AI demand puts pressure on chips, data centers and copyright law

Google is reportedly developing a server chip called Frozen v2 that embeds parts of Gemini directly into hardware to serve responses with less computing power. The design could deliver six to ten times more tokens per unit of electricity than Google’s latest custom AI hardware, while complementing rather than replacing its TPUs.

Moonshot AI temporarily stopped accepting new Kimi subscriptions after demand for its K3 model exceeded available computing capacity. K3 carries 2.8 trillion parameters, and the pause came as the Chinese startup seeks fresh capital and explores a Hong Kong listing.

A federal judge approved Anthropic’s $1.5 billion settlement with authors who alleged the company used pirated books while developing Claude. More than 91% of eligible authors and publishers have claimed a share, with the court having previously distinguished potential fair use in training from the storage of pirated titles.

Hut 8 signed a second 15-year AI data-centre lease worth $9.8 billion covering 352 megawatts at its Texas campus, while Tempus agreed to acquire Personalis in a deal valued at roughly $1.5 billion. ASML’s valuation has moved close to $700 billion as demand for advanced chipmaking equipment rises, though future gains depend on sustained cloud data-center spending.