To The Moon

The AI race is entering a new phase.

The biggest AI companies are no longer just competing to build better models.

They are competing for something even more fundamental:

Compute.

And an unexpected trend is emerging:

Some of the world’s largest technology companies are becoming landlords of AI infrastructure, renting computing capacity to the very companies challenging them in AI.

1. Meta and SpaceX Are Becoming Compute Landlords

For months, investors questioned whether massive AI infrastructure spending would generate enough returns.

$Meta Platforms, Inc.(META)$ ’s answer may be emerging:

Don’t just consume compute.

Own it—and monetize it.

Meta is reportedly in discussions with Anthropic for a potential deal that could reach $10 billion over two years, allowing Anthropic to access computing capacity from Meta’s AI infrastructure.

The significance is not only the size of the deal.

It is what it represents.

A company once viewed as a social media platform is potentially becoming a major AI infrastructure provider.

The market initially interpreted Meta’s expanding data-center plans as a sign of possible "excess capacity."

But if frontier AI companies are actively seeking access to Meta’s infrastructure, the message is different:

Compute is not abundant. It is scarce.

$SpaceX(SPCX)$ is following a similar path.

After building massive AI computing infrastructure through its integration with xAI, SpaceX has moved beyond simply using compute internally.

It is also positioning itself as a provider of AI infrastructure.

The company has reportedly secured large-scale compute agreements with AI customers, including Anthropic, while exploring additional opportunities such as government AI workloads.

The strategy is becoming clear:

Build large-scale data centers.

Secure power.

Acquire GPUs.

Rent compute capacity.

Turn infrastructure into recurring revenue.

2. Anthropic Shows How Valuable Compute Has Become

The other side of this trend is even more important:

AI labs are running out of affordable, available compute.

Anthropic is becoming one of the most aggressive buyers of computing resources.

The company has been expanding access to compute through multiple partners rather than relying on a single provider.

Its agreements show that frontier AI development is increasingly becoming a resource allocation problem.

The bottleneck is no longer only:

"Who has the best model?"

It is also:

"Who has enough chips, power, and data-center capacity to train and run those models?"

This creates a new dynamic.

AI companies that compete against each other at the model layer are becoming customers and suppliers at the infrastructure layer.

Anthropic can compete with Meta, Google, OpenAI and others on AI products.

But all of them still need the same scarce resource:

Compute.

3. The Rise of the AI Compute Economy

The most important trend may not be any individual deal.

It is the emergence of a new asset class:

Industrial compute leasing.

Just as cloud computing transformed servers into a utility business, AI may transform advanced computing capacity into a strategic infrastructure market.

The emerging ecosystem looks increasingly complex:

• Chip companies provide the hardware.

• Data-center operators provide the infrastructure.

• Power providers support expansion.

• Cloud companies package computing access.

• AI labs consume enormous amounts of compute.

The companies owning scarce compute capacity may become the next generation of infrastructure landlords.

4. What Does This Mean for AI Investors?

The AI investment story is gradually shifting.

The first phase was about:

Who builds the smartest models?

The second phase is becoming:

Who controls the computing resources behind those models?

This explains why companies such as Meta and SpaceX are investing aggressively in data centers.

They are not simply preparing for future AI products.

They may also be building a new revenue stream.

The Bigger Question

The AI race may eventually be decided not only by model performance.

It may be decided by control over the scarce layer underneath:

Compute.

The winners could be the companies that own:

• the GPUs,

• the energy,

• the data centers,

• and the contracts that turn computing capacity into long-term cash flow.

The AI industry is slowly evolving from a software race into an infrastructure economy.

And the biggest question investors should watch is:

Will the future belong to the companies creating AI—or the companies owning the machines that make AI possible?