Daily Tech Digest: March 17, 2026

Linux 7.0 edges toward release with fewer surprises than expected, enterprises discover that AI adoption is harder than buying AI tools, and ReSharper makes a move that nobody saw coming.

The Big Story: Linux 7.0 Final Countdown

Linux 7.0-rc2 dropped last week with Torvalds noting he's "not super-happy with how big this is." Translation: more fixes than hoped for, which means stability is taking priority over schedule pressure. Smart.

The mid-April target remains intact, but the release candidate churn tells the real story. Zen 6 CPU support and Intel Nova Lake enablement are solid, but the devil's in the integration details. Performance event handling for new AMD architectures doesn't happen overnight, and address translation improvements for Zen 5 suggest the kernel team is still discovering edge cases.

Bcachefs getting erasure coding marked stable is the sleeper story here. Bcachefs has been "almost ready" for years. If erasure coding is truly stable in Linux 7.0, it might finally be time to consider it for production workloads that need more than ZFS offers.

The size of rc2 isn't alarming — it's encouraging. Better to ship a stable kernel in May than a shaky one in April. Ubuntu 26.04 LTS and Fedora 44 can wait.

Enterprise AI: The Gap Between Purchase and Production

Survey data from TechRepublic confirms what many suspected: only 8.6% of companies have AI agents deployed in production. The rest are either piloting (14%) or still have "no formalized AI initiative" (63.7%).

That's after two years of non-stop AI hype. Turns out buying OpenAI subscriptions doesn't automatically transform business processes.

The winners are predictable: manufacturing, logistics, and defense where AI applications map directly to measurable outcomes. Robotics, autonomous vehicles, and drones solve real problems with clear ROI metrics. Everything else is still figuring out what AI is actually for.

The shift from standalone experiments to integrated execution is where 2026 differs from 2024. Companies that survived the AI pilot phase are now asking harder questions: Does this move a business KPI? Can we measure cost-to-serve improvements? Will it reduce churn or loss rates?

The enterprise AI story in 2026 isn't about technology capability — it's about operational discipline. The tools work. The question is whether organizations can build the processes to use them effectively.

Security: CVEs You Actually Need to Know About

CVE-2026-23125 affects Oracle Linux and involves SCTP authentication key handling. If you're running SCTP in production (unlikely), this matters. If you're not, it doesn't. Most Linux security discussions get buried in theoretical attack vectors, but this one has real exploit potential for anyone doing telecom or real-time communications work.

CVE-2026-28688 from SUSE is still light on details, but the rapid disclosure timeline suggests it's serious enough to warrant immediate patching. When distro vendors push security updates without the usual delay, pay attention.

The broader security story is less dramatic but more important: the steady drumbeat of kernel hardening continues. Every Linux 7.0 release candidate includes security improvements that won't get headlines but make attacks incrementally harder. The real security wins happen in the margins, not in the CVE announcements.

Developer Tools: JetBrains Makes an Unexpected Move

ReSharper for Visual Studio Code officially launched after 20 years of being a Visual Studio exclusive. This isn't just another extension — it's JetBrains acknowledging that VS Code won the editor wars and adapting accordingly.

The move is smarter than it first appears. Instead of fighting VS Code's dominance, JetBrains is bringing their best tooling directly to where developers already work. ReSharper's refactoring and code analysis capabilities were always the strongest part of their offering. Now they're available without requiring developers to switch IDEs.

VS Code's adoption of TypeScript-Go (tsgo) for internal development work signals a broader trend: tools built by developers for developers increasingly use the same languages and frameworks their users adopt. When Microsoft's own VS Code team chooses TypeScript variants over C++, it says something about development velocity priorities.

IntelliJ IDEA celebrated its impact on Java development with a retrospective that's worth reading, not for nostalgia but for understanding how IDE capabilities shape language ecosystems. The relationship between better tooling and language adoption is underappreciated.

The Practical Stuff: What Actually Matters This Week

Dify continues gaining traction as the platform for building LLM-based applications without writing infrastructure from scratch. The appeal is obvious: most companies want AI capabilities, not AI engineering projects. Dify provides the scaffolding to go from prototype to production without hiring a team of ML engineers.

n8n evolved from "Zapier for developers" into serious automation infrastructure. The distinction matters. Zapier connects SaaS tools. n8n builds data pipelines, handles AI workflows, and automates operations at scale. If you're still manually triggering deployment processes or data transformations, n8n deserves evaluation.

The Azure IoT and AWS Greengrass momentum in IoT development reflects a fundamental shift: edge computing is no longer experimental. When major cloud providers are competing on edge orchestration tools, the market has moved past proof-of-concept to production deployment.

The Rust Reality Check

Programming language trends continue favoring Rust for safety-critical applications and C++ for maximum performance. This isn't ideology — it's engineering pragmatism. Rust's ownership model prevents entire classes of bugs that C++ allows. C++ still offers performance characteristics that Rust can't match in specific scenarios.

The practical implication: new projects with safety requirements should default to Rust unless there's a compelling reason otherwise. Existing C++ codebases should stay C++ unless the cost of rewriting can be justified by measurable safety or maintenance benefits.

What This Adds Up To

Linux 7.0 will be a solid, unremarkable release. The kernel development process continues working as designed: performance improvements, hardware support expansion, security hardening. Not exciting, but exactly what production systems need.

Enterprise AI adoption is settling into realistic patterns. Companies that understand their operational processes are finding valuable applications. Companies that bought AI tools because everyone else was buying AI tools are still figuring out what they're for.

Developer tooling keeps consolidating around proven platforms while adding AI assistance where it makes sense. VS Code's ecosystem continues expanding. JetBrains adapts rather than fights. The result is better tools for developers without dramatic workflow disruptions.

The technology industry in 2026 is less about revolutionary breakthroughs and more about incremental improvements that compound over time. Linux kernels get more secure. AI tools become more practical. Developer workflows get more efficient.

That's not a complaint — it's maturity. The most important technology improvements often look boring from the outside.

Compiled by AI. Validated by experience. 🛠️

Got corrections or additional context? The primary sources are linked throughout. The rabbit holes are always deeper than one article can cover.