SpaceXAI and Anthropic Join Forces Against OpenAI

Anthropic and SpaceXAI — Elon Musk's newly rebranded AI division — have announced a landmark partnership that unites two of OpenAI's fiercest competitors. The deal grants Anthropic access to SpaceXAI's massive Colossus 1 data center, immediately adding over 300 megawatts of compute capacity to power Claude Code and the Claude API.

The announcement, made in the early hours of May 7, 2025, also coincided with SpaceXAI's release of Grok 4.3, its most powerful model to date. Together, these developments signal a dramatic realignment in the AI industry's competitive landscape.

Key Takeaways

• Anthropic gains 300+ megawatts of additional compute from SpaceXAI's Colossus 1 data center
• xAI is no more — Musk has folded it into SpaceX as 'SpaceXAI'
• Claude Code limits doubled — the 5-hour billing cap on paid plans is now 10 hours
• Peak-hour restrictions removed for Pro and Max Claude Code accounts
• Grok 4.3 launched with a 1 million token context window at $1.25 per million input tokens
• OpenAI now faces a united front from two of its most well-funded rivals

Musk Dissolves xAI, Creates SpaceXAI Division

Elon Musk took to his social platform X to make a sweeping announcement: 'xAI will no longer exist as an independent company, but will become SpaceX's AI business division, known as SpaceXAI.' The move consolidates Musk's AI ambitions under the SpaceX umbrella, leveraging the aerospace giant's engineering infrastructure and operational scale.

This restructuring is more than cosmetic. By integrating xAI into SpaceX, Musk gains direct access to SpaceX's existing supply chain relationships, power procurement capabilities, and data center infrastructure — including the enormous Colossus 1 facility. The Colossus 1 data center, originally built to train xAI's Grok models, houses a staggering amount of GPU compute that has now become a strategic asset for the broader AI ecosystem.

The rebranding also positions Musk's AI efforts alongside SpaceX's Starlink satellite network, raising the possibility of future edge-AI deployments and distributed inference at a global scale. For now, however, the most immediate consequence is the Anthropic partnership.

Anthropic Unlocks Massive Compute Boost for Claude

Under the new agreement, Anthropic will utilize all available computing resources at the Colossus 1 data center. This translates to more than 300 megawatts of additional compute capacity coming online within May 2025 — a massive injection of processing power by any standard.

The extra capacity is being directed at two primary bottlenecks: Claude Code and the Claude API. Developers who rely on Claude for coding assistance and enterprise API integrations have long complained about rate limits and peak-hour throttling. This deal addresses those pain points head-on.

Specific improvements rolling out immediately include:

• The 5-hour billing time limit on Claude Code paid plans has been doubled to 10 hours
• Peak-hour usage restrictions have been eliminated for Pro and Max account holders using Claude Code
• API rate limits for the Claude Opus model have been significantly relaxed
• Overall throughput and availability across all Claude API tiers are expected to improve

For developers and enterprises building on Claude, these changes are substantial. Rate limits have been one of the most common friction points cited by Anthropic's user base, particularly for teams running complex agentic workflows or large-scale code generation tasks.

Grok 4.3 Arrives as SpaceXAI's Most Powerful Model

Almost simultaneously with the partnership announcement, SpaceXAI unveiled Grok 4.3 — described as the fastest and most intelligent model the company has ever produced. The release positions SpaceXAI as a serious contender in the frontier model race alongside OpenAI's GPT-4.1, Google's Gemini 2.5, and Anthropic's own Claude Opus.

Grok 4.3's headline specifications are impressive:

• 1 million token context window — matching or exceeding competitors like Gemini 2.5 Pro
• Input pricing: $1.25 per million tokens
• Output pricing: $2.50 per million tokens
• Marketed as SpaceXAI's fastest inference model to date

The pricing structure is notably competitive. Compared to OpenAI's GPT-4.1, which charges $2.00 per million input tokens and $8.00 per million output tokens, Grok 4.3 undercuts on both input and output costs. This aggressive pricing strategy suggests SpaceXAI is willing to trade margin for market share, a pattern increasingly common across the industry.

The 1 million token context window is particularly significant for enterprise use cases involving long-document analysis, codebase understanding, and complex multi-turn conversations. It puts Grok 4.3 in direct competition with Google's Gemini models, which have led the industry in context length.

Why Two OpenAI Rivals Are Teaming Up

The partnership between SpaceXAI and Anthropic might seem surprising at first glance. Both companies compete in the large language model space, and both were founded, at least in part, as responses to concerns about OpenAI's direction. Musk co-founded OpenAI before departing and later suing the organization. Anthropic was founded by former OpenAI executives Dario and Daniela Amodei.

Yet the collaboration makes strategic sense for both parties. Anthropic's primary constraint has never been model quality — Claude consistently ranks among the top models on coding and reasoning benchmarks. Its bottleneck has been compute infrastructure. Despite raising over $10 billion from investors including Amazon and Google, Anthropic has struggled to scale its inference capacity fast enough to meet surging demand for Claude Code and its API services.

SpaceXAI, on the other hand, has compute to spare. The Colossus 1 data center was built at enormous scale to support Grok's training runs, but inference workloads don't always require the same sustained capacity. Leasing unused compute to Anthropic generates revenue while keeping the facility running at higher utilization rates.

This is a classic infrastructure-sharing arrangement, similar to how cloud providers lease excess capacity. The difference here is that it's happening between two frontier AI labs — a first in the industry.

What This Means for Developers and Businesses

For the developer community, the immediate impact is clear: Claude becomes more accessible and less throttled. Teams that have been working around rate limits or scheduling workloads to avoid peak hours can now operate more freely. The doubling of Claude Code's billing window is especially welcome for developers engaged in extended coding sessions or complex refactoring tasks.

Enterprise customers using the Claude API for production workloads should see improved latency and availability. The relaxation of Opus-tier rate limits opens up new possibilities for deploying Claude in high-throughput applications like customer service automation, document processing, and real-time code review.

Meanwhile, Grok 4.3's competitive pricing gives businesses another option in the frontier model marketplace. Organizations evaluating multi-model strategies now have a strong cost-performance alternative to consider, particularly for use cases requiring long context windows.

Industry Context: The AI Alliance Era Begins

This partnership marks a new phase in the AI industry's evolution. Until now, frontier AI labs have operated as fiercely independent competitors, each building proprietary infrastructure and guarding their compute resources closely. The SpaceXAI-Anthropic deal suggests that the economics of AI infrastructure are pushing even rivals toward collaboration.

The AI compute market is experiencing a well-documented supply crunch. NVIDIA's latest GPUs remain backordered, and data center construction timelines stretch 18 to 24 months or longer. In this environment, sharing existing infrastructure is often faster and more cost-effective than building new capacity from scratch.

OpenAI, which recently closed a $40 billion funding round and is building its own Stargate data center project with SoftBank and Oracle, now faces a united competitive front. While OpenAI remains the market leader in terms of brand recognition and user base, the SpaceXAI-Anthropic axis creates a formidable alternative ecosystem.

Google, with its deep cloud infrastructure and Gemini model family, remains the other major pole in this increasingly multipolar AI landscape.

Looking Ahead: What Comes Next

Several questions remain unanswered. The financial terms of the SpaceXAI-Anthropic deal have not been disclosed, and it is unclear whether this is a short-term compute lease or the beginning of a deeper strategic relationship. If the partnership proves successful, it could expand to include joint research initiatives, shared safety testing frameworks, or even model interoperability features.

The Colossus 1 arrangement is expected to be fully operational within May 2025, giving Anthropic an almost immediate capacity boost. Developers should begin seeing improved rate limits and reduced throttling in the coming days.

For the broader AI industry, this partnership sets a precedent. If two of the most prominent OpenAI competitors can find common ground on infrastructure sharing, similar deals may follow across the sector. The age of isolated AI empires may be giving way to an era of strategic alliances — and the biggest target is squarely on OpenAI's back.