SpaceX neo cloud strategy: Why Anthropic’s claude is turning to Elon Musk’s colossus data centers for critical AI compute

For much of the past two years, Elon Musk’s AI-related public profile has been shaped by the development of Grok and his ongoing rivalry with OpenAI. However, a different shift may now be taking place behind the scenes. According to the host of The AI Daily Brief, Musk has effectively moved into a new role within the AI ecosystem. Rather than competing primarily through AI models, SpaceX is increasingly acting as a provider of large-scale computing infrastructure.


The change centers on SpaceX supplying both its Colossus-1 and Colossus-2 data centers to support Anthropic’s Claude models. The development comes as Anthropic has faced significant compute constraints throughout the year, making additional infrastructure capacity increasingly important.

The host describes SpaceX as a “neo cloud,” a category of companies focused on building and renting AI computing capacity. The business model is straightforward but requires enormous investment.

Neo cloud operators purchase NVIDIA GPUs, combine them with high-bandwidth memory supplied by companies such as Micron and SK Hynix, build or lease large data centers, and then rent that compute power to AI companies through multi-year agreements.

The significance of SpaceX entering this segment extends beyond AI operations. The host argues that the move helps explain the strategic logic behind a potential SpaceX IPO. Public market investors already understand infrastructure-driven businesses and have shown a willingness to assign substantial valuations to companies operating in this space.


According to the host, the shift places Musk in a position where he is focusing on an area in which he has a proven track record: building large-scale physical infrastructure. The host characterizes this transition as Musk becoming a “self-appointed czar of compute.”

Anthropic’s largest financial backer remains Amazon. During the first quarter, Amazon recorded a $16.8 billion pre-tax investment gain related to Anthropic. At the same time, AWS revenue rose 28% to $37.59 billion, marking its fastest growth rate in 15 quarters.

Anthropic has also committed to using up to 5 GW of AWS Trainium capacity. Despite that relationship, Claude users have continued to experience throttling throughout the year.

This situation has highlighted a growing challenge facing advanced AI developers. The issue is no longer only access to capital or model development. Instead, reliance on a single source of computing power has become a major operational risk that can directly affect revenue growth.

As a result, Anthropic has been expanding its infrastructure footprint. The company now spreads its compute requirements across CoreWeave, Google Cloud, and SpaceX’s data centers. SpaceX’s Colossus campuses provide a scale advantage that smaller neo cloud providers cannot easily match.


The growing demand for AI infrastructure has created significant momentum across the broader hardware supply chain, particularly in memory.

Micron Technology has recorded substantial gains this year and recently surpassed a $1 trillion market capitalization. UBS currently maintains a $1,625 price target for the company, while the stock trades at approximately 10 times estimated earnings. That valuation remains the lowest multiple within the Philadelphia Semiconductor Index.

The host points to the sharp rise in AI memory companies, noting that firms such as SK Hynix and Micron have become trillion-dollar businesses. The trend suggests strong investor demand for exposure to the infrastructure components powering the AI industry.

NVIDIA remains at the center of this ecosystem. The company reported first-quarter data center revenue of $75.25 billion, representing 92% year-over-year growth. NVIDIA currently trades at a forward multiple of 26, which now appears relatively restrained compared with some of the neo cloud companies purchasing and deploying its hardware.


Another company attracting attention in the infrastructure discussion is Meta Platforms.

Meta recently increased its 2026 capital expenditure guidance to between $125 billion and $145 billion. Despite the spending plans, the company’s stock has declined year to date as investors question whether those investments will generate sufficient returns.

According to the host, Meta has begun discussing the possibility of becoming a cloud provider itself. The company could potentially rent out portions of the computing infrastructure it is building and “sell back the $130 billion or whatever worth of compute that they’re investing in at a premium.”

Such a model could materially reduce concerns around Meta’s large capital spending commitments. If its data centers can generate neo cloud-style margins, investors may view the expenditures differently than they do today.

The emergence of SpaceX as a neo cloud provider points to a broader transformation occurring across the AI industry. As frontier AI companies push for larger and more capable models, access to computing infrastructure is becoming one of the most important competitive factors.


For Anthropic, diversifying beyond a single compute provider appears increasingly necessary. For SpaceX, supplying infrastructure rather than only developing AI products opens a new role within the market.

The implications may extend even further. The host notes that Meta is exploring cloud opportunities while Jeff Bezos is discussing orbital data centers as an eventual reality. If such facilities become viable, launch providers would be among the few organizations capable of operating them at scale.

Taken together, these developments suggest that the next phase of AI competition may be defined not only by models and applications, but also by who controls the infrastructure that powers them. In that environment, SpaceX’s Colossus data centers position the company as a growing force in the rapidly expanding neo cloud market.