Today’s theme comes from the racetrack. ‘Out of the gate’, and away.

Three first results on the board today. SpaceXAI’s first quarter as a public company. The White House’s AI framework, finished yesterday and showing some of its details today. And AMD’s first quarter since their Helios AI data center launch.

All three are notable n this AI Tech Wave, with the semiconductor and infrastructure movements we have been tracking for days now running through all of them.

(1) SpaceXAI Reports Its First Quarter Post Mega-AI IPO

MP TAKE: We talked about it yesterday, and now it is out. SpaceXAI’s first public quarter, out of the gate.

Remember this stock priced at 135 and ran all the way to the 220s. That was a market cap of almost 2.8 trillion dollars. It is now about 1.5 trillion.

The results held few surprises. Revenue about 7.8 billion dollars, up 92%, a good headline number against expectations of 6.9 billion. The stock reacted a little negative, down 5% after hours.

Recognize that the spending was 2x of that revenue, and that math is going to continue. About 16 billion dollars burned in the quarter on 7.8 billion of revenue, on 18.4 billion of capital spending, mostly the AI data center build.

The CFO reasserted that they are looking at an annualized revenue run rate of over 100 billion dollars by the end of this year, against first half revenues of 12.5 billion. Now remember, the 100 billion is not recorded revenues. ARR is but one month multiplied by 12.

And Elon pointed out that his target by 2030, which is not that far from now, is a trillion dollars in actual revenues, not ARR. Then he added it may occur a year earlier, by 2029. Ambitions and aspirations, definitely aloft.

The other element to keep in mind: SpaceXAI President Gwynne Shotwell reasserted that their direct satellite to phone service would acquire quite a few customers, with the major US cellphone companies being the target. Possibly by 2027, which is earlier than most people would have thought possible. The company did acquire some spectrum of late, and all of that has to be deployed, with more to invest.

Going up against major carriers around the world directly means massive investments in marketing, customer acquisition and maintenance. Apple is focused on this area. Amazon bought Globalstar, which I wrote about separately. A very competitive area over the next three or four years, so the notion of being up and running by next year is on the ambitious side.

Then add the AI data centers and the Terafabs, the chip manufacturing capabilities in space. These are investments very different from the peers. The Mag-7s are investing 150 billion dollars plus each on AI infrastructure. SpaceX has all of that, plus all these other things it also wants to do.

So you have to add them all up and keep tabulating. They are out of the gate, but the calculations will continue to run. And the market now watches the lockup: about 900 million shares, over 120 billion dollars at current valuations, lifting in a few days. A technical factor to mark on the calendar.

Sources:

• Bloomberg, ‘SpaceX Exceeds Revenue Estimates in First Earnings Since IPO’

• WSJ, ‘Elon Musk’s rocket company issues its first quarterly earnings report since IPO’

• Axios, ‘SpaceX tops revenue expectations in first earnings report after IPO’

• The Information, ‘SpaceX reports lots of red ink and makes big promises’

For longtime readers:

• ‘SpaceX/xAI IPO filing outlines Elon’s boundless AI Ambitions’ in AI-RTZ #1066

• ‘Elon Musk boosts SpaceX/xAI pre-IPO story with Terafab’ in RTZ #1034

• ‘Why Elon Might Merge SpaceXAI and Tesla’ in ARD #133

(2) White House AI Framework Leaves US Open Source Alone

MP TAKE: We talked about the framework yesterday, when it was finished and classified. Today we got details we did not have. And one of the core surprises, in a good way, in the context of things I have been writing about.

The White House is exempting US open source models from the government review.

The focus lands on the safety governance elements around the frontier closed models, mostly from Anthropic, OpenAI and others. Remember, Anthropic has been raising its hand saying regulate us, put safety guardrails on. So has OpenAI. They got their wish. And the government did not include the US open source models.

And no mention of Chinese open source models, which has been an issue. Jensen Huang put a coalition together, which I wrote about a few days ago, that basically said the US tech industry could benefit by accessing Chinese and global open source AI models.

The details worth having on the record, via the reporting. Covered frontier models are defined as closed-source, with state-of-the-art capabilities and national security risks, with no clear definition of either test. Government access runs up to 30 days before release, in high-security environments with access logs. The benchmarking stays classified, and the framework itself stays unpublished: details only for the companies in the process.

So far the administration is threading the needle, which is what I was hoping would be the case. The center route, given the multiple voices around these issues.

That seems to be where we are coming out, out of the gate, on this AI framework. A lot to monitor, a lot to understand. We will come back with updates.

Sources:

• WSJ, ‘White House’s AI Guidelines Exempt US Open Models from Government Review’

• Axios, ‘Inside Trump’s AI framework’

• Axios, ‘White House plans to keep AI framework under wraps’

For longtime readers:

• ‘Nvidia Assembles AI Avengers, Open Secure AI Alliance’ in AI-RTZ #1161

• ‘Anthropic flagged to a screeching Halt Globally by the US’ in AI-RTZ #1117

• ‘The White House Finishes Its AI Framework, and Classifies It’ in ARD #133

(3) AMD’s Good News, Bad News

MP TAKE: AMD, Nvidia’s nemesis for 30 years in graphics chips and now in AI data center chips, reported its quarter. Good news and bad news.

The good news first, because there was plenty. The numbers were as expected, and they beat expectations a little. Revenue came in about 11.5 billion dollars, up 50%.

And their new Helios AI data center product, which I wrote about when it launched, has gotten good reception with early customers. OpenAI, Oracle, Meta, potentially Microsoft. This was the first quarter out of the gate from that launch.

The surprise was Elon Musk, on his own quarterly call, saying SpaceX would focus exclusively on Nvidia and their chips for its data centers. Earlier there had been intimations that they would potentially use AMD as a second source. That is not to be the case. The stock was down about 8% in the aftermarket, and we will see how it opens.

It is tough being a second source. You do not win them all.

My take is that AMD does have a lot of opportunities, with a lot of other players around the world that could be customers. It is just that Elon is very visible and very vocal, and this public announcement of going with Nvidia only is a speed bump they will have to navigate through. We will see how that goes.

Sources:

• WSJ, ‘AMD Shares fall 8% as Elon Musk commits to Nvidia chips for SpaceX’

• Bloomberg, ‘AMD Shares slide after AI Growth Outlook underwhelms Investors’

• Barron’s, ‘AMD stock slides after solid earnings report’

For longtime readers:

• ‘AMD’s Helios up against Nvidia’s Vera Rubin & beyond’ in AI-RTZ #1154

My overall take on all of these: they are all out of the gate items.

The first public quarter for SpaceXAI. The White House now public, somewhat, with its classified AI framework. And AMD with its first quarter post the Helios launch, now having to run really hard to capture its share of the burgeoning market for data center infrastructure.

Against, of course, market leader Nvidia, which is out there with 80 plus percent market share on a global basis.

They are all out of the gate. But it is a long race. So we will see how that goes.

GADGET AI

Apple’s MacBook Neo Keeps Beating Windows Laptops in the ‘RAMageddon’ Age

MP TAKE: A topic I have talked about before, with new data. The Verge ran a detailed analysis of Windows laptops against Apple’s MacBook Neo, the value notebook.

The Neo, until they raised the price about a hundred dollars, was about a 500 dollar item. Now it is about 600. It uses an iPhone chip with eight gigabytes of memory.

The reason all the detail matters, as I discussed in my ‘RAMageddon ‘piece: we are going to see a very noticeable decline in both Windows PCs and Android phones over the next couple of years. The burgeoning increase in prices for RAM and solid state drives is shifting a lot of the memory supply to AI data centers.

As a result, Windows is more disadvantaged relative to Apple. Apple has a very robust supply chain. They had to raise prices as well, but they are less negatively impacted than the Windows computers.

The Verge evaluated three laptops in the 500 to 700 dollar price range, with a lot of detail, as we go into the second half and the school season. All of them relatively lacking against the Neo on overall value proposition. If you are in the market for a laptop, do take a look at the review. It is very well done.

We are going to see a lot more comparisons and reviews like this. Not just Windows computers, but on the smartphone side too, Android against Apple’s iPhones and iPads. And recognize that Apple announces its next generation products this September. All of this will continue to get reset.

Sources:

• The Verge, ‘I tested three Windows laptops in the MacBook Neo’s price range. There’s no contest’

For longtime readers:

• ‘Apple’s Macbook Neo a new Arrow in its AI quiver’ in RTZ #1017

• ‘’RAMageddon’ really here to stay’ in AI-RTZ #1145

• ‘Apple’s supply chain locks out most tech memory pressures’ in RTZ #1010

Q1. What is my favorite feature of the Apple Neo vs Windows?

The out of the box experience. Apple’s computers, when you buy them fresh, give you the new computer feel, the ‘new car smell’ as it were. They launch, and you are up and ready to work in five minutes.

Any brand new Windows laptop, and I have several over there, typically takes about an hour to launch. It updates forever. Both the operating system from Microsoft and all the third party software from the OEMs that make the individual laptop.

There are a lot of YouTube videos on this if you are interested. And it continues when you reboot these computers as well. The Mac startup speed is a signature feature I am rather attached to.

Q2. What is the worst feature of Windows laptops, post those boot up times?

The advertising. Everything from Windows to the third party applications is constantly trying to get you to update and upgrade things while you are trying to get work done on the laptop. Upgrade this. Take a look at that promotional offer. Your computer needs protection. They are constantly looking for subscriptions, which is a way they make up the margins on these machines.

I understand the economic model. It is just a very bad user interface and user experience. The Macs do not do any of that.

Add to that the stickers. Every Windows computer comes with a whole bunch of promotional stickers from third parties. The OEMs get paid for that, and that is their margin. Tough to remove, and aesthetically they do not look very good. Macs come out of the box clean.

To be fair, I use Windows machines for playing games on the higher end GPU machines from Nvidia. But for daily computing, the Mac is racing ahead. In this era of RAMageddon, they are able to get out there, out of the gate, a little bit further. No contest indeed.

WRAP

Today’s AI-RTZ #1169 is on the pricing of AI models. There were a couple of articles basically saying AI models are racing to zero in pricing. That is, to my mind, a little bit of an exaggeration. What is happening is the pricing is widening, on the high end and on the low end, and there are important dynamics to understand in that. If you are interested, do take a look at that post today.

AI Ramblings Daily on AI-RTZ is here to think through AI and reset. Together.

Tomorrow, ARD 135 and AI-RTZ #1170.

Thanks for joining us today, AI Curious Folk. Stay tuned.

— MP

Full Source Reading

For the broader context, see the canonical sources for ARD 134, in today’s narrative order:

Event 1. SpaceXAI’s First Public Quarter

• Bloomberg, ‘SpaceX Exceeds Revenue Estimates in First Earnings Since IPO’

• WSJ, ‘Elon Musk’s rocket company issues its first quarterly earnings report since IPO’

• Axios, ‘SpaceX tops revenue expectations in first earnings report after IPO’

• The Information, ‘SpaceX reports lots of red ink and makes big promises’

• AI-RTZ #1066, ‘SpaceX/xAI IPO filing outlines Elon’s boundless AI Ambitions’

• RTZ #1034, ‘Elon Musk boosts SpaceX/xAI pre-IPO story with Terafab’

• ARD #133, ‘Why Elon Might Merge SpaceXAI and Tesla’

Event 2. The White House AI Framework Details

• WSJ, ‘White House’s AI Guidelines Exempt US Open Models from Government Review’

• Axios, ‘Inside Trump’s AI framework’

• Axios, ‘White House plans to keep AI framework under wraps’

• AI-RTZ #1161, ‘Nvidia Assembles AI Avengers, Open Secure AI Alliance’

• AI-RTZ #1117, ‘Anthropic flagged to a screeching Halt Globally by the US’

• ARD #133, ‘The White House Finishes Its AI Framework, and Classifies It’

Event 3. AMD’s Good News, Bad News

• WSJ, ‘AMD Shares fall 8% as Elon Musk commits to Nvidia chips for SpaceX’

• Bloomberg, ‘AMD Shares slide after AI Growth Outlook underwhelms Investors’

• Barron’s, ‘AMD stock slides after solid earnings report’

• AI-RTZ #1154, ‘AMD’s Helios up against Nvidia’s Vera Rubin & beyond’

Gadget AI. MacBook Neo vs Windows Laptops

• The Verge, ‘I tested three Windows laptops in the MacBook Neo’s price range. There’s no contest’

• RTZ #1017, ‘Apple’s Macbook Neo a new Arrow in its AI quiver’

• AI-RTZ #1145, ‘’RAMageddon’ really here to stay’

• RTZ #1010, ‘Apple’s supply chain locks out most tech memory pressures’

Clips from today

Clip 1. White House Exempts US Open Source AI

The White House is exempting US open source models from the government review.

MP Take: Anthropic has been raising its hand saying regulate us, put safety guardrails on. So has OpenAI. They got their wish, and the government did not include the US open source models. And no mention of Chinese open source models, which has been an issue. Jensen Huang put a coalition together arguing the US tech industry could benefit by accessing Chinese and global open source AI models. So far the administration is threading the needle, the center route, which is what I was hoping would be the case.

Clip 2. Windows vs MacBook: Startup Speed

Apple’s computers, when you buy them fresh out of the box, launch and you are up and ready to work in five minutes.

MP Take: Any brand new Windows laptop typically takes about an hour to launch. It updates forever, from the Microsoft operating system to all the third party software from the OEMs. And it continues when you reboot. This has been a core Windows complaint for a long time, and the Mac startup speed is a signature feature I am rather attached to. In the RAMageddon age, with memory prices squeezing every PC maker, the value gap is widening. For daily computing, the Mac is racing ahead, out of the gate a little bit further.

Clip 3. SpaceX’s $100B ARR Target by 2026

The investments are running 2x of revenues, and that math is going to continue.

MP Take: The CFO reasserted an annualized revenue run rate of over 100 billion dollars by the end of this year, against first half revenues of 12.5 billion. Remember, the 100 billion is not recorded revenues. It is one month multiplied by 12. And Elon pointed to a trillion dollars of actual revenues by 2030, maybe a year earlier by 2029. Ambitions and aspirations, definitely aloft. The market now watches the lockup, about 900 million shares, over 120 billion dollars at current valuations, lifting in a few days.

Clip 4. Elon Musk’s Exclusive Nvidia Stance

It is tough being a second source. You do not win them all.

MP Take: AMD beat expectations with revenue about 11.5 billion dollars, up 50%, and good early reception for Helios with customers like OpenAI, Oracle, Meta and potentially Microsoft. The surprise was Elon Musk on his own quarterly call, committing SpaceX’s data centers exclusively to Nvidia, after earlier intimations AMD could be a second source. AMD has a lot of opportunities with a lot of other players around the world. But Elon is very visible and very vocal, and this public announcement of going with Nvidia only is a speed bump they will have to navigate through.

About AI Ramblings Daily (ARD), and AI-RTZ

Both are daily. Both are free. Both are about AI. But they’re different mediums carrying different messages.

AI-RTZ is the morning text, a deeper written take on one idea, published by at least 5 AM EST. Today: post #1169.

AI Ramblings Daily is the afternoon video + podcast, my ad hoc takes and perspective on the day’s AI issues and news flow, around 20 minutes, with short 1-2 minute clips for quick topic views. Today: episode #134.

Subscribe to either or both on michaelparekh.substack.com. They run as separate Sections you can opt into or out of.

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here.)