AI tooling shifts and cloud platform outages #26

Today's Letter

1. Microsoft, internal Claude Code licenses pulled back
2. Railway says Google Cloud account suspension caused 8-hour outage

Microsoft, internal Claude Code licenses pulled back

• According to Windows Central, Microsoft has pulled back internal Claude Code licenses and is shifting developers toward GitHub Copilot CLI
• The reported change affects internal tooling choice rather than a public product launch or pricing update
• GitHub Copilot CLI appears to be the replacement path cited in the report, aligning the workflow with Microsoft's own GitHub developer stack
• The report frames cost control as the likely reason for the change, but Microsoft has not publicly confirmed that rationale
• No official Microsoft statement, migration scope, or team-by-team rollout details were included in the cited material
• The move suggests tighter internal standardization around Microsoft-owned AI coding tools instead of third-party command-line agents
• For developers, the practical impact would be reduced access to Claude Code inside Microsoft and a stronger internal preference for Copilot-based CLI workflows

Source: windowscentral.com
More: techradar.com · venturebeat.com · news.hada.io

Railway says Google Cloud account suspension caused 8-hour outage

• Railway said Google Cloud incorrectly placed its production account into suspended status on May 19, causing a platform-wide outage from 22:20 UTC to 06:14 UTC on May 20
• The suspension took Railway's API, dashboard, databases, and Google Cloud-hosted compute offline, with users seeing 503 errors and failed logins early in the incident
• Railway said edge proxies depended on a Google Cloud-hosted control plane API for routing tables, and when cached routes expired, workloads on Railway Metal and AWS also became unreachable and returned 404 errors
• Access to the Google Cloud account was restored at 22:29 UTC, but persistent disks, networking, and compute required separate recovery steps, extending the outage by several hours
• Persistent disks began returning at 23:09 UTC and were fully ready by 23:54 UTC, while networking was restored around 01:38 UTC and core API, dashboard, and OAuth endpoints were confirmed operational by about 04:00 UTC
• Builds and deployments were paused during recovery to avoid overload, and Railway later drained a backlog of queued deploys incrementally after infrastructure came back online
• GitHub also rate-limited Railway's OAuth and webhook integrations during recovery, temporarily blocking some logins and builds as retried requests surged after cache resets
• Railway said it takes responsibility for the architectural dependency that allowed a single upstream provider action to cascade across all regions and said it is preparing preventive changes

Source: blog.railway.com
More: theregister.com

Jocoletter curates AI, software, and product trends for developers and builders.

#Microsoft #Railway

Subscribe to Jocoletter