Linux 7.0 — The Version Number Means Nothing, the Code Means Everything
Linus dropped Linux 7.0 on April 13, and before anyone panics: no, this is not some massive architectural overhaul. Torvalds himself clarified that "major version numbers do not necessarily indicate architectural changes." He bumped the number because 6.x was getting unwieldy — same reason he jumped from 3.20 to 4.0 back in 2015. The man just doesn't like big minor version numbers.
That said, the actual contents of this release? Genuinely significant. Let me break it down.
Rust Is No Longer Experimental
This is the big one. Linux 7.0 formally strips the experimental label from Rust support. Rust-based drivers and kernel modules are now first-class citizens in the kernel tree. This has been a multi-year journey since Rust first appeared in Linux 6.1, and the implications are massive.
What this means in practice: subsystem maintainers can now accept Rust code without the "experimental" caveat. Memory safety bugs — buffer overflows, use-after-free, data races — are the single largest category of kernel CVEs. Rust eliminates entire classes of these bugs at compile time. This isn't theoretical anymore; it's production.
Will the entire kernel get rewritten in Rust? No. Should new drivers and modules consider Rust as a first choice? Absolutely. The C diehards will complain, as they always do, but the code quality speaks for itself.
Post-Quantum Cryptography Enters the Kernel

Linux 7.0 adds ML-DSA (Module-Lattice-Based Digital Signature Algorithm) post-quantum signatures for kernel module authentication. At the same time, SHA-1-based module signing has been removed entirely.
Let that sink in. SHA-1 — an algorithm that's been considered broken for over a decade — was still being used for kernel module signing until literally this release. Better late than never, I suppose. The move to ML-DSA is forward-looking: when quantum computers eventually threaten current cryptographic schemes, the kernel's module verification infrastructure will already be hardened.
This is the kind of unsexy infrastructure work that doesn't make headlines but matters enormously. Every kernel module you load — every driver, every filesystem module — now gets verified with post-quantum-resistant signatures.
PREEMPT_LAZY Scheduling
The scheduler gets a meaningful improvement with PREEMPT_LAZY, which defers non-critical background tasks to prioritize latency-sensitive workloads. If you're running a desktop or gaming setup (I use Arch with CachyOS kernel patches, btw), this translates to snappier input response and smoother frame pacing.
The sched_ext framework from 6.12 continues to mature as well. BPF-based custom scheduling policies are getting more capable with each release, and 7.1-rc already has improvements in the pipeline. This is particularly exciting for gaming — projects like CachyOS have shown what custom schedulers can do for frame time consistency.
Memory and swap operations also got faster, with improved zram efficiency and better container startup performance. Server admins running container orchestration will notice the difference.
Hardware Enablement

The usual hardware enablement work continues: Intel Nova Lake and AMD Zen 6 processors get initial support, KVM gains advanced virtual machine capabilities for AMD EPYC 5 CPUs, and there are updates across ARM, RISC-V, and Loongson architectures.
In a somewhat nostalgic move, new code was also added for SPARC and DEC Alpha processors. Yes, in 2026. The kernel still cares about legacy architectures, and honestly, that's part of what makes Linux, Linux.
Self-healing XFS filesystem technology landed as well, improving resilience against data corruption. Between this and the Btrfs improvements that have been accumulating over the last several releases, Linux filesystems are getting genuinely impressive at protecting your data.
Torvalds on AI in Kernel Development
Linus weighed in on AI-assisted development, noting that "AI tools are likely to continue uncovering corner cases" in kernel code. The kernel now has formal documentation for AI-generated code contributions and bug reports — contributors must disclose AI tool usage and remain fully accountable for their submissions.
Greg Kroah-Hartman highlighted AI as useful for bug detection, which is fair. The kernel's test infrastructure can always use more fuzzing, and AI tools are decent at finding edge cases. Just don't submit AI-generated patches without understanding every line. The maintainers will know, and they will not be kind about it.
Who Gets It First?
Ubuntu 26.04 LTS "Resolute Raccoon" ships with Linux 7.0 as its default kernel on April 23. Fedora 44 and Arch Linux (the superior choice, obviously) already have it in their repos. If you're running any rolling release distro, you probably already have it.
For the rest of you — `make menuconfig && make -j$(nproc)` awaits. Or just wait for your distro to package it. Either way, this is a solid release that moves the kernel forward in all the right directions.