Even ifColossus 2 is the new training cluster, handing Anthropic 220k GPUs frees Anthropic to then redirect its own compute toward training and research is not what you’d do in a tight race for AGI.
But I think it’s not at all trivial to integrate a company, take on all its debt and obligations, change its corporate reporting structure with a very distinct mission, while continuing to do frontier AGI development that’s going to take off.
It just doesn’t seem very characteristic of Musk to throw in the towel like that. They do seem to be falling behind, but I think the intent is still to have a frontier lab.
Idk, I could possibly see it through the lens of “Musk has decided that the future will be owned by those who control the hardware, not those who make the best software (since software stuff is an easier, less bottlenecky, and more easily commoditized task).”
I’m not sure if this hardware focus is correct, but it’s at least plausible, and it must be very tempting for Musk to believe (with so much hardware expertise / experience, a whole career based around winning at hard physical engineering & integration challenges, etc!). No fussy issues with mechahitler or white genocide in south africa, no annoying nuances where you do a bigger training run but end up producing a worse model than your competitors. Just good old-fashioned rapidly ascending the kardashev scale with expanding shells of massive solar-powered orbital datacenters. Hence the “macrohard” jokes, the xAI integration with SpaceX that doesn’t make any sense from a software perspective, etc.
Note this is very different from “dropping out of the AGI race”. If this is indeed Musk’s perspective, he is still hell-bent on racing to AGI (see also the new chip-foundry plans and potential intel partnership, etc). But he might be dropping out of the “frontier AI lab” race, which after all seems so hopelessly competitive (as soon as one lab pulls ahead, the others seem to catch right back up! or at least this has been the case so far). Better, he might be thinking, to focus on a more neglected “vertically integrated, AGI-pilled hardware scaling” strategy that might pay off bigger in the long term.
I agree this sounds plausible and appreciate with the distinction you’re drawing between still competing for the future, but no longer racing for the software intelligence explosion AGI.
I never saw xAI as really in “the race”. For me it’s always been OpenAI vs Anthropic vs GDM.
And seeing you connect the dots that xAI could become a GDM-like lab inside SpaceX totally makes sense. I think Musk first want to IPO SpaceX then will come the merger with Tesla. And somewhere in between now and end of 2027 I could see xAI become a GDM-like lab.
The other route I see possible is xAI becoming more like Apple and rely on partnerships. Apple failed hard with Siri and now is relying on Gemini to power their own AI.
The deal with Cursor. Then the deal with Anthropic. Something is cooking.
xAi seems like they are dropping out of the frontier AGI race.
“xAI will be dissolved as a separate company, so it will just be SpaceXAI, the AI products from SpaceX.” (Musk on X, May 6)
xAI was acquired in Feb 2026 through a reverse triangular merger, keeping it separate with its own debt, liabilities, and structure. It’s now being folded into SpaceX as a division focused on integrating AI across the SpaceX stack.
Also 11 of the 12 original co-founders have left, Grok DAU fell from 13.9M to 12.2M March→April while Claude went 16M→23M, and xAI burned $7.8B in the first 9 months of 2025.
Letting Anthropic use all of Colossus 1 for inference, for an estimated $3-4B/yr.
Even if Colossus 2 is the new training cluster, handing Anthropic 220k GPUs frees Anthropic to then redirect its own compute toward training and research is not what you’d do in a tight race for AGI.
It’s possible this is all a play for the IPO, to get a better narrative for raising huge sums, and that they turn xAI into a GDM like organization. And Colossus 2 still has the biggest compute capacity of any cluster in the world, so it’s certainly can train big models.
But I think it’s not at all trivial to integrate a company, take on all its debt and obligations, change its corporate reporting structure with a very distinct mission, while continuing to do frontier AGI development that’s going to take off.
Also filings haven’t reflected this yet because the only one that exists is the April 1 confidential draft. The public S-1 is expected the week of May 18-22, working back 15 days from the targeted June 8 roadshow.
It just doesn’t seem very characteristic of Musk to throw in the towel like that. They do seem to be falling behind, but I think the intent is still to have a frontier lab.
Idk, I could possibly see it through the lens of “Musk has decided that the future will be owned by those who control the hardware, not those who make the best software (since software stuff is an easier, less bottlenecky, and more easily commoditized task).”
I’m not sure if this hardware focus is correct, but it’s at least plausible, and it must be very tempting for Musk to believe (with so much hardware expertise / experience, a whole career based around winning at hard physical engineering & integration challenges, etc!). No fussy issues with mechahitler or white genocide in south africa, no annoying nuances where you do a bigger training run but end up producing a worse model than your competitors. Just good old-fashioned rapidly ascending the kardashev scale with expanding shells of massive solar-powered orbital datacenters. Hence the “macrohard” jokes, the xAI integration with SpaceX that doesn’t make any sense from a software perspective, etc.
Note this is very different from “dropping out of the AGI race”. If this is indeed Musk’s perspective, he is still hell-bent on racing to AGI (see also the new chip-foundry plans and potential intel partnership, etc). But he might be dropping out of the “frontier AI lab” race, which after all seems so hopelessly competitive (as soon as one lab pulls ahead, the others seem to catch right back up! or at least this has been the case so far). Better, he might be thinking, to focus on a more neglected “vertically integrated, AGI-pilled hardware scaling” strategy that might pay off bigger in the long term.
I agree this sounds plausible and appreciate with the distinction you’re drawing between still competing for the future, but no longer racing for the software intelligence explosion AGI.
Very thoughtful analysis.
I never saw xAI as really in “the race”. For me it’s always been OpenAI vs Anthropic vs GDM.
And seeing you connect the dots that xAI could become a GDM-like lab inside SpaceX totally makes sense. I think Musk first want to IPO SpaceX then will come the merger with Tesla. And somewhere in between now and end of 2027 I could see xAI become a GDM-like lab.
The other route I see possible is xAI becoming more like Apple and rely on partnerships. Apple failed hard with Siri and now is relying on Gemini to power their own AI.
The deal with Cursor. Then the deal with Anthropic. Something is cooking.