303 stories from 25 sources
The month in AI and Technology
Longest running story: 28 days
Click on a headline to see stories that shaped it.
Threads
6 threads
The month moved from major financing and long-term capacity deals to a broader contest for GPUs, data centers, power, memory, and serving capacity. China also challenged the computing lead as export controls raised chip costs and local providers filled gaps, while companies pursued custom silicon and passed infrastructure costs into hardware prices.
Anthropic's enterprise expansion gave way to tool suspensions, export-style restrictions, and negotiations over trusted access to its most capable models. Government control over model releases then widened, while cyber warnings and alleged capability extraction made both misuse and model theft central risks.
OpenAI paired exceptional user growth and IPO preparations with a push toward sustainable economics. Rising costs, tighter access, cheaper Chinese models, and weak return-on-investment evidence forced providers and buyers to question whether widespread deployment creates enough value.
OpenAI, Microsoft, and established enterprise platforms shifted from selling models toward agents that execute tasks across applications and business processes. Salesforce's acquisitions and finance tools, alongside broader commerce and product-design efforts, showed companies competing to own the workflows where automation acts.
Nvidia extended its strategy from data centers into laptops and local workloads. Apple simultaneously positioned Apple Intelligence as a system-wide layer across its devices and apps, making distribution and on-device integration central parts of the platform competition.
Legal and regulatory pressure widened around training data, personalization, deepfake safeguards, advertising, and sensitive information. At the same time, enterprise adoption remained broad but shallow, leaving companies with unresolved questions about oversight and measurable returns.
Week’s Recap