Canonical vs HashiCorp

Canonical announces Zephyr 26.04 LTS, delivering up to 15 years of support for MCU-grade devices

Canonical today announced the upcoming release of Zephyr 26.04 LTS, an enterprise distribution providing a trusted Real-Time Operating System (RTOS) to solve critical Cyber Resilience Act (CRA) compliance and developer experience challenges. Also announced: Ubuntu coming soon to Snapdragon X2 Series platforms, and a new kernel release strategy for faster CVE fixes.

Migrate Cloud TPU API Workloads to Compute Engine

This technical migration guide explains moving TPU workloads from Google Cloud's deprecated Cloud TPU API to Compute Engine instances. The Cloud TPU API is no longer under active development and future TPU hardware generations (starting with TPU7x) are supported only through Compute Engine or Google Kubernetes Engine. Migration requires flag and command mapping (e.g., accelerator-type -> machine-type, tpu-vm ssh -> compute ssh), checking different quota metrics (preemptible vs family quota) and provisioning models (FLEX_START, SPOT, STANDARD, RESERVATION_BOUND), and adjusting startup scripts and images (some Compute Engine accelerator images lack tools like docker). The guide documents practical troubleshooting: using SPOT to probe capacity, checking both quota metrics via the Cloud Quotas API, handling silent failures where RUNNING != ready, and other pitfalls encountered during real migrations.

• Google's Cloud TPU API is no longer under active development; new hardware generations starting with TPU7x are supported only via Compute Engine or GKE.
• Compute Engine uses different flags and flows (e.g., --machine-type=ct6e-standard-1t, --image-family, --request-valid-for-duration, --provisioning-model=FLEX_START) compared with the Cloud TPU API.
• Flex-start provisioning on Compute Engine consumes preemptible quota (PREEMPTIBLE-TPU-V6E-per-project-region) and falls back to the family quota; quota and capacity are separate and reported by different APIs.

IBM Launches AI Agent Identity and Control Suite

IBM announced new capabilities at TechXchange 2026 to secure and govern AI agents in the enterprise. The portfolio includes Agent Identity for HashiCorp Vault, which gives each agent a verified identity and enforces least-privilege access with real-time revocation. Vault 2.2 adds agent-specific accounts, auto-renewing credentials, and post-quantum encryption. IBM watsonx Orchestrate introduces Discover, Contain, and Evaluate capabilities to unify agent discovery, sandboxing, and performance testing across platforms like Amazon Bedrock, Microsoft Foundry, and Google Gemini. The company also launched Secure Hub for centralized Vault visibility, Integration Intelligence Hub for agent context, and IBM Z Atlas for mainframe change intelligence. IBM is partnering with NVIDIA to integrate with OpenShell runtime. The announcements aim to address the challenge of AI agent sprawl, citing Gartner's prediction that Fortune 500 enterprises will use over 150,000 agents by 2028.

• IBM announced Agent Identity for HashiCorp Vault, giving AI agents a governed identity with least-privilege access and real-time revocation.
• HashiCorp Vault 2.2 is generally available, featuring agent-specific accounts, auto-renewing credentials, and post-quantum encryption.
• IBM watsonx Orchestrate introduced Discover, Contain, and Evaluate capabilities for cross-platform agent management.

AI/Infra/VC News Roundup: AI Investments, Funding, Security

This newsletter from 'What's Hot in Enterprise IT/VC' compiles recent AI, infrastructure, and venture capital news. It highlights trends in AI funding, including Gimlet Labs' $300M Series B at a $3B valuation with multiple tranches, and Clay's $115M Series D at $7.1B. It also covers the decline in AI spending among top enterprises, the increasing demand for AI compute, and the emergence of cybersecurity threats using AI, such as Anthropic's report on Iran's misuse of Claude. The roundup includes commentary on the AI infrastructure buildout, the acquisition of Miro for $1.355B, and the rise of physical AI deployments like Skild AI's $100M ARR. Overall, it reflects the dynamic and rapidly evolving landscape of AI technology, infrastructure, and investment.

• Gimlet Labs raised $300M Series B at $3B valuation, led by a16z, with money in at $2.5B, $3B and higher tranches.
• Clay raised $115M Series D at $7.1B valuation, with clients including Anthropic, Google, OpenAI, Stripe, Visa, and UPS.
• AI spend among top 1% of businesses declined 10% in August to $7.2K per employee per month.

AI Projects Close PRs, Deploy Agent 'Software Factories'

AI-native open source projects are increasingly shutting down external pull requests and using agent-based 'software factories' to manage contributions. Vercel deployed such a system for its AI SDK project, which now authors 25–35% of merged PRs and closes 70–80% of issues. The Astro web framework adopted agent-driven triage and regained control of its backlog. Flue and tldraw now automatically close external PRs, converting them into issues or discussions, partly to prevent 'drive-by AI slop PRs.' Maintainers say they trust internally optimized agents more than community-generated code, though they acknowledge risks for community onboarding. Mitchell Hashimoto predicts large open source projects will eventually close contributions completely, while projects still invite reporting, discussion, and perspective from outside contributors.

• Vercel's agent-based 'software factory' for AI SDK authors 25–35% of merged PRs and closes 70–80% of issues within four weeks.
• Flue and tldraw automatically close external pull requests and convert them into issues or discussions.
• The Astro web framework adopted agent-based triage to reestablish control over its GitHub issue backlog.