Evan Rodgers 🏳️🌈 retweeted
Fuck I’m starting to worry Republicans firing all the experienced people in government so they could give no bid contracts to friends might not be panning out
CNN exclusive: "The highly classified network that the National Security Agency uses for its internal communications is down, according to six people familiar with the matter." cnn.com/2026/11/17/politics/…
Evan Rodgers 🏳️🌈 retweeted
Gen Z Men got a first hand account of what touching the stove is like and they fucked all the way off from it
I find him absolutely disgusting. This is censorship and control.
说个暴论,听完 Anthropic 掌门人达里奥阿莫迪对开源模型的这番最新炮轰,你才会彻底看清硅谷闭源巨头那副为了人类安全的面具底下,藏着怎样一颗想要垄断全球智能的野心,
达里奥在访谈里把话说得极其直白,甚至连掩饰都懒得掩饰了,他说如果继续放任开源大模型这条路走下去,世界可能会到达一个非常危险的地方,
原因只有三条,当你亲自控制并部署一个闭源模型时,你就有能力监控每一个用户的全部使用情况,
你可以随时撤销任何人的访问权限,你还可以随时改变模型愿意做的事情,而模型一旦开源发布,就彻底脱离了你的掌控,
把这几句冠冕堂皇的英文翻译成大白话,意思简直清晰到让人背脊发凉,说白了就是,
智能这东西只能由我们少数几家巨头攥在手里,我们得能随时偷看你在干嘛、随时决定哪句话不准说、甚至随时一脚把你踢出局,
一旦权重开源交到了普通人自己电脑上,我们就再也没法遥控和收税了,
更有意思的是,
这根本不是 Anthropic 一家的独角戏,OpenAI 的萨姆奥特曼最近也在唱着完全相同的调子,直言底层计算与存储资源根本不应该无限期地让每个人都能随便访问,
两大闭源寡头在这一点上默契得惊人,他们嘴上挂着的都是全人类安全与风险防范,心里算的全是如何把智能变成像水电煤一样的中心化特许经营权,
普通人如果看不穿这层窗户纸,未来付出的代价会极其惨重,一旦全球所有顶尖智力全部锁死在几家巨头的云端机房里,
企业和个人就彻底沦为了数字时代的佃农,你每天写的代码、跑的业务流程、沉淀的商业机密全在人家的实时监控之下,
今天人家心情好按 token 收你过路费,明天政策风向一变或者规则一改,后台轻轻点一下撤销访问权限,你的整套业务当场就会变成一堆废铁,
这里有一个特别反直觉的真相,在 AI 时代,开源模型存在的真正意义,从来不只是为了给开发者省几块钱 API 调用费,而是为了给全世界普通人保留一道不被任何巨头随时拔掉网线的数字防线,
所以当闭源寡头正在联手筑起一道名为安全的算力高墙时,唯有把开源权重死死攥在自己本地的硬盘里,
我们才算真正拥有了属于自己的独立智能,而不是随时看人脸色讨饭吃的云端租客。
73
The only reason I was able to get 4x R9700s from Microcenter this weekend was because they were open box.
I had to sign an affidavit asserting that I wouldn't sell or use these cards abroad, and they copied my driver's license, took all contact details, etc
There's been a series of weird reports here of people saying they're being "denied by micro center" buying GPUs/SSD/etc. so I decided to read about this a bit deeper:
It turns out that ?? [Some entities] were sending such a high volume of people to buy 5090s with cash that micro center managers had to get armored trucks for cash pickups 😂
the other discoveries made by some local micro center managers were a bit revealing
1) nationwide they implemented a policy where buyers had to state they wouldn't be exporting them. When one of the locations implemented this, the manager stated it made the [Nvidia] chip sales fall by more than half.
2) when they started running some id checks on suspicious transactions, there were groups of buyers coming in who all lived in the same apartment complexes; others stated they didn't know what it was, just getting it for a friend
3) it wasn't just 5090s but all freshly stocked GPUs
lol how many of these are being bought & shoved in warehouses or sent to chyna fr?
lol curveball at the end there "so lots of people won't be able to retire"
You can see how all logic returns to luxury boomer communism
The NY Times has the receipts today. Here's how the Nvidia Blackwell chip laundering worked.
1. Nvidia was blocked from selling Blackwell chips to China. So Jensen Huang sold them to Alice Huang at Megaspeed which installed them in data centers in Singapore, Malaysia, and Indonesia.
2. Howard Lutnick's family and company (Cantor Fitzgerald) were also heavily invested in these southeast Asian data centers and Lutnick blocked regulatory enforcement against Nvidia.
3. The data centers then either sold compute to China or hollowed out the servers and shipped the Blackwell chips directly to China. (Reports of hollowed out servers are from others on X, not in the articles.)
4. Nvidia could claim correctly that they had not shipped Blackwell chips to China. But the Megaspeed middlewoman (Alice Huang) was clearly facilitating Chinese access.
5. The Trump administration has known this for a year. Jensen Huang and President Trump were eager to sell Blackwells to China so they opposed the restrictions and didn't seem to mind the circumventions of the embargo. Secretary Rubio seems to be the main character within the administration trying to enforce the law.
It is also true that no one can run Nvidia other than Jensen Huang. So we are left in the unfortunate position where the CEO of the most important tech company in the world messed up and now the stock is going to crater and lots of people won't be able to retire because their stock portfolio just got nuked.
nytimes.com/2026/11/14/techn…
nytimes.com/2026/11/15/us/po…
74
"The open weights of smaller Qwen4 models will be released."
📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery!
Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window.
Highlights:
- Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation.
- Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables.
- Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery.
- $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%.
The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder.
Start building today! 🚀
- Blog: qwen.ai/blog?id=qwen4-max
- API: qwencloud.com/models/qwen4-m…
- Token Plan: qwencloud.com/pricing/token-…
More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀
131
YES!!
Judge denies motion to dismiss DRAM price-fixing lawsuit against Micron, Samsung, and SK hynix — says chipmakers acted like ‘hawks flying in formation,’ finds evidence of DRAM collusion plausible after 500% price surge tomshardware.com/tech-indust…
2
229
Send this to the stupidest chud in your life
Gen Z worker questions Kevin O'Leary on why Starbucks profits go to shareholders instead of employees and Kevin claps back saying “Nobody's making you work there”
“What if Starbucks took the roughly $2B in profit they made last year and they gave it back to the people who actually pour coffee?”
“What if we actually gave workers a chance to build wealth?”
Kevin O'Leary: “The problem with that scenario is if you don't leave any for the people that took…”
Gen Z worker: “I said some”
Kevin O'Leary: “I believe that company actually does do profit distribution”
Gen Z worker: “They don’t. Last year they actually gave all their profit away back to… take a guess. Where do you think it went?”
Kevin O'Leary: “Well, shareholders I guess”
Gen Z worker: “Dividends. There you go. So why should all the money we make doing our like work, our labor, why should it go to people like you? We did the work”
Kevin O'Leary: “Nobody's making you work there”
2
528
Evan Rodgers 🏳️🌈 retweeted
We optimized DeepSeek-V4.1-Flash on vLLM for Vera Rubin NVL72 and reached 58× the throughput of GB300 NVL72 at 1.7 million tokens/s/GPU and 387 tokens/s/user on SemiAnalysis AgentX.
Across the curve:
🔹 2.52 million tokens/s/GPU at peak throughput (2.7× GB300)
🔹 1.7 million tokens/s/GPU at P90 387 tokens/s/user (58× GB300)
🔹 1,810 tokens/s at one concurrent session with TP
1/4
So glad I got two of these, kicking myself for not getting more 🥲
🪙It's like finding gold. An 8GB NVIDIA mining GPU just became an 80GB AI accelerator.
And it's outperforming a retail A100 in an FP16 compute benchmark!
Hardware researcher Kate discovered how to unlock disabled hardware on NVIDIA's CMP 170HX cryptocurrency mining card.
So what's the secret? The 8GB CMP 170HX uses NVIDIA's GA100 silicon, the same GPU architecture as the A100.
And it physically contains much more memory than NVIDIA originally exposed.
👀Look at this ...
Stock CMP 170HX → Hardware-unlocked CMP 170HX
💾 VRAM: 8GB → 80 GiB
🧠 Streaming multiprocessors: 70 → ~120–126
🔗 NVLink: Disabled → Enabled
⚡ PCIe: Gen1 → Gen4
Now look at the reported FP16 compute benchmark:
🟢 Retail NVIDIA A100 → 72.49 TFLOPS
🔥 Unlocked CMP 170HX → 85.86 TFLOPS
According to Kate, of 10x 8GB mining cards examined
✅ 9x had a functional 80 GiB of memory.
✅ 1x had 64 GiB.
Kate's new method uses hardware voltage glitching and permanent fuse modifications to unlock additional memory and compute resources.
Imagine 80GB of HBM on 1x GPU.
Enough raw memory capacity to accommodate a 70B model at 8-bit or a much larger model at 4-bit, depending on runtime overhead.
And Kate has demonstrated 2x unlocked cards connected via NVLink.
From two cards originally sold as 8GB cryptocurrency mining accelerators!
🔗 Research and follow-up details in threads
178
The Ascend chips are already on the market, this is the rear-view mirror
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.
200
Evan Rodgers 🏳️🌈 retweeted
Obsessed with these semi surrealist digital corporate illustrations by Glenn Mitsui | Studio MD found in “Showcase Stock ~ Premier Illustration Volume 1” from 1998





































