Anthropic’s SpaceX Deal, Murati Testifies Against Altman & Anthropic’s $200B Cloud Deal

Anthropic & SpaceX Partner in New Compute Deal 🚀

The geopolitical lines of the artificial intelligence industry are being redrawn. Anthropic has signed a shock agreement with Elon Musk’s SpaceXAI to lease the massive Colossus 1 supercluster, marking a sudden alliance between the two major OpenAI rivals.

Here is the breakdown of the massive infrastructure partnership:

• The Hardware: Anthropic is leasing the entirety of Colossus 1, a massive 300+ megawatt supercomputer in Memphis equipped with over 220,000 Nvidia GPUs that SpaceX built in record time.
• The Capability Boost: Thanks to the immediate influx of processing power, Anthropic announced it is instantly doubling the 5-hour usage caps for Claude Code across all paid tiers and entirely removing peak-hour restrictions.
• The Detente: The deal marks a sharp reversal for Elon Musk, who just months prior had publicly insulted Anthropic as “Misanthropic.” Musk stated on X that he decided to lease the compute because Anthropic is taking the right steps to ensure AI is “good for humanity.”
• The Broader Ecosystem: This lease comes on the heels of reports that Anthropic is simultaneously committing over $200 billion to Google Cloud and $100 billion to AWS, proving its insatiable demand for frontier infrastructure.

Why it matters: In the race for Artificial General Intelligence, the enemy of your enemy is your friend. Elon Musk wants to defeat OpenAI, but xAI’s software models aren’t quite there yet. Anthropic has the software models to beat OpenAI, but desperately needs raw physical compute. By leasing Colossus 1 to Anthropic, Musk is effectively weaponizing his massive hardware advantage to directly attack OpenAI’s market share, turning SpaceX into a kingmaker in the AI infrastructure wars.

UrviumAI Suggestion: Compute capacity is forcing unexpected corporate alliances. If you are an enterprise developer relying heavily on AI coding assistants, immediately migrate your heavy workloads to Claude Code. Because Anthropic just secured the raw power of 220,000 Nvidia GPUs from SpaceX, they are eliminating the peak-hour rate limits that constantly throttle developer speed on other platforms. In the short term, Anthropic now possesses the most stable and generous inference infrastructure on the market.

Mira Murati Speaks Out in Musk vs. OpenAI Trial ⚖️

The courtroom battle over the soul of OpenAI has brought its former top executives to the witness stand. Ex-OpenAI Chief Technology Officer Mira Murati testified via video deposition in Elon Musk’s ongoing lawsuit, leveling severe accusations against CEO Sam Altman.

Here are the key takeaways from Murati’s explosive testimony:

• Allegations of Dishonesty: Murati testified that Altman explicitly told her OpenAI’s legal team had cleared a specific model to bypass standard safety reviews. She later verified with Chief Counsel Jason Kwon that this statement was entirely false.
• Leadership Chaos: The former CTO described Altman’s management style as actively destructive, accusing him of giving conflicting directions to different executives to pit leaders against one another.
• The 2023 Crisis: Reflecting on her brief stint as interim CEO during Altman’s firing in 2023, Murati stated that the chaotic board process put the entire company at “catastrophic risk of falling apart.”
• Counter-Testimony: Former OpenAI board member Helen Toner also testified during the proceedings, reportedly criticizing Murati for being “afraid to stick her neck out” during the leadership crisis out of fear of career blowback.

Why it matters: Elon Musk is attempting to prove that Sam Altman and Greg Brockman deceitfully manipulated OpenAI’s non-profit mission to build a massive commercial empire for personal gain. Having the company’s former CTO swear under oath that Altman actively lied about safety reviews provides massive, credible ammunition for Musk’s legal narrative. If the jury believes OpenAI’s leadership routinely bypasses safety protocols using deceptive tactics, it could trigger devastating regulatory crackdowns and permanently damage the company’s enterprise trust.

UrviumAI Suggestion: Enterprise trust requires transparent governance. The internal chaos at OpenAI should serve as a massive warning for enterprise software buyers. When an AI vendor’s former CTO testifies under oath that the CEO lies about passing safety reviews, you cannot rely entirely on that vendor’s internal safety guarantees. If you are integrating frontier models into highly sensitive corporate environments, you must implement your own strict, independent red-teaming and compliance audits before letting the AI touch your proprietary data.

Anthropic’s $200B Cloud Deal Backlog ☁️

The capital expenditure requirements for frontier artificial intelligence are reaching incomprehensible, macroeconomic scales. Anthropic has reportedly finalized an agreement to spend a staggering $200 billion with Google Cloud over the next five years.

Here is the breakdown of the historic cloud computing commitment:

• The Deal: According to The Information, Anthropic will spend $200 billion to secure massive amounts of Google’s cloud infrastructure and custom tensor processing units (TPUs) to train and run its Claude models.
• The Revenue Backlog: This single commitment is so massive that it accounts for more than 40% of the entire future revenue backlog recently disclosed by Alphabet (Google’s parent company) to its investors.
• The Multi-Cloud Strategy: Anthropic is heavily diversifying its hardware dependence. This Google contract runs in parallel with a massive multi-billion-dollar commitment to Amazon Web Services and a separate deal with CoreWeave.
• The Industry Impact: The combined contractual commitments from just two companies, Anthropic and OpenAI, now account for more than half of the $2 trillion in combined revenue backlogs across Amazon, Microsoft, and Google.

Why it matters: AI is single-handedly propping up the balance sheets of the world’s largest tech companies. By committing to spend $200 billion on servers they haven’t even used yet, Anthropic is proving that the demand for enterprise AI is limitless, but the physical hardware to run it is critically scarce. However, this also creates a massive, precarious financial feedback loop: Silicon Valley’s giants are investing billions into AI startups, and those startups are immediately handing that money right back to the giants to rent server space.

UrviumAI Suggestion: Cloud computing is a zero-sum hardware game. If you are a mid-sized enterprise trying to secure dedicated cloud computing power for your own internal operations, you are about to be priced out of the market. Anthropic and OpenAI are buying up the entire global supply of cloud infrastructure years in advance. You must optimize your software immediately to run on smaller, highly efficient open-source models (like Llama or Kimi) that require vastly less hardware, because the raw compute required for frontier models is being entirely monopolized by the tech giants.

Last AI News: OpenAI’s AI Phone, Span’s Home Data Centers & Anthropic Finance Agents

Other AI News Today:

1. Chinese AI startup DeepSeek is nearing a massive funding round led by China’s state-backed “Big Fund,” valuing the company at up to $45 billion.
2. Anthropic introduces “dreaming” for Managed Agents, allowing AI to review past sessions and self-improve, alongside new outcome grading and multi-agent delegation tools.
3. OpenAI partners with AMD, Intel, Nvidia, and Microsoft to open-source Multipath Reliable Connection (MRC), a networking protocol that prevents AI training crashes.
4. Google partners with Believe and TuneCore to put its new Lyria 3 Pro music model and Flow Music creative assistant in the hands of global artists.
5. Google DeepMind takes a minority stake in Fenris Creations to use the 23-year-old space game EVE Online as a sandbox for long-horizon AI research.

Jigar Chaudhary

Jigar Chaudhary is the Editor-in-Chief at UrviumAI, where he oversees coverage of artificial intelligence news, tools, and in-depth studies. With over 5 years of experience analyzing AI and robotics, he focuses on maintaining high editorial standards, accurate reporting, and clear explanations to help readers understand how AI is shaping the future.