To The Moon

This is the biggest scam ever and obviously is lead by Elon Musk's long time buddy Adam Jonas who would sell his mother for SpaceX share sells from his clients and other people who can't do math and need video summaries like this one and Wall Street Bets "advice". Read the SpaceX S1 prospectus on what their plan and expectations is for reaching profit. Then think twice because you might now live long enough to ever see this happening. Not to mention this is all build on the fragile and overhyped AI cycle in an overcrowded market where no one, for sure not Grok, is making any profit in the next 5 years, if ever. Because AI is a commodity, you can easily switch from provider to provider or use Openrouter to run several of them with no effort at all.

Guess who isn't getting more energy? The same one who will pay more for all this energy consumption surge whilst cheap renewable energy isn't on the agenda of the current administration (in favor of the oil, gas and nuclear industry)

Still the biggest scam ever, no matter what the name. Starships won't be able to transport the payload, if lucky maybe half of it. AI in Space is a joke and will be quietly stopped (as many of Musks previous heroic projects like wireless charging, using solar energy at every Supercharger, Dojo, 4280 battery cells everywhere etc).

Legend

Spot on article and analyst review. Let's not forget their core business which is still selling ads and Google is doing an amazing job there with their AI approach. It doesn't need to compete with OpenAI/Anthropic or Chinese models for AI leadership at all.

The majority of SpaceX valuation comes from their AI projects like Microhard and so. Starlink and their rocket business is a tiny part of the valuation. In the AI business SpaceX has achieved nothing apart from creating overcapacity in their data center which they rent out for a small margin vs their huge debt to build it. Don't trust analysts who helped on the IPO and are part of Musk's scam business! Do your due diligence yourself and check their profit (easy, there's none), debt (huge, low quality bonds) and growth projection (in all three business areas).

Commitment to leading the sector? I can commit to anything but let's be realistic and fact based. They are so far behind Waymo, in revenue, number of fully autonomous vehicles (which actually serve more than two people and two duffel bags), locations (including London now) and miles to critical incident. Then there's even Zoox which is backed by a dude called Jeff Bezos who apparently has some money to spent. It's on the road for a while and also has no steering wheel as it's the first purpose build autonomous taxi (not Tesla like Musk said).

It will be interesting to see their S-1, to me the revenue increase isn't so much of interest vs their prediction as to when they will achieve profit. The rise of the cheaper Chinese models (supported by the Chinese Governments push for domination in that area as well) and thus the pressure on Anthropic and OpenAI to lower their prices is what is worrying me. Anthropic has been able to get a large portion of the "coding" companies (unless you're deeply integrated with Microsoft) and has been able to switch over users from OpenAI. However, this comes at a cost, their loss with the subscription models is huge and they can't easily switch to API cost as that would lead to an exodus of their user base (those that are doing AI engineering at least). Beyond this use case nothing else has really

I'm running the GLM model 5.3 for adversarial agent duties like code reviews and to check the plans that Claude Code Opus 5 comes up with. I'm positively surprised about the quality of the models work and responses, in almost every review it finds flaws in Opus 5 plans (which is more a testament to low quality work that Claude's frontier model puts out regularly). Token usage is a bit on the higher side but response time is good, especially given the large repositories and complex infrastructure I run. The company seems like a valid competitor in the AI engineering landscape.

While it’s a smart Jensen Huang move, I do see a big risk for anyone who is coming late to the dinner table. Even right now, us on the ground using Frontier models and also exploring the capabilities of the cheap, Chinese open source models can foresee that this isn’t going well for OpenAI and Anthropic. I’m using GLM 5.2 right now and not for adversarial agents to correct all the errors in the plan and even in the implemented code that Claude Code’s Opus 5 on XHigh or Ultracode produces! The advantage is little and doesn’t justify the 10-20x price difference and Opus 5 is slow as hell. So all those investments in NVIDIA chips will depreciate fast and users will just switch to the cheaper and better option. Other than eg on cars, countries like the US can’t block OpenRouter etc to th