Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124

[ad_1]
On Wednesday, xAI and Anthropic announced amazing collaboration which makes Claude-maker buy “all the computers at (xAI’s) Colossus 1 data center,” about 300MW which allowed Anthropic to raise its user limit immediately. That’s a lot of money for xAI, which is probably in the billions of dollars. Most importantly, it made money immediately one of the most impressive features of the industryturning xAI from a consumer to a provider of compute.
It’s interesting to see this plan as a shot at OpenAI in the midst of ongoing litigation. But Musk’s explanation for X was that xAI was already there moved to a new data center, Colossus 2and xAI didn’t want them both.
In the short term, there is a clear sense of purpose at work. Existing xAI products focus on Grok, which has seen misuse since the photo generation is confusing earlier this year. If xAI’s data center buildout is more than what Grok needs to use, partnering with Anthropic adds more green to the scale. This is especially useful as the company, which is now merged with SpaceX, rushes towards an IPO. In addition, having Anthropic as a customer makes it easier to believe that SpaceX’s orbital data center play. it might work.
But beyond the short-term benefits, the Anthropic partnership sends a strange message about what Elon Musk values ​​most. It shows that the company’s real business may be building data centers rather than training AI models.
It’s rare to see a big tech company doing something like this when companies like Google and Meta, which are also case studies, are building data centers. It’s easy to miss, because many of these companies are operating as AI businesses, web services, and cloud providers at the same time. But when forced to choose between selling existing computers to customers and retaining others to build their own devices, they choose door No.
Last month, Sundar Photosi admitted on a call that Google Cloud costs were lower than they should have been because the company was “complex” – and when given the choice of leasing their GPUs or using them to create AI products, Google chose AI products.
Facebook has faced similar challenges, managing new cloud infrastructure to ensure it has enough GPU power to chase Mark Zuckerberg’s AI ambitions. As he said at the Meta Compute announcement in January, “The way we engineer, fund, and collaborate to build this building is going to be a great opportunity.”
Techcrunch event
San Francisco, CA
| |
October 13-15, 2026
The key word there is “strategic”. Both Zuckerberg and Fox are looking to the future of how AI is helping popular and profitable systems around the world. Computing power is not the only way to meet the needs of today, but to create the things of tomorrow – and missing out on computing means missing out on that opportunity.
Focusing on data centers (global and others), xAI is positioning itself as a neocloud business: buying GPUs from Nvidia and leasing them to emulate developers like Anthropic. It’s a very complex business, squeezed by both chip suppliers and the flexibility of demand. A count of many active neoclouds shows the following: xAI was its value is $230 billion in his part of the income of January; CoreWeave, which controls the amount of computing power, is important less than a third of that.
Musk’s version of the neocloud is ambitious, as you might expect. Some of the data may be in space – at least by 2035, if things go according to plan. xAI is making its own chips in Teraphabwhich will remove some but not all of Nvidia’s pricing power. But nothing changes the basic economics of the neocloud business.
Recently if February and allxAI had real ambitions in software. It was the presentation that unveiled the orbital data center project, but it also teased the big ambitions in writing (with the help of Cursor connection) and interesting ideas such as using computers to become digital twins (unfortunately called the Macrohard project). These are remote projects that require computer hardware to run successfully. As long as xAI is selling more computing power to its competitors, it’s hard to imagine that this new ambition has much of a future.
When you purchase through links in our articles, we can get a little work. This does not affect our authorship.
[ad_2]
Source link