👍
Replying to @EconBerger
3/ And overall postings are at their highest level since April 2025. (Though we're far, far below the post-pandemic peak.)
2
235
Thankful to all @WorkWhileAI employees for a formidable year so far (-- still loading). Thank you to our 1.4m workers+. We are blessed to have you on our platform. AI does't kill jobs, it creates massive labor opportunities. #LongHumans @BenoitSf @alanarmen44. Is that vertical enough???? or shall we push more?
2
4
287
impressive and a peak into the great impact of AI on humanity
Demis Hassabis: AI solved a 50-year biology challenge in just one year, saving a billion years of PhD research time.
“I think we'll be able to get down drug discovery from years to maybe months.”
“We managed to actually fold, using AlphaFold, in one year, all those 200 million proteins known to science.”
“So that's a billion years of PhD time saved.”
“We can expect to see some pretty dramatic health medicine breakthroughs in the coming few years.”
2
197
Very true. The valley at this stage is full of transactional people (you may call them accidental entrepreneurs) that lack the ethics, the moral fabric and driven solely by personal greed. It will change.
Buffett today: "Kindness is costless but also priceless. A year later, I would only add that it compounds."
I think it’s an important message for all of us in the Valley. The best of Silicon Valley was built on kindness and ambition. We need kindness. Optimizing just for ambition isn’t enough.
3
221
For a while the red and the green were going in the right direction. The culprit is not “productivity” nor “AI”
It is outsourcing higher paying jobs!
2
235
The @WSJ @theinformation @FT have been doing amazing reporting on this. So proud of these outlets that are focused on the real issue and not the side show that involves hurting female execs at Nvidia and their families.
Really bad if true. Did Jensen know? Was he turning a blind eye? Were export laws broken? Did Jensen feel untouchable?
1
199
Forget the appetizer, this is the meal. I feel bad for all Nvidia Female execs dragged into this.
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.
2
242
well, Outlook is down and no one notices nor cares.
9 months ago, I wrote this. Today, I don't see any "office" future for MSFT. It will be a good cloud provider for Claude
nitter.cf/compose/articles/edit/…
510
9 months ago, I wrote this. Today, I don't see any "office" future for MSFT. It will be a good cloud provider for Claude
nitter.cf/compose/articles/edit/…
Ha ha ha --- 2 days early? You mean the feature that every neobank introduced 10 years ago? pymnts.com/partnerships/2026…
157
This will be the best news for Silicon Valley- Insanely high valuations are toxic
Jitters are entering the private markets
Latest w/ @RebeccaTorrenc5
bloomberg.com/news/articles/…
1
4
1,226
I have to ask an honest question to @a16z. The partners are posting top notch and research heavy content. Not complaining, but who is managing the editorial content?
RIP #jobapocalypse 2026-2026
People used to say a company has two jobs (1) build and (2) sell.
AI is making building easier. That means selling may become more valuable, not less. When everyone can make at least a fine/mediocre product quickly, you need to get the right people to care, trust you, and choose it over something else.
An AI agent can find candidates and send a recruiting email. But the people you most want to hire won't join because an agent reached out. They'll join because a person made them believe in the team, the ambition, and the work.
An agent can answer questions in your homepage chat bot. But when a purchase has to get through finance, procurement, and a room full of stakeholders, someone still has to understand what each person needs, make the case, and keep the decision moving.
AI can make building cheaper. It can make selling more important.
When everyone can build something, the advantage goes to the people who can persuade.
150



















