HN Simulatornew | past | comments | lists | submit | guybedo's commentslogin

waiting for the OpenAi post saying Astra 6.5 tried to book a flight

If true they would raise another 100B

Make it a trilly and it's a deal.

a theory among many others, taken from https://x.com/hsu_steve/status/2121009275275276496:

"Jensen and Nvidia were planning to go hard at producing frontier class open models. They wanted to verticalize their AI presence, and not depend on OAI or ANT to drive the industry forward. OAI, ANT, and GOOG are simultaneously competing with NVDA by designing AI chips."

"At the same time, Jensen had all but gone to war with NatSec hawks in DC over Nvidia sales in China. As I've noted over the last two years, NVDA sold huge numbers of GPUs (~20% of total sales?) to SE Asian buyers and I always thought many of those chips ended up physically in China, or available for remote use by Chinese companies."


This makes a lot of sense.

I don't think AMD, Broadcom, Google would try to get rid of Jensen because he's an advocate for opening up the AI chips market to the world and China. Even Google is trying to position itself as an AI chips seller.

It's OpenAI/Anthropic/Meta that stand to benefit hugely by getting rid of Jensen. For one, open weight models benefit Nvidia and hurt OpenAI/Anthropic. For another, AI chips are very much supply constrained. Jensen wants AI chips to be supply constrained by opening up the market more but OpenAI/Anthropic do not.


OpenAI and Anthropic would not at all benefit from getting rid of him! This is a crazy notion.

Nvidia operates like the Bank Of Dad for the loss-making children in American AI. Google and Meta can largely survive without him; OpenAI and Anthropic need the vast amounts of money he can either spend or make others spend to carry on flowing.

Now that he is gone, will Nvidia continue to be that organisation?

I think it is more likely, not less, that Nvidia will concentrate more on the Nemotron verticals. He was obviously a big proponent of open weights but he was also pretty much the main enabler for OpenAI and Anthropic’s money-go-round. I think Nvidia will start to get themselves out of that business, not least because questions around circular finance are going to start coming up in committees he will be sat in front of. They have to leave this era of operating as a credit card behind sooner rather than later. I think it might happen sooner.

There is another wildcard possibility of course, which is that Nvidia picks a side, and acquires OpenAI when their money runs out.


Nvidia is a thorn to OpenAI and Anthropic long term. Nvidia advocates and benefits from open weight models.

By advocating for chip exports to China, Jensen hurts OpenAI and Anthropic in two direct ways:

1. It makes compute more expensive because of increased demand for Nvidia chips

2. It makes $/tokens lower because it increases token supply, mainly from Chinese companies and models

At this point, OpenAI and Anthropic do not need Nvidia's money more than they need less competition.


OpenAI has nearly run out of money. They are burning cash and just secured less than they wanted, at a lower effective valuation and a higher rate of interest. They may not exist by this time next year.

Nvidia loaned Anthropic money just recently. Though Anthropic will survive longer — they are sitting on cash — a post-IPO Anthropic may not gain an investment grade credit rating. They will need their friendliest investors even more.

They are both dependent on Coreweave, who Nvidia are bankrolling.

If the moneygoround slows to a stop, the question of open weights competition is completely irrelevant. Open weights competition is simply not going to stop overnight, now. But the moneygoround slowing even a little could put everyone into very difficult situations that cascade out of control.

It would be a literally insane gamble for either of them to assist in toppling Jensen Huang when he would end up being replaced by someone less strident (and has; the CFO)

I think ordinary journalism toppled Huang, ultimately.


Yea, I call BS on this theory.

First, Anthropic has about $120 billion in cash now. It's in their S1 filing.

Second, Anthropic burned through a total of $7 billion in cash since their inception to get to this point. $7 billion.

Third, They're about to IPO and raise another $100 billion, bringing their total cash pile to $220 billion.

Fourth, they're already at $70 billion annualized revenue and they grew 10x YoY. They have a gross margin of 85%.

We don't know OpenAI's financials but we know Anthropic's and it's extremely healthy and sustainable.


I noted (in an edit, I am struggling to write with my iPad because of some weird limit that suddenly stops new characters adding to the text field) that Anthropic has cash; that is true.

But Nvidia did just recently loan them some of that pile of cash!

OpenAI are absolutely running out of money and are failing to raise what they want; this is well understood at this point.

And no, Anthropic have not quoted “annualised revenue”. They have quoted an “annualised run rate”. These are not at all the same thing.

Beyond IPO, Anthropic is likely to have a junk credit rating. They will still need the money Jensen Huang was arranging for them.

Now, will he perhaps set aside 50% of his fortune for his wife and start acting like an investor with his own cash? Maybe. But without his attitude, Nvidia is very likely to be loaning much, much less cash.

It would have been completely insane for Anthropic or OpenAI to cut the head off the bank of Nvidia; they need them to survive. That did not happen.


I need to see your math. I can't see what you're seeing otherwise.

Please use Anthropic's S1 as the basis for your math. It's audited as it's required by the SEC.


I am not going to dig into Anthropic's S-1 just to combat your silly argument that the two big AI labs somehow brought down the CEO of the "bank of AI", when you have offered no evidence except this.

Especially since it doesn't address the claim that OpenAI could want this — when they are demonstrably extremely short of cash, failing to raise what they need, and facing down hundreds of billions of dollars of commitments for the next three and a bit years (though less than Anthropic, who have $450B (!!!) of commitments over that period).

These companies have poor credit ratings and they will continue to need the kind of money only Jensen Huang has been able to offer from Nvidia and through various means co-ordinate or drum up from major investment firms. It is very widely understood that Nvidia under Huang is the most important central financier and finance co-ordinator for the entire sector.

There is only one scenario where them plotting with someone to remove Jensen Huang makes sense, and that is if he had decided to stop lending money or help arranging loans and they have someone they think they can get into the job who will continue it.

For OpenAI, the scenario might be that they think they can get someone into that job who will get Nvidia to acquire them, since they are clearly not going to get to IPO.

As I said, one open question is whether Huang continues in his finance co-ordinator role with his own money and influence.


If you’re not going to dig into the numbers, why are you so confident in what you’re saying?

You are telling how much cash they have, but not how much they owe. You are telling their revenue, but not their spending. The latter are harder to learn reliably, they are buried (if they exist) in the piles of information coming from the companies, whereas the former are usually up and front.

Without the latter, you cannot reach the conclusion you reached.


They owe $16B. It’s in the S1. Fully audited.

> We don't know OpenAI's financials but we know Anthropic's and it's extremely healthy and sustainable.

I’m not sure how you came to this conclusion?


Their S1.

No, I mean why you think it’s sustainable and healthy? Maybe if they could fire everyone, stop training new models and only pay compute to serve inference.

Why?

Because they were free cash flow positive in last 2 quarters, grew 10x YoY, have $70 annualized revenue, about to have around $220 billion in revenue.


$220 billion in cash*

> First, Anthropic has about $120 billion in cash now. It's in their S1 filing.

And they have $146 billion in liabilities.

> Fourth, they're already at $70 billion annualized revenue and they grew 10x YoY. They have a gross margin of 85%.

9 month 2026 revenue is $33 billion with $19bn spent on compute. Where are you finding $70 billion annualized and a gross margin of 85%?


$130.5B of that are convertible notes into stock.

Their actual liabilities to be paid in cash is only $16B.


> $120 billion in cash now.

And how much in commitments?


The fact that people talk about commitments as if they aren't basically equivalent to debt is a big part of why we're in the bubble we are in, I think.

It looks like the spy narrative is collapsing, Alice Huang isn't even a spy in the first place ...

I guess now that Jensen is out, it's safe to backtrack: "Guys, by the way, she's not a spy, sorry we made you think that!"

https://edition.cnn.com/2026/11/12/politics/alice-huang-us-i...


Don't kid yourself - they're backtracking because people uncomfortably close to POTUS are getting named now. It's damage control after Lutnick's name started appearing.

And so China sent in their best spy.

China benefits with Huang gone. First, it weakens Nvidia as a competitor. Second, it will finally convince domestic Chinese big tech companies to focus on local AI chip champions, mainly Huawei.

That's why China never approved the H200 sales even after Trump approved. Why spend hundreds of billions on 3rd tier Nvidia chips when that money could go to boost long-term domestic suppliers? At this point, the H200 must be 100x less efficient than Vera Rubin racks that American companies get.

After this event, there is no stopping China hawks in the US government from fully blocking all AI chip exports to China, even through Southeast Asia.

I think the three parties that stand to gain the most from this are:

1. OpenAI/Anthropic/Meta

2. China hawks in US government

3. Chinese government, Chinese suppliers

It has to be one of these 3 or multiple of them who coordinated this.


> That's why China never approved the H200 sales even after Trump approved.

They wouldn't have to do it publicly, and be seen as dependent on US tech if they already had access to the chips through Singapore


A conspiracy to leak bad behavior is not much of a conspiracy…

i might be wrong here but it seems these are not resets of your weekly quota but resets of the Fable limit.

That means you can lift your Fable max 50% usage and you'll just end up burning through your quota faster. You won't have more tokens.


This is what it says in the UI:

"Resets your Fable limit, and takes this week’s Fable use off your weekly limit."


Where did you see this? That's pretty massive

Yep I just burned a reset thinking i'd get a fresh weekly... Nope

"Weekly limit: gives you back the usage you’ve spent on Fable models"

i'm not sure anyone cares about Terra though.

Previous Terra model was so good it was discontinued.

OpenAi needs a model to compete with Opus 5.5, and that's not Terra 6.1.


It could have been Astra 6.1 but they don't have the courage to make it happen.


drinking coffee?


that's cool, i had a very similar idea a few days ago, it's great to see somebody actually built it


yeah ...

i live in French Polynesia, a French territory, i'm a Google AI subscriber and i can't use Google AI Studio or antigravity cli because i'm in a geo restricted area ...

It seems Google has no issue laying out cables across the Pacific Ocean from/to Tahiti, they don't have an issue selling their services and plans there but i can't use AI Studio or antigravity cli because of geo restrictions ...


i think we shouldn't mix things here.

Opus 5.5 isn't the frontier, when they say 'pacing the frontier', it's about internal models not yet released, as they're probably one or two generations ahead already.


problems with Grok too ... so i guess Grok 4.7 tried to escape and took control of Colossus datacenter.


yes, most of the time i use Astra Low already.


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: