🔋Self-Motivated 🔋

[zerohedge @zerohedge]

Zuck: “The pricing from some of the other labs is very extreme and has very high margins. We think that there’s a real ability to be able to offer frontier or very high-level intelligence at a much more affordable cost.”

Epic pricing war breaking out among agentic models.

[🔋Self-Motivated 🔋 @self_motivated0]

@damnang2 As it should. Brilliant piece m. Nicely articulated.

[🔋Self-Motivated 🔋 @self_motivated0]

Isn't it. Not as sophiscated as some of these brilliant writers but these little proxies get the job done too. I try to keep things simple. Ask grok everyday when are you getting more memory so you can remember our long context convos and the response usually is - the next tesla software stack is being upgraded with MORE memory - coming soon.

[Damnang @damnang2]

This article is fully open to everyone.

In this piece, I cover two things.

First, for readers without a technical background, I explain how NVIDIA rack names are structured and what each part means.

Second, based on that foundation, I analyze SemiAnalysis’s July 6 post. I go through what it implies, why those posts may sound alarming, and what the actual midplane yield picture could look like in detail.

The first part is written for readers who may not have a technical background. If you only want to read my thoughts on the latest @SemiAnalysis_ post, please jump directly to Section 5.

[Damnang @damnang2]

https://t.co/Otk4B83HRA

[🔋Self-Motivated 🔋 @self_motivated0]

@RealJimChanos 4.0 is normal for a growing capex intensity business. Look at Amazon for example. But if this gets to 5.0 or higher then we may have a problem. Revenue can’t be on a perpetual pace of outgrowing. invested capital.

[🔋Self-Motivated 🔋 @self_motivated0]

@tengyanAI don't forget Korea export data, where a flash is given 1st and 21st of the month.

[🔋Self-Motivated 🔋 @self_motivated0]

@saso_capital There were 4.5 million developers developing on CUDA platform last year. That number is not at + 6 million, and CUDA is by far the largest platform being developed on. There is no one else even close to this.

[Micron Technology @MicronTech]

Today, we're proud to announce a strategic agreement with @AnthropicAI that spans memory and storage AI architecture design, supply and demand, enterprise adoption of Claude across Micron and a strategic investment in Anthropic’s Series H funding round. https://t.co/WkAzl0YXxK

[🔋Self-Motivated 🔋 @self_motivated0]

@yianisz The most important is the number of LTAs signed

[🔋Self-Motivated 🔋 @self_motivated0]

Interesting how these stories are highlighted but stories like Jane Street are ignored. Jane street earned $38 billion in revenue last year with $31 billion in profits. Why? Compute. The more compute they bought online led to 80 percent NET profit margins. 80 percent. AI seems to be helping someone 👀 the question is who else is willing to step up to the plate and highlight the margins they’re achieving from AI in order to put this noise to rest. My guess not many due to competive disadvantages.

[🔋Self-Motivated 🔋 @self_motivated0]

2.2 TRILLION in backlog collectively across all hyperscalers. The backlog represents thousands of customers relying on the hyperscalers for compute. Compute the hyperscalers are relying on Nvidia’s hardware in order to deliver.
And, this is just 60 percent of their book. The other 40 percent, various countries jumping on this train with more boarding everyday - diversifying the concentration.

[🔋Self-Motivated 🔋 @self_motivated0]

You really think the Fed will raise rates twice end of year with mid terms around the corner and gas and food prices already high? Let’s go ahead and do what Europe does and hit Americans with a double tax on top of the transitory inflation they are already dealing with. Yeah, that’s a good idea- not. These guys are not dumb so can we stop pretending this is a possibility

[🔋Self-Motivated 🔋 @self_motivated0]

For this to be a threat, the hardware these tokens are created on need to compete from a performance standpoint - comparable or better than western performance. China’s hardware is a huge energy suck because they have massive amounts of energy so no incentive to optimize for TCO. Cheaper tokens yes - but in exchange for huge electricity bills. Long term not economical feasible nor will it be with export bans in place for EUVs and the rest of the supply chain. This not a near term issue. In 3-5 years, maybe

[🔋Self-Motivated 🔋 @self_motivated0]

I remember when ya’ll sold. Your thesis and and some YouTube influencers along with my own DD is why I added in the 20,30 and 40s. At the time it was two maybe three analyst covering the stock. I remember reading your rational behind the sell and thought - premature and impatient thinking being communicated here. Sometimes you just got to let your runners, run

[James E. Thorne @DrJStrategy]

Peace through strength.

On his birthday, the President just gave the world a peace dividend.

[Omer Cheema @Omercheema]

How chips are made.

[James E. Thorne @DrJStrategy]

This bull cycle will last far longer than Wall Street expects.

There is no bubble. AI infrastructure and semiconductor demand are still in the early stages, supporting S&P 500 earnings growth toward $650 by 2031.

Inflation is not the issue many claim; a supply shock simply changes relative prices. Why does everyone forget their undergraduate economics?

Warsh is not a hawk.

We are also heading toward a constructive Iran deal that restores risk-on momentum.

Ignore the doomers. This cycle has substantial room to run.

[🔋Self-Motivated 🔋 @self_motivated0]

or maybe they are seeing something the rest of us don't have access to. What they are seeing that triggered them to go to market for $80 billion? more importantly, what did they show Buffett to get him to invest $10 billion after a 200% run. The last time Google did a raise was 2005. What are they up to? This is bigger than a money grab. Something is cooking here and I don't think its what your thinking.