8am☕Coffee

Kai-Fu Lee warns that the United States is already losing the AI hardware race to China, framing the contest around compute supply, chip manufacturing, and system-level scale. His assessment underscores intensifying competition over semiconductors and infrastructure critical to training and deploying advanced models.

DeepSeek releases an open-source model that compresses text roughly 10x by representing it through images, challenging standard assumptions about text encoding and storage. The approach targets bandwidth and cost reductions while rethinking how language data is packaged for transmission and retrieval.

Qwen’s Deep Research update adds rapid export that turns its generated reports into live webpages and podcasts within seconds. The capability streamlines publishing and distribution by automating formatting and output, reducing manual steps from content creation to delivery.

Microsoft CEO Satya Nadella outlines themes in a shareholder letter with practical implications for enterprises, from platform strategy to operational priorities. The message offers guidance on sustaining growth, navigating technology transitions, and aligning corporate roadmaps with long-term vision.

Anthropic launches a web app for its Claude AI coding assistant, enabling browser-based access beyond IDE-only workflows. The move broadens availability for teams that prefer a lightweight, hosted interface for AI-assisted programming.