The enormous energy requirements of modern AI models are driving tech giants into space. However, the hurdles are immense - from cooling to radiation. But companies are thinking in terms of decades, not years.

The article AI in space requires new cooling tech and cheap rockets appeared first on THE DECODER.

The US Department of Defense has officially launched "GenAI.mil," a centralized platform designed to give roughly three million civilian and military employees and contractors direct access to generative AI. To kick things off, the system is rolling out Google Cloud's "Gemini for Government."

The article Pentagon debuts GenAI.mil platform with Google while eyeing rival models appeared first on THE DECODER.

A new report warns that a massive energy gap in the US could threaten the expansion plans of OpenAI, Microsoft, and their peers.

The article Report: Aging power grid puts OpenAI and Microsoft's growth at risk appeared first on THE DECODER.

A year after publishing its Titans paper, Google has formally detailed the architecture on its research blog, pairing it with a new framework called MIRAS. Both projects target a major frontier in AI: models that keep learning during use and maintain a functional long-term memory instead of remaining static after pretraining.

The article Google outlines MIRAS and Titans, a possible path toward continuously learning AI appeared first on THE DECODER.

Anthropic has acquired Bun, the popular JavaScript and TypeScript runtime and toolchain. The company says the deal strengthens the infrastructure behind both its coding tool Claude Code and the Claude Agent SDK. Bun already powers these systems today, and Claude Code ships to millions of users as a Bun executable.

The article Anthropic brings Bun in-house, the runtime powering Claude Code’s $1B ARR appeared first on THE DECODER.

Nvidia used the NeurIPS conference to debut new AI models for autonomous driving and speech processing.

The article Nvidia shows new AI models for autonomous driving and speech processing appeared first on THE DECODER.

HSBC has signed a multiyear partnership with the French startup Mistral AI to bring generative AI across its global operations.

The article HSBC relies on Mistral AI for global AI integration appeared first on THE DECODER.

A large analysis of more than 170,000 reasoning traces from open-source reasoning models shows that large language models rely heavily on simple, default strategies when tasks get harder. A new cognitive science framework for categorizing thinking processes makes it easier to see which abilities are missing and when extra reasoning guidance in the prompt actually helps.

The article New study maps how AI models think and where their reasoning breaks down appeared first on THE DECODER.

With Gemini 3 Pro, Google aims to reclaim AI leadership. But Deepmind-CEO Demis Hassabis is already looking ahead to the next, expensive stage of AI development, warning of a bubble in the private market and explaining why Google's long-term strategy is paying off.

The article Deepmind CEO Hassabis: World models are the future, but the AI bubble is real appeared first on THE DECODER.

Microsoft has introduced Agent 365, a new platform designed to help organizations manage their AI agents as if they were part of the workforce. 

The article Agent 365: Microsoft launches management platform for AI agents appeared first on THE DECODER.

Gemini 3 Pro debuts with major gains in logic, multimodality, and speed, along with a new agent platform that hints at where Google wants AI to go next.

The article Analysts say Google now leads the AI performance race with Gemini 3 Pro appeared first on THE DECODER.

Arm says its CPUs will be able to connect directly to Nvidia's AI chips using NVLink Fusion, making it easier for customers to pair Neoverse CPUs with Nvidia GPUs.

The article Arm and Nvidia enter into partnership for AI chips with NVLink Fusion appeared first on THE DECODER.