This week is not really a feature week.

It is a capacity week.

Anthropic's compute expansion, Project Glasswing, and Microsoft's nation-scale AI commitments in Japan and Singapore all point to the same shift: compute, defense, and skills are starting to move together.

Anthropic expanded its Google Cloud and Broadcom partnership to secure multiple gigawatts of additional TPU capacity starting in 2027.
Anthropic launched Project Glasswing, a defensive cybersecurity initiative with more than 40 organizations, up to $100 million in credits over five years, and access to non-public defensive model support.
Microsoft paired nation-scale AI infrastructure and workforce commitments in Japan and Singapore, including infrastructure investment, cybersecurity collaboration, and large-scale skills programs.

If the serious players are organizing compute, security, and skills together, AI is no longer just another software category.

That means leaders need to think about provider concentration, defensive coordination, and workforce readiness as one operating board instead of three separate conversations.

Name your top two AI-provider dependencies.
Identify one workflow where provider concentration is now a real operating risk.
Confirm who owns external AI-safety or cyber-intake if a report lands.
Check where AI training is actually happening, not just where it is theoretically available.
Write down one future-access assumption your team is making without evidence.

Anthropic: Google/Broadcom partnership on compute (link)
Reuters: Anthropic expands Google and Broadcom partnership on AI compute (link)
Anthropic: Project Glasswing (link)
The Verge: Anthropic launches Project Glasswing (link)
Microsoft: Japan AI infrastructure, cybersecurity, and workforce commitment (link)
Microsoft: Singapore AI access and skills commitment (link)
Reuters: Microsoft to invest $10 billion in Japan for AI infrastructure and skills (link)
The Straits Times: Microsoft to invest more in Singapore and expand AI skilling (link)

This is operational analysis, not legal advice.

AI-assisted tools may be used in parts of the research and production workflow. Final editorial judgment, risk posture, and release approval stay human-led.

AI Change Desk | EP016: National Capacity Check