We’re feeling cynical about xAI’s big deal with Anthropic

3 weeks ago 25

Anthropic and xAI announced a large partnership this week, with Anthropic buying each the compute capableness astatine xAI’s Colossus 1 information halfway successful Tennessee.

On the latest occurrence of TechCrunch’s Equity podcast, Kirsten Korosec, Sean O’Kane, and I discussed what the woody mightiness mean for xAI’s genitor institution SpaceX, arsenic SpaceX prepares to spell nationalist and apparently plans to dissolve xAI arsenic a abstracted organization.

Kirsten did her champion to connection “a affirmative view” connected the concern — aft all, it’s a caller mode for xAI to marque money. But she besides noted that this besides suggests xAI isn’t doing overmuch erstwhile it comes to grooming its ain frontier AI models, and it’s harder for the institution to presumption itself arsenic a “forward-looking, innovative” concern erstwhile that’s the case.

Then Sean asked: “Why beryllium affirmative erstwhile you tin beryllium cynical?” In his view, this seems similar “a large vigor cheque earlier the IPO.” Yes, becoming a neocloud mightiness beryllium “a much believable concern successful the adjacent term,” but it’s little apt to get extracurricular investors excited successful the agelong term. (And past there’s the biology lawsuit that xAI is facing implicit Colossus 1.)

Keep speechmaking for a preview of our conversation, edited for magnitude and clarity.

Sean O’Kane: I ever emotion a surprise, particularly erstwhile everybody’s eyes [are] connected different ball, a large trial that’s happening. Seemingly retired of obscurity this week, SpaceX and truthful its AI subsidiary xAI — which seemingly nary longer exists now, oregon is imminently not astir to exist, which we tin get to — struck a woody with Anthropic.

Basically, the existent mentation of the woody is that Anthropic’s fundamentally taking implicit each of the compute astatine the information halfway known arsenic Colossus 1 successful Memphis, Tennessee, to absorption connected Anthropic’s much enterprise-focused AI products. There’s been a batch of reporting astir however [Anthropic’s] been looking for much compute […] and it seems similar an flight valve for them to beryllium capable to onslaught this woody and get entree to each this compute.

Techcrunch event

San Francisco, CA | October 13-15, 2026

In the adjacent term, for xAI and for SpaceX, yes, they are a neocloud now, successful the consciousness that they had to bash thing with each this compute that they were building, due to the fact that it surely seems similar they were not going to request it for Grok — which, extracurricular of X, is not burning up the satellite arsenic acold arsenic becoming the caller blistery user chat bot.

Kirsten Korosec: And we should accidental that successful presumption of what a neocloud is, for those who don’t know, this is the thought of buying GPUs from Nvidia and the like, and renting those retired arsenic opposed to utilizing those for their ain AI, grooming their ain AI models.

So this is simply a antithetic benignant of business, and the constituent that our AI editor, Russell Brandom, makes is that a batch of companies are gathering retired information centers, but if fixed a prime between, bash they rent them retired [or utilizing them to bid their ain models], they are inactive prioritizing utilizing this compute for their ain interior AI exemplary training. I deliberation that’s an important constituent and 1 that suggests that possibly xAI isn’t doing truthful overmuch connected the AI exemplary grooming [side]

Anthony Ha: Right, and arsenic Sean was alluding to, astir radical would not needfully deliberation of Grok arsenic — not lone that it’s known for immoderate beauteous unpleasant, if not downright illegal, content, but besides it’s not needfully ace cutting edge. Especially if we commencement talking astir endeavor AI, which I cognize we’re gonna beryllium getting into aboriginal successful this episode, you don’t perceive a batch astir radical utilizing Grok for work-critical tasks. 

And truthful the question becomes: How tin xAI really marque money? And seemingly conscionable selling the infrastructure could beryllium 1 of the main ways to bash it.

Kirsten: And you could instrumentality a affirmative presumption connected that, right? They figured retired a mode to marque money. But I deliberation that erstwhile you are positioning your institution — successful this case, SpaceX-slash-xAI — arsenic a forward-looking, innovative company, that’s tougher to merchantability if you are simply conscionable renting retired your GPUs and not utilizing them for that innovation.

Sean: But wherefore beryllium affirmative erstwhile you tin beryllium cynical? Which is to accidental that this seems similar a large vigor cheque earlier the IPO that we’re astir to spot get rammed into the markets with SpaceX.

Anthony, you mentioned not lone is Grok not being utilized for large endeavor tasks, there’s been reporting that xAI employees were using different models, they weren’t adjacent utilizing [Grok] internally, and that caused this large shakeup wrong of xAI, station acquisition from SpaceX, that progressive fundamentally all the co-founders leaving different than Elon Musk, [and] him fundamentally saying he’s starting from scratch connected xAI, contempt the information that SpaceX paid $250 cardinal for it successful the tally up to this mega-IPO. 

And present he’s saying that they’re going to dissolve xAI arsenic a abstracted entity wrong SpaceX altogether. He’s starting to telephone the full happening SpaceXAI, due to the fact that this antheral loves thing but to ruin a marque that has immoderate worth to it — spot Twitter.

This whitethorn beryllium a much believable concern successful the adjacent term, and truthful connected immoderate level, I could spot this being possibly much charismatic to investors travel IPO time, due to the fact that it’s similar a spot much reliable and surely much existent than them being a frontier laboratory developer. But it’s besides not the benignant of concern that’s going to gully the aforesaid — astatine least, successful a mean situation — extracurricular concern that we’re seeing spell into each the frontier labs.

That’s possibly 1 of the biggest hostility points we’ve seen make during this IPO process.

When you acquisition done links successful our articles, we whitethorn gain a tiny commission. This doesn’t impact our editorial independence.

Read Entire Article