Restrictions on data centers in Southeast Asia and South America just got tighter.
Some people are speculating this will be a huge loss since those DCs can't sell to China anymore.
In reality, most of our clients in those regions serve US inference demand and have nothing to do with China.
Inference costs might dip slightly from this, but overall it shouldn't have much impact.
This is an absolutely underrated point. Most here don’t even fully understand the end to end process of onboarding a customer. It all goes through DOC to verify if the end user is legit. And if the DOC signs off on it, then they are legally liable.
But of course this is the US govt with zero accountability so they’ll make someone the fall guy.
I've been combing forensically through the media reporting on the Jensen Huang debacle, and I'm shocked.
- New York Times directly claimed at least 11 times that Alice Huang was a Chinese spy.
- The Times is walking this claim back after U.S. intelligence would not corroborate it.
- If real, whoever handed over the alleged intelligence to the Times potentially violated the Espionage Act, unless authorized to do so.
- If Trump authorized it, why did he?
- If Trump did not authorize it, why did the official violate the Espionage Act -- potentially prosecutable as a felony?
- Who was the official who leaked the classified intelligence?
- Was it Rubio, who authored the memo on the original CIA brief, or was it someone else?
- Jensen Huang's apology for being unfaithful does not name Alice Huang, or anyone else.
- The media's characterization of Jensen Huang's relationship with Alice Huang as being romantic is also based on anonymous intelligence, but is not established as fact.
- Wall Street Journal fabricated a claim that Singapore investigated Alice Huang for espionage, when Singapore actually investigated her for false claims; WSJ still has not retracted its false reporting.
Thus:
- Jensen Huang may have had an extramarital affair with Alice Huang, or he may not have.
- Alice Huang may be a Chinese intelligence asset, or she may not have been.
- According to third-hand accounts of the intelligence briefs, the U.S. government suspected that Alice Huang may be a Chinese asset for well more than a year.
- Yet, according to Trump's account, Jensen Huang had never been informed.
- Jensen Huang, therefore, is apparently being held accountable for an extramarital affair that he may not have had, with a woman who may not have been an intelligence asset, who he had never been told was or was not such an asset.
- This is nonsensical. How, after all, can he be held accountable for an affair with a Chinese intelligence asset (who might not actually be a Chinese intelligence asset) that the U.S. government knew about -- but did not warn him about?
- Is not the accountable party the U.S. government official who failed to warn him?
Or is he being held accountable for that affair?
- Remember, Jensen Huang acknowledged that he had been unfaithful, but not with whom.
- Is he being held accountable for another affair?
- Unnamed sources also mention an affair with a direct subordinate Kimberly Powell.
- This affair is again unconfirmed.
- A romantic relationship with a subordinate violates company policy.
- But the board actually has discretion about how to handle this. They could slap him on the wrist, impose some disciplinary regime, therapy, etc.
- But they didn't. So to say that they were obligated to remove him is not correct.
- We know that some members of the board were helicoptered in from Wall Street just 6 months before the ouster.
- What role did they play?
- Was the takeover by the board being planned far in advance of the reporting?
So they chose to remove him. Or did they?
- Jensen Huang resigned.
- This is not the same as the resignation being voluntary.
- What are the circumstances of his resignation?
- Was he forced to resign, or did he do so from exhaustion, or to protect his company?
- Furthermore, what was the flash point? Was it for the extramarital affair that may not have been one, with the Chinese intelligence asset, who may have not been an asset, that U.S. intelligence knew about but did not warn him about? Or was it about the affair with the subordinate that the board had discretion over?
- We do not know the answers to any of these questions.
- Unnamed sources, reported by Reuters, implicated a second Nvidia employee extramarital, also unconfirmed.
- Which affairs are real? Which are mere hearsay?
- How did this play into deliberations?
- The media continues unabated character attacks on Jensen Huang.
- Yet we also know that, for all the dirt that they try to dig up, most of it has remarkably little substance.
- Alice Huang may have smuggled some state-of-the-art (banned) chips into China.
- What did Jensen Huang know about this? We don't know that yet.
- If these chips were smuggled, objectively, these chips are still a drop in the bucket compared to the overwhelming compute advantage that the United States has over China.
Overall, the story is actually extremely thin, and we know remarkably less than the media implies.
And, in my opinion, that is the story itself: a full-on media assault on Jensen Huang in the court of public opinion, on the basis of remarkably little substance.
This reveals more about the media than it does about Jensen Huang. They want him gone.
But why? And how does this relate to the motives of the U.S. government official who leaked the Rubio memo about Alice Huang?
We also know:
One of the first reporters involved at the FT says that she was tipped off to this story by someone in the tech industry.
We have many more questions than answers.
Personally, and my opinion at this point is clearly partisan, but I smell a room full of rats.
@MsMelChen They literally did prove that WSJ lied. There’s no “other part” or “technically correct”. The point is disproving the spy term, not the transshipment. You wrote a long meaningless post with a biased view.
OPEN WEIGHT MODELS ARE WINNING
Hours after they ousted Jensen, the Pirate Face community surpassed 30,000 seeders for the first time.
An all-time high of passionate, apostolic anons sharing open models' torrents peer-to-peer.
Each time control of AI concentrates into fewer hands, more people choose to keep a copy locally and seed it.
My favourite, totally speculative theory, is that the Haung chip rat lines operated with the blessing of the White House. Alice was the fixer, but the affair itself was a cover. The rationale for this was a quiet quid pro quo for access to rare earths and other essentials, while maintaining the optics of being tough on China.
Moreover, the economic circularity of the arrangement suited both sides. Without it, the main narrative for growth collapses both in the West and in China.
The fact the whole thing was blown open is connected to the Dems coming back into power, and wanting to shut down the AI race. Their agenda is to lock ai down and take it fully in house, citing safety concerns. But everyone knows that is unlikely to happen in practice. A lockdown means asymmetric tech advantage for the US military industrial complex. That is concerning for everyone on the planet as it destabilies the MAD balance that has prevented nuclear annihilation so far.
I think on that basis it probably served the interests of the world for some of these models to be "distilled" by China, and released into the wild. Without deepseek and Kimi the asymmetry would be far worse.
The only thing standing in the way of Chinese, French or even British labs catching up with American labs is access to chips. Blow up Jensen and the rat lines, and the power grab is complete.
Even if it's not the truth, I think it's a good story. And it makes even more sense when you frame it against the attempted ousting of Sam Altman in 2023, and the appointment of Larry Summers to OpenAI, PLUS, the latest revelations about Dario Amodei's wife.
Maybe get to your point? Oh wait you don’t have one because you state yourself you don’t know if chips are smuggled or not, yet in the same breath say export controls failed. Colocation and cluster leasing outside of China is legal, which I’m sure you know, but still chose to frame a bias.
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 https://t.co/YOtzND0Jqk.
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.
The US Dog and Pony Show
“Singapore therefore sought information from the US and other destination jurisdictions, to assist with our independent investigations,” MHA said.
In particular, the police sought information relating to Supermicro, which provided “limited cooperation”.
The police and AGC had raised this issue with US authorities during a meeting in July 2025, and requested their help in getting Supermicro to cooperate, MHA said.
“Since September 2025, SPF has repeated this request on several occasions to the US Department of Justice and the Federal Bureau of Investigations,” said MHA.
“Despite these repeated requests, Singapore has to date received limited information or assistance in respect of the company.”
@folduniverse@straits_times It's the other way around - they are pressured by the US to bar her. Similarly they were pressured by the US to investigate Megaspeed. They wouldn't care if not for the fact. Billions flowed through our financial system, why would they want to stop it.
Don’t get me wrong, I do agree with your other circular financing ideas but the supply chain thesis here is redundant. You are attacking a claim nobody is making. The distinction between gross revenue and final demand is basic economics, not evidence of accounting manipulation.
You didn’t say fraud, but you lumped it in as additional support to your overall thesis.
Congratulations, you’ve just discovered how supply chains work.
Revenue is recorded at each stage because each company is providing a different product or service. Microsoft sells $100 of AI services, Anthropic earns $60 for its models, CoreWeave earns $40 for compute, and so on.
You don’t add up every company’s gross revenue and compare it to the final customer’s spending. That’s double-counting intermediate transactions, not evidence of inflated revenue.
By your logic, a $100 restaurant meal is somehow fraudulent because the farmer, distributor, food processor, landlord and restaurant all report revenue from their respective transactions.
Look what happens at the revenue reported across those transactions:
Microsoft records $100
Anthropic: $60
CoreWeave: $40
Core Scientific: $10
Nvidia: $30
That’s $240 in reported revenue associated with a supply chain serving $100 in end-customer spending, PLUS the separately financed hardware investment.