Their growth is insane. The operating income (not revenue which is double digits, but irrelevant) in 2025 was 4,4B and with the growth they have, they are profiting multiple billions a year.
First "breakeven cash flow" does not equal profit. But that's just Musk, as the CEO (and thus accountable for financial statements he makes to the SEC), cannot actually say profit.
More generally, never believe any kind of financial engineer when they say this. It's bullshit. Guaranteed. Why? Simple: if it was true, they would never tell you. They would take a shovel and put their own money in, they would charge large banks to let them put their money in (yes, seriously). They'd literally charge other rich people money to let them put their money in ... You would hear about it 20 years after the fact in an interview done after retirement, and only then.
Specifically, what's wrong here Starlink pays for it's own launches in stock (they pay by issuing stock to SpaceX). If you think about it, that's creating money out of nothing:
SpaceX holds Starlink debt, and counts that as assets (as a positive amount)
Starlink issues and "succesfully" sells shares to SpaceX, and counts that as income (as a positive amount)
So you see what's happening here: these are two positive numbers, but ... obviously have to sum to zero! After all if I buy an ice cream from you for 5$ the change in my assets and the change in your assets has to sum to zero (-5$ for me, +5$ for you = 0$ for both of us together). Not for SpaceX and Starlink. SpaceX gives rocket launches to Starlink and the sum of both their assets is billions!
If you look at actual cash flow, Starlink's last quarter was 4.4B operating income, minus 2.4B depreciation and amortization, minus 4.17B capital expenditures. How do you calculate the company worth? Easy capital expenditures are zero (because they "convert" capital to goods, which can be resold, except of course they can't really).
So for "total assets" on the financial statements that's 4.4B - 2.4B PLUS 4.17B per year.
For actual cash or resources, it's obviously 4.4B - 2.4B - 4.17B or MINUS 2 billion per year.
Which one is true depends on whether SpaceX can succesfully sell Starlink's shares at the internal valuation they're using. Obviously the safe accounting value is the one with the minus sign. Or to put it differently: Starlink has "increased in value" by 2 billion, so investors (for now SpaceX) need to give it 2 billion to continue operating.
That's an incredible achievement. But you seem to be ignoring the goals of Starship program - to make economically viable and reusable means of space transportation.
It's like after Jobs unveils iPhone to the world you'd comment that, meh, we could make telephone calls more than 100 years ago?
I wonder how this type of comment would apply to software development world where something innovative is being built, currently not ideal but the devs would just look at it: Meh, it doesn't do X, it doesn't do Y, it does Z worse and it doesn't send email!
Maybe AI applies here - from a few years of supercharged intellisense to builds stuff on its own. But even here I don't remember such a skepticism some folks are aiming at Starship.
It's like running into R&D department and just pointing out at the flaws/not implemented features while you iterate on your product. Annoying.
AI and the success of agentic coding is an excellent example. It works because validation is external to the AI coding process. If it had to validate itself it wouldn't work. Luckily, humans created a lot of tools to validate the work of humans. If your use case doesn't match the characteristics of agentic coding, don't expect the same success. That kind of serendipity is rare when conditions for it are good, and the conditions for serendipity around rockets is never that good.
Mine doesn't. It mostly adds and keeps referencing previous decisions. Correctly, referring to them as being overridden. Even tweaking the original text or occasionally deleting a part. But text would look differently had it been written from scratch.
Unless a LLM works truly as a compiler, there will be a drift between the code and the spec.
For me they mostly append the end of the file with more sections or details. Heck it often adds very specific implementation details that it worked on recently that have no business being written in a repo wide summary.
Weird. I changed/amended model/integration and it updated table, provided explanation.
But yeah, I like to reference the docs so it can refresh his memory. But I may have had the .md files updated even if I didn't explicitly provide/ask for it, but a refactor that touched particular areas.
But yeah, my .md references others, it also maybe helps that my dir is named docs/ and it lists it somewhere in .md files... maybe its just the way it is structured helps infer agent what to do.
Don't know about US, but I work within European Datacenter where we intend to have excess heat to be used for residential building heating [1]. You must be in proximity of the consumer for this to be possible. Not necessarily center is needed, but even then, DCs can be pretty :)
For example our office is heated using heatpump that has >22C of water heated by equipment as a source.
> excess heat to be used for residential building heating
In the late 90s I worked at an insurance company in leafy Surrey[1]. I was told that they used to use their IBM mainframe to heat the swimming pool[2] effectively for free; when they upgraded they had to buy and run a heating system for the pool as the new mainframe didn't pump out enough heat.
[1] In England, south west suburbs of London to be precise.
[2] yes it had an indoor pool, badminton courts, outdoor tennis courts and other facilities - a great place to work.
Thousands of people Google "why is my building cold" causing the AI DC to wake up and pump out heat. AI sees the heat and reports "heating is on, you're fine" then shuts down.
The fascinating thing is that it wasn't a primary part of the mission but got lots of excitement. For me it was the thing that it features many off-the shelf parts from smartphones, uses linux and other open source software and it beat the expectations on how long it will service - Planned ~30 days with 5 flights and it did almost 3 years with 72 flights. https://www.nasa.gov/news-release/after-three-years-on-mars-...
Reviewing and reading everything makes this feeling, yeah.
However where I can say 100% it saves time is discovery by answering these questions:
- At what state does bug X manifests?
- Explain how integration/feature works.
- I want to integrate with system X: Audit what items/features are used and what model changes are needed on my side
No.
Starlink profits are positive since 2023: https://x.com/elonmusk/status/1720098480037773658
Their growth is insane. The operating income (not revenue which is double digits, but irrelevant) in 2025 was 4,4B and with the growth they have, they are profiting multiple billions a year.
reply