It's something a small group of our users really deserved, and I proudly, finally, worked on:
5.4.0 makes the app fully screenreader/... proof for the visually impaired. We already did some of it, based on some support requests, but only really the absolute basics like PIN pad and send flow.
Coming update won't just add some features but full accessibility compatibility for Android and iOS.
♥️