After years of public criticism, Elon Musk’s AI company xAI has struck a major infrastructure deal with rival Anthropic, giving the maker of Claude AI access to one of the world’s most powerful supercomputing systems.
The agreement is remarkable not just because of its scale but because of who is involved.
Musk has repeatedly attacked Anthropic in the past, accusing the company of ideological bias and mocking its approach to AI safety. Now, the same company appears set to rely on xAI’s massive computing infrastructure to train and operate its flagship AI systems.
In the AI arms race, yesterday’s ideological enemy can quickly become today’s electricity provider.
What is the xAI-Anthropic deal?
Under the agreement, Anthropic will reportedly gain access to the full capacity of xAI’s Colossus 1 supercomputer in Memphis, Tennessee.
The system is said to contain more than 220,000 Nvidia processors, making it one of the largest AI computing clusters ever assembled.
The deal could reportedly provide Anthropic with:
- Roughly 300 megawatts of additional compute capacity
- Higher compute density for Claude AI
- Faster model training and inference
- Expanded AI scaling capabilities
Anthropic is reportedly expected to pay between $3 billion and $4 billion annually under the arrangement.
That transforms xAI from merely an AI model developer into something potentially even more valuable: an AI infrastructure landlord.
Why compute power is becoming the new oil in AI
Modern AI systems are constrained less by ideas than by compute.
Training frontier AI models now requires staggering amounts of the following:
- Graphics processors
- Electricity
- Cooling systems
- Networking hardware
- Physical data center space
The result is a tectonic shift in the industry.
A handful of companies capable of assembling massive AI superclusters are becoming gatekeepers to the future of artificial intelligence.
That is why this deal matters so much.
Anthropic already competes with:
- OpenAI
- Google DeepMind
- Meta AI
- xAI itself
Yet, despite being rivals, Anthropic apparently still needs external compute muscle to keep pace in the escalating AI race.
In 2026, AI companies increasingly resemble countries scrambling for energy supplies during an industrial revolution.
Why the partnership is so surprising
The partnership appears deeply ironic because Musk has spent years publicly criticizing Anthropic.
Among his past comments:
- Calling the company “hypocritical.”
- Saying it was “doomed to become the opposite of its name.”
- Accusing it of hostility toward Western civilization
- Mocking its “Constitutional AI” philosophy
Anthropic positioned itself as one of the AI industry’s safety-focused alternatives, emphasizing structured behavioral guardrails and governance principles.
Musk, meanwhile, often framed xAI as an “anti-woke” counterweight to competitors he viewed as ideologically constrained.
The two companies cultivated opposite public identities.
Now they are business partners.
Silicon Valley occasionally resembles a gladiator arena where competitors throw spears during the day and share generators at night.
Musk’s “evil detector” comment adds fuel online
The deal gained even more attention because of Musk’s sudden rhetorical pivot.
After meeting Anthropic’s team, Musk posted online:
“Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector.”
The comment spread rapidly online because it contrasted sharply with his previous attacks.
Musk added that Claude would “probably be good” if the company continued engaging in “critical self-examination.”
The phrasing sounded half endorsement, half warning label.
What is Colossus 1?
The centerpiece of the agreement is Colossus 1, xAI’s enormous AI supercomputer facility in Memphis.
The system reportedly operates at a scale beyond most traditional cloud deployments and could rival infrastructure controlled by:
Large AI companies increasingly face a difficult problem: public cloud infrastructure is becoming too expensive and too generalized for frontier AI workloads.
Owning or leasing dedicated compute capacity offers the following:
- Better performance efficiency
- Greater scheduling control
- Faster experimentation cycles
- Reduced dependency on traditional cloud providers
That may explain why Anthropic pursued the deal despite the unusual political optics.
Why Wall Street may love this deal
For investors, the agreement potentially changes how xAI is valued.
Most AI startups are judged based on:
- Model performance
- Consumer adoption
- Enterprise revenue
- Research breakthroughs
But infrastructure businesses often command enormous long-term value because they become foundational layers for entire industries.
If xAI evolves into both
- A frontier AI developer
- A major AI infrastructure provider
Then Musk’s company could occupy a uniquely powerful position.
Analysts increasingly view compute capacity as the strategic choke point of the AI economy.
Owning the chips may eventually matter as much as owning the models.
The deeper power struggle inside AI
The deal also highlights a growing contradiction in the AI industry.
Companies publicly compete on philosophy:
- Safety vs speed
- Open vs closed models
- Regulated vs. Unrestricted AI
- “Aligned” vs “maximalist” approaches
But behind the scenes, nearly all frontier AI labs depend on overlapping infrastructure ecosystems.
The rivalry can appear theatrical at times because the same companies often
- Share chip suppliers
- Lease cloud capacity from competitors
- Collaborate on standards
- Recruit from the same talent pool
AI firms increasingly behave like rival kingdoms, drawing water from the same reservoir.
Anthropic’s “dreaming” feature hints at where AI is heading
The partnership announcement coincided with Anthropic unveiling a new feature called “dreaming” during its developer event in San Francisco.
The feature reportedly aims to help AI systems:
- Work between sessions
- Identify patterns over time
- Update persistent user context
- Retain longer-term memory structures
That direction reflects a broader industry trend toward more persistent, personalized AI systems.
The next phase of AI competition may revolve less around raw chatbot responses and more around memory, continuity, and autonomous reasoning.
Those capabilities require massive computing resources, which makes infrastructure deals like this increasingly critical.
Why this deal could reshape the AI industry
The xAI-Anthropic partnership suggests the AI race is entering a new phase where compute alliances matter as much as model releases.
The industry is consolidating around three pillars:
- Chips
- Energy
- Data centers
The companies controlling those resources may ultimately wield more power than the companies with the flashiest chatbot demos.
And perhaps the strangest part of the story is this: one of Elon Musk’s fiercest AI rivals may now depend on Musk’s machines to compete against everyone else.
TL;DR
Elon Musk’s xAI has signed a major compute deal with Anthropic, giving Claude AI access to the Colossus 1 supercomputer in Memphis. The partnership is surprising because Musk previously harshly criticized Anthropic and its AI philosophy. The deal highlights how computing power is becoming the most valuable resource in the AI industry, with infrastructure increasingly shaping the balance of power between competing AI companies.