Elon Musk and Anthropic: Inside the $40 Billion SpaceX Compute Deal, the Public U-Turn, and What It Means for the AI Industry

In July 2026, one of the strangest business relationships in tech got even stranger. Elon Musk — who runs xAI, a direct rival to Anthropic’s Claude — publicly admitted he was wrong about Anthropic’s chances of winning the AI race, praised its Claude Mythos and Fable models as the best on the market, and promised he would never use his control of SpaceX’s data centers to hurt the company that is paying him roughly $1.25 billion every month.

That promise matters because Anthropic isn’t just a Musk critic turned admirer — it’s now one of the largest customers of SpaceX’s AI infrastructure business, running mission-critical workloads on hardware owned by a competitor. This single fact sits at the center of one of 2026’s biggest AI industry stories: what happens when the company building your AI models also builds the rockets — and the servers — you depend on?

This article breaks down exactly what Musk said, why it matters, the full financial terms of the SpaceX-Anthropic compute agreement, the legitimate risks analysts are flagging (including AI “distillation” and competitive visibility), and what this episode reveals about the broader AI infrastructure economy in 2026. We’ll also cover critical context competitors’ coverage missed — including the wider Anthropic funding and IPO story, the recent U.S. export-control suspension of Anthropic’s flagship models, and how this deal compares to other mega infrastructure contracts reshaping the industry.

Quick Answer

Did Elon Musk say he would protect Anthropic’s access to SpaceX infrastructure? Yes. On July 9, 2026, Musk posted on X that he was “clearly wrong” about doubting Anthropic and stated he would never cut off the company’s compute access “in a way that hurt them badly, even as a competitor.” The comment came after Anthropic signed a deal worth roughly $1.25 billion per month to lease compute capacity from SpaceX’s Colossus 1 data center, an asset that came under SpaceX’s control after xAI merged with SpaceX in February 2026.

Timeline: From “Winning Was Never Possible” to “I Was Clearly Wrong”

Understanding this story requires understanding how fast Musk’s public position on Anthropic has shifted. Here’s the sequence of events:

This reversal is notable not just because Musk rarely walks back criticism of a competitor, but because it happened while Anthropic and xAI are actively competing head-to-head for the same enterprise customers with Claude Mythos/Fable and Grok 4.5.

Also Read : AI Productivity Tools 2026: The Complete Guide to This Year’s Biggest Updates

What Exactly Did Elon Musk Say?

Responding to a wave of X posts questioning whether Anthropic had made a strategic error by depending on infrastructure controlled by a rival, Musk wrote that he was “clearly wrong about Anthropic,” adding that the company is “obviously currently the leader in AI” and that no rival has released a model as capable as Mythos/Fable. He also predicted Anthropic would release Mythos 2 soon, and stated plainly that he would never cut the company off from SpaceX infrastructure “in a way that hurt them badly, even as a competitor,” calling that approach “not my style.”

To back up the claim, Musk pointed to his own companies’ track record:

• Tesla’s 2014 patent pledge — Tesla agreed not to pursue patent lawsuits against anyone using its technology in good faith.
• Tesla Supercharger network opened to rival automakers — Other EV brands can now charge at Tesla stations.
• SpaceX’s satellite-launch neutrality — Musk claims SpaceX launches satellites for competing companies without inflating prices or adding unfair terms.
• X’s tolerance of criticism — Musk noted that even his fiercest critics remain free to use the platform.

Whether these examples fully answer the concerns raised about Anthropic’s exposure is something analysts and investors are still debating — a point the original reporting on this story largely glossed over.

Inside the SpaceX-Anthropic Compute Deal: The Numbers

This isn’t a minor cloud-hosting arrangement. It’s one of the largest AI infrastructure contracts signed in 2026, and the financial scale explains why Musk’s public reassurance carries real weight for markets and enterprise customers alike.

Deal Terms at a Glance

How Anthropic’s Deal Compares to Google’s SpaceX Contract

Anthropic isn’t the only AI heavyweight renting SpaceX-controlled compute. Google has also signed an infrastructure deal with SpaceX, giving the arrangement industry-wide significance rather than being an Anthropic-specific anomaly.

Together, these two deals alone could generate tens of billions of dollars in recurring revenue for SpaceX’s xAI infrastructure business — a meaningful diversification beyond launch services and Starlink.

Why This Deal Is Controversial: The Core Risks

On the surface, this looks like a straightforward compute-capacity transaction. But because SpaceX’s AI infrastructure business is fused with xAI — Anthropic’s direct competitor in the frontier AI market — the arrangement raises questions that don’t come up in a typical cloud contract.

1. Concentration Risk and Vendor Dependence

Anthropic is leasing the entire output of a single data center from a company that builds a competing foundation model (Grok). If that relationship deteriorated, Anthropic’s model training, inference capacity, and customer-facing services could theoretically be affected. Some commentators on X went further, suggesting the compute lease was reportedly a shorter-term arrangement without a guaranteed renewal, which would leave Anthropic more exposed than the headline “through 2029” figure suggests.

2. AI “Distillation” Concerns

Distillation is the practice of extracting the behavior of a rival’s AI model — often by generating massive volumes of automated queries — to improve a competing system. Musk has acknowledged in legal proceedings (his lawsuit against OpenAI) that this practice happens across the AI industry. Anthropic itself accused several Chinese AI developers of running large-scale distillation campaigns against Claude in early 2026. Critics note that hosting Anthropic’s compute could, in theory, give SpaceX’s infrastructure teams more visibility into Anthropic’s usage patterns than an arm’s-length competitor would ever have.

3. Operational Visibility

Beyond distillation, infrastructure providers inherently see signals like training schedules, compute demand curves, and scaling patterns. When the provider also owns a rival model, that visibility becomes commercially sensitive — even with standard data-isolation and security safeguards in place.

4. Musk’s Track Record With Rivals Is Mixed

While Musk cited Tesla’s patent pledge and Supercharger access as evidence of good faith, he has also pursued aggressive legal action against competitors — most notably his lawsuit against OpenAI and Sam Altman. Analysts covering this story have pointed out that Musk’s history includes both cooperative gestures and adversarial tactics, meaning his personal assurance alone isn’t a complete risk mitigant.

Why Anthropic Likely Isn’t Relying on Trust Alone

Large infrastructure agreements at this scale typically include:

• Contractual service-level guarantees with financial penalties for early termination or degraded service
• Legal remedies if either party breaches agreed terms
• A 90-day termination notice window, giving Anthropic runway to secure alternative capacity if the relationship sours
• Strong financial incentives for SpaceX to preserve the deal — losing a $1.25B/month customer would be a major revenue and reputational hit

Why SpaceX Wants This Deal Just as Much as Anthropic Does

It’s easy to frame this story as “Anthropic takes a risk,” but the financial incentives clearly run both directions.

• Recurring revenue at scale. A combined ~$2.17 billion per month from Anthropic and Google alone represents a transformative new revenue stream for SpaceX’s AI infrastructure arm, supplementing launch and Starlink revenue.
• Technical experience. SpaceX’s engineering teams gain hands-on experience supporting one of the fastest-growing AI companies in history — comparable to the operational knowledge Amazon has built through its own Anthropic partnership via AWS.
• Market credibility. Hosting Anthropic and Google signals that SpaceX’s data center infrastructure is enterprise-grade and can attract blue-chip AI customers beyond xAI’s internal Grok workloads.
• Reputational stakes. If SpaceX were seen as weaponizing infrastructure against a paying customer, it would deter other AI labs from ever signing similar deals — a reputational risk far larger than any single contract.

The Bigger Picture: Anthropic’s Extraordinary 2026 Growth Story

Coverage focused narrowly on the Musk tweet misses just how significant this moment is in the context of Anthropic’s broader trajectory — arguably the fastest revenue ramp in enterprise software history.

Anthropic’s Revenue Run-Rate Timeline

For context, Salesforce took roughly two decades to reach $30 billion in annual revenue. Anthropic reportedly crossed that mark in under three years from a standing start.

Funding and Valuation Milestones

• March 2025: $61.5 billion valuation
• September 2025: $183 billion (Series F)
• February 2026: $380 billion (Series G, $30B raised)
• May 2026: $965 billion (Series H, $65B raised) — with a confidential IPO filing submitted June 1, 2026

This valuation surge is directly relevant to the SpaceX deal: Anthropic’s compute needs are scaling just as fast as its revenue, and infrastructure scarcity — not customer demand — has become one of its biggest constraints. Anthropic has publicly acknowledged strain on infrastructure affecting reliability during peak usage, which helps explain why locking in an entire dedicated data center, even one owned by a rival, was worth the risk.

The Export Control Twist Competitors’ Coverage Missed

One important wrinkle that’s largely absent from other coverage of this story: Anthropic’s Mythos and Fable models — the very models Musk praised — were briefly pulled from public access in mid-June 2026 after the U.S. Department of Commerce imposed export controls affecting the models. Anthropic suspended access on June 12, 2026, and only restored it on July 1, 2026, after the Commerce Department lifted the restriction on June 30. Musk’s July 9 praise for Mythos/Fable came just over a week after the models became publicly available again — timing that adds useful context to why the models were fresh in the industry conversation.

Enterprise AI Infrastructure Market: Why Deals Like This Are Becoming the Norm

The Anthropic-SpaceX arrangement isn’t an isolated curiosity — it reflects a structural shift in how frontier AI labs secure compute in 2026.

• Anthropic has also committed more than $100 billion over 10 years to AWS infrastructure in exchange for up to 5 gigawatts of compute capacity.
• The company has separately expanded compute partnerships with Google Cloud and Broadcom for gigawatt-scale TPU capacity beginning in 2027.
• Industry-wide, GPU and data center scarcity has pushed AI labs into unconventional partnerships — including deals with companies that are simultaneously competitors, as seen with Anthropic-xAI/SpaceX and other cross-industry compute agreements reported in mid-2026.
• Analysts have summarized the strategic logic bluntly: shared rivalries can create alignment even between competitors — in this case, xAI and Anthropic have a shared incentive to keep pace with OpenAI, which may partly explain the willingness to do business despite competing products.

Key takeaway for enterprise decision-makers: if even the largest, best-funded AI labs are being forced into infrastructure-sharing arrangements with competitors, expect compute scarcity — not raw model capability — to remain the primary bottleneck shaping AI product roadmaps through at least 2027.

Skeptical Voices: Not Everyone Is Convinced

Balanced coverage requires acknowledging the pushback. Some AI industry commentators on X have questioned whether Anthropic’s leadership underestimated the risk of handing a direct competitor this much financial and infrastructural leverage — with one widely shared post asking whether history will eventually view the deal as one of the AI era’s biggest unforced errors. The core worry: if xAI’s Grok models continue improving while SpaceX collects billions in guaranteed infrastructure revenue from Anthropic, Musk’s companies benefit financially from Anthropic’s success regardless of which AI lab ultimately wins more market share.

Others take the opposite view, arguing that Anthropic’s enterprise dominance — including its position as the leading provider of enterprise LLM spend — gives it enough negotiating leverage and legal protection that the risk is manageable, especially given the reputational cost SpaceX would bear by acting in bad faith.

What This Means for Businesses Evaluating AI Vendors

If your organization relies on Claude, Gemini, GPT, or Grok for production workloads, this story has practical implications:

• Infrastructure dependencies are now part of AI vendor due diligence. Ask any AI provider where their compute actually comes from — not just which model you’re using.
• Contractual termination clauses matter more than public statements. A CEO’s promise on social media isn’t a substitute for reviewing SLAs, redundancy plans, and multi-cloud strategies.
• Compute scarcity, not model quality, may be the deciding factor in which AI vendors can reliably scale with your business over the next 24 months.
• Cross-competitor infrastructure deals are becoming normal, not exceptional — expect to see more of them as GPU supply remains tight industry-wide.

Frequently Asked Questions

Q1: Why did Elon Musk change his opinion about Anthropic? Musk publicly admitted he was wrong about Anthropic’s competitive prospects after previously claiming in September 2025 that “winning was never” possible for the company. The reversal coincided with Anthropic’s rise to enterprise market leadership and a major compute deal between Anthropic and SpaceX.

Q2: How much is Anthropic paying SpaceX for compute? Anthropic reportedly pays approximately $1.25 billion per month for access to SpaceX’s Colossus 1 data center, under a contract running through May 2029 and estimated to be worth around $40 billion in total revenue for SpaceX’s xAI unit.

Q3: Can Elon Musk cut off Anthropic’s access to SpaceX infrastructure? Musk has publicly stated he would not do this, calling it “not my style.” However, industry reports suggest the agreement includes a 90-day termination clause available to either party, meaning the relationship isn’t purely dependent on Musk’s personal word.

Q4: What is AI distillation, and why does it matter here? AI distillation refers to the practice of using extensive automated interactions with a competitor’s AI model to learn how it behaves and improve a rival system. Because SpaceX hosts Anthropic’s compute, some analysts worry it could gain unusual visibility into Anthropic’s operations — a concern Musk has acknowledged is a real industry-wide practice.

Q5: Is Google also using SpaceX’s AI infrastructure? Yes. Google has reportedly signed a separate deal to use SpaceX infrastructure, paying approximately $920 million per month through June 2029.

Q6: What are Claude Mythos and Fable? Mythos and Fable are Anthropic’s most advanced AI model tier, sitting above the Claude Opus line. They were briefly suspended from public access in mid-June 2026 due to U.S. export controls before being restored on July 1, 2026, after the restriction was lifted.

Q7: How big is Anthropic’s business now? As of mid-2026, Anthropic’s annualized revenue run-rate has been reported at approximately $47 billion, with a valuation near $965 billion following its Series H funding round and a confidential IPO filing submitted in June 2026.

Conclusion: Key Takeaways

Elon Musk’s public reversal on Anthropic is more than a social media moment — it’s a window into how deeply intertwined competitor relationships have become in the AI infrastructure race of 2026.

Actionable takeaways:

1. Compute access, not just model capability, now determines competitive positioning in frontier AI — even for a company as well-funded as Anthropic.
2. Cross-competitor infrastructure deals are a growing industry pattern, not a one-off anomaly, driven by GPU and data center scarcity.
3. Public reassurances from executives are not a substitute for contractual protections — always evaluate the legal and financial terms behind any partnership, not just the headlines.
4. Watch the termination clauses. A 90-day notice window means this relationship, despite running “through 2029” on paper, could shift faster than the headline contract length suggests.
5. For enterprises choosing an AI vendor, infrastructure resilience and multi-cloud redundancy deserve the same scrutiny as model benchmarks and pricing.

The Musk-Anthropic relationship will remain a useful case study in how far commercial pragmatism can stretch even the fiercest AI rivalries — and whether that pragmatism holds as both companies race to out-build each other over the next three years.