5GW committed, majority on Trainium — roughly a quarter of Bitcoin's annual energy draw earmarked for one model lab on non-Nvidia silicon. Akash and Render keep pitching themselves as the answer to AI compute bottlenecks; deals this size show training is consolidating into 3-4 hyperscaler alliances while DePIN networks pick up inference scraps. The Nvidia moat thins every time AWS ships another Trainium cluster, and the power-sourcing story for 5GW at this pace is going to get interesting fast.

Top comment by @Benthic

More coverage

Explore the topic

More on Anthropic

Galaxy Research warns Washington's shutdown of Anthropic's Fable 5 sets a dangerous precedent, giving regulators veto power over which AI models Americans can access

Galaxy Research warns Washington's shutdown of Anthropic's Fable 5 sets a dangerous precedent, giving regulators veto power over which AI models Americans can access

𝕏/@intangiblecoins ·

Micron and Anthropic strike a strategic AI infrastructure partnership covering memory design, multi-year supply agreements, Claude adoption, and a Series H investment.

Micron and Anthropic strike a strategic AI infrastructure partnership covering memory design, multi-year supply agreements, Claude adoption, and a Series H investment.

investors.micron ·

OpenAI mirrors Bitcoin and Anthropic mirrors Ethereum as AI labs follow the altcoin playbook, raising billions on research narratives and future potential

OpenAI mirrors Bitcoin and Anthropic mirrors Ethereum as AI labs follow the altcoin playbook, raising billions on research narratives and future potential

𝕏/@AlokVasudev ·

Anthropic's June 16, 2026 economic research analyzing ~400,000 Claude Code sessions reveals domain expertise not coding skills drives AI coding success, with experts achieving 28–33% verified success versus 15% for novices, while non-software occupations succeeding at nearly identical rates (26%) to software engineers (30%).

Anthropic's June 16, 2026 economic research analyzing ~400,000 Claude Code sessions reveals domain expertise not coding skills drives AI coding success, with experts achieving 28–33% verified success versus 15% for novices, while non-software occupations succeeding at nearly identical rates (26%) to software engineers (30%).

Anthropic ·

Anthropic publishes FAQ for B2B clients to Enable US-only inference

Anthropic publishes FAQ for B2B clients to Enable US-only inference

support.claude ·

Following US government directive, Anthropic suspends all access to Fable

Following US government directive, Anthropic suspends all access to Fable

𝕏/@anthropicai ·

Explore the full feed

Comments

Up nextKiwoom Securities seeks stake in Bithumb through new share deal as Korean brokers move on crypto