SpaceX Leases Data Center to Anthropic — Repercussions of Technical Hurdles Faced in Grok Development

According to Bloomberg, SpaceX has leased the full capacity of its Colossus 1 data center in Memphis to Anthropic. The move comes after SpaceX ran into latency issues between Colossus 1 and two other data centers located more than 10 miles away while developing and operating its own Grok AI models. Aging network infrastructure reportedly compounded the problems.

SpaceX had aimed to train cutting-edge AI models by linking its three data center campuses together, but the physical distance between the facilities ended up being the bottleneck. As a result, Anthropic will now put Colossus 1's compute resources to use — an unexpected partnership in the AI infrastructure market.

Claude Fable 5 Tops the Benchmarks, but at 2x the Cost for a 5.7% Performance Gain

According to The Decoder's analysis, Anthropic's latest model Claude Fable 5 scored 64.9 points on the Artificial Analysis Intelligence Index, setting new records on 5 of 10 benchmarks. Compared with the previous-generation Opus 4.8, however, the performance gain is just 5.7% — while per-token pricing has doubled.

On top of that, fallback routing triggered by safety filters is structured to push actual costs even higher, which is likely to reignite the debate over the balance between peak performance and cost efficiency. As gains in model performance slow, whether the pricing is justified is becoming a key decision factor for users.

Anthropic Partners with TCS — Claude Rolling Out to 50,000 Employees Across 56 Countries

Anthropic has entered into a partnership with Tata Consultancy Services (TCS). TCS, one of the world's largest technology services companies, will provide Claude to its 50,000 employees and deploy it across 56 countries.

Specifically, the plan is to develop Claude-powered products for regulated industries such as financial services, healthcare, and the public sector. Several initiatives are already underway: Diligenta, TCS's UK life insurance subsidiary, will use Claude for its more than 22 million policyholders, while banking and financial products teams aim to boost software engineering productivity with Claude Code.

Anthropic CEO Dario Amodei said the company has "built Claude to be safe and reliable, especially where accuracy matters," and TCS will join the Claude Partner Network.

Rumor: Mistral Raising €3 Billion at a €20 Billion Valuation

According to TechCrunch, rumors have surfaced that French AI startup Mistral is raising €3 billion (roughly ¥2.3 trillion) at a €20 billion valuation — nearly double its €11.7 billion valuation from its previous Series C round.

If confirmed, it would be one of the largest funding rounds for a European-born AI company, further raising Mistral's profile as a challenger to OpenAI and Anthropic. For now, though, this remains at the rumor stage, and an official announcement is awaited.

OpenAI Loosens Codex Rate Limits — The Opening of an AI Price War

OpenAI has made it possible for users of the Codex coding agent to trigger rate limit resets at any time of their choosing. Previously, resets expired on a fixed schedule; going forward, banked resets can be drawn on whenever they're needed.

The Decoder frames this as "the beginning of the AI price war." Users on the Go, Plus, Pro, and Business plans each get one free reset, and Plus and Pro users can unlock additional resets by inviting friends. While it improves the developer experience, the move also looks like a countermeasure against competitors.

Apple WWDC 2026: Running Agentic AI Locally with MLX

At WWDC 2026, Apple published a session on running agentic AI locally on the Mac using the MLX framework. It has been a hot topic on Hacker News, with the direction of running agentic AI locally without relying on the cloud drawing attention as a very Apple-like, privacy-first approach.

Details on specific benchmarks and supported models remain to be seen, but this looks like a significant step toward edge AI that taps the power of Apple Silicon.