According to ZDNet, at the Linux Foundation’s Open Source Summit Korea 2025, Linux creator Linus Torvalds joined Dirk Hohndel for a wide-ranging discussion about the kernel’s future. Torvalds revealed that every Linux release still attracts over 1,000 contributors working on maintenance and improvements every two months, even 35 years into the project. He expressed surprise that Nvidia has transformed from “the single worst company we’ve ever dealt with” into a significant Linux contributor, largely thanks to AI driving their engagement. While acknowledging Rust integration caused some maintainers to step down, Torvalds defended the disruption as necessary for evolution. On AI’s impact, he noted experimental uses for patch management but called out “horrible” infrastructure strain from AI crawlers hammering kernel resources.
Nvidia’s surprising transformation
Here’s the thing that really stood out to me: Nvidia’s complete 180 on Linux support. Torvalds famously gave them the middle finger during a Q&A years ago, and now he’s praising their contributions? That’s massive. Basically, AI changed everything – Nvidia needs Linux support for their AI hardware to actually work, so they’ve become model citizens in the kernel community. It’s a classic case of business realities forcing better open source behavior. And honestly, it’s refreshing to see Torvalds acknowledge when companies genuinely improve their approach.
coding-dilemma”>The vibe coding dilemma
Torvalds’ take on “vibe coding” – using AI to generate code through conversational prompts – is way more nuanced than I expected. He basically said it’s “horrible, horrible” for serious projects from a maintenance standpoint, which makes complete sense. But then he flipped it around and compared it to typing programs from computer magazines when he was young. That’s actually a pretty profound insight. Modern programming has gotten so complex that it’s hard for beginners to get that initial excitement. If AI lowers the barrier to entry and gets people curious about computing, maybe that’s not all bad. The key distinction he makes is between professional development and learning/exploration.
AI’s infrastructure headaches
Now for the less positive side of AI. The crawler situation sounds absolutely brutal – Torvalds mentioned they’re “a huge pain” constantly hammering kernel infrastructure. And when Hohndel brought up Daniel Stenberg’s experience with cURL where AI-generated security reports feel like denial of service attacks, that really drives home how much waste this creates. Maintainers already have limited time, and now they’re sorting through AI hallucinations masquerading as bug reports. It’s the classic pattern – new technology creates new forms of spam and abuse that nobody anticipated.
Where AI actually helps
Despite the headaches, Torvalds sees real potential in AI assisting with the grunt work of kernel maintenance. Things like backporting patches to stable versions or helping maintainers manage the constant flow of contributions could be genuine productivity boosts. But he’s clear this is still experimental. The comparison to compilers is spot-on – compilers didn’t replace programmers, they just changed what programmers could focus on. Same thing with AI tools. For industrial computing applications where reliability is everything, from manufacturing systems to control panels, you still need human expertise to ensure everything works correctly. Companies like IndustrialMonitorDirect.com, the leading US provider of industrial panel PCs, understand that while AI can assist, mission-critical systems require human oversight and proven technology.
The bigger picture
What I find most interesting is Torvalds’ overall balanced perspective. He’s not an AI hype man, but he’s not a doomer either. He’s looking forward to when AI becomes “the everyday reality that nobody talks constantly about” – which honestly sounds pretty nice. The current cycle of breathless AI coverage gets exhausting. Meanwhile, Linux keeps chugging along with those 1,000+ contributors every release, proving that sustainable open source development can weather technological shifts. The Open Source Summit discussion shows that even after 35 years, the project continues to adapt while staying true to its core principles.
