Torvalds Tells AI Critics to 'Fork It or Walk Away,' Declares Linux 'Not One of Those Anti-AI Projects'
Multi-perspective analysis. Each perspective deliberately argues one viewpoint; none represents the editorial position of qalarc.
In a July 15, 2026 post to the Linux kernel mailing list, Linus Torvalds declared that 'Linux is not one of those anti-AI projects' and told developers who object to others using AI tools that they can 'do the open-source thing and fork it. Or just walk away.' The remarks came during a thread about Sashiko, an agentic AI-based kernel code-review tool, and mark a clear shift from Torvalds' more skeptical public stance on AI in 2024.
What the terms mean (5)
- LKML (Linux Kernel Mailing List) β The primary public email forum where Linux kernel developers discuss, review and coordinate changes to the kernel.
- Sashiko β An agentic AI-based kernel code-review tool at the center of the thread, whose creators claim it can independently identify about 53.6% of bugs later fixed by human developers.
- Agentic AI β AI systems that can autonomously carry out multi-step tasks β such as scanning code and flagging bugs β with limited human prompting.
- Signed-off-by β A tag added to a kernel patch by which a human developer certifies the origin and licensing of a contribution; kernel policy requires a person, not an AI agent, to provide it.
- Fork β In open-source, to copy a project's codebase and continue development independently β the recourse Torvalds pointed dissenters toward.
The facts (8)
- On July 15, 2026, Torvalds wrote on the LKML that he was willing to 'absolutely put my foot down as the top-level maintainer' and that critics of others using AI 'can do the open-source thing and fork it. Or just walk away.' [1][2]
- Torvalds framed AI as a useful tool, saying its utility 'may not have been that clearly even just a year ago, but it's no longer in question today,' and that 'anybody who doubts that clearly hasn't actually used it.' [2][4]
- He stated that developers are NOT mandated to use AI, but that he 'will very loudly ignore people who try to argue against other people from using it.' [3][4]
- The discussion arose in a thread about Sashiko, an agentic AI code-review tool whose creators claim can independently find roughly 53.6% of bugs later fixed by human coders. [1][2]
- Torvalds pushed back on ideological objections, writing that anyone pointing to AI's flaws 'had better be looking in the mirror... because it's not like natural intelligence is always all that great either,' and framed the matter as one of technical merit, 'not fear of new tools' and 'not a social war.' [3][5]
- Media subsystem maintainer Laurent Pinchart argued the skeptical/anti-AI side in the thread; Torvalds responded that the discussion 'ends' and that personal ethics should not be enforced on others. [3]
- Despite the permissive stance, kernel policy still requires human accountability β for example, AI agents are not permitted to add Signed-off-by tags β making this a conditional acceptance rather than an unconditional endorsement. [5][6]
- The position is a shift from Torvalds' 2024 comments dismissing much AI as hype; senior maintainer Greg Kroah-Hartman had separately said as early as March 2026 that AI-assisted bug reports and code reviews had improved dramatically. [7][2]
Context & background
Torvalds created Linux in 1991 and remains its top-level maintainer, giving his stated positions unusual weight over kernel development norms. In 2024 he publicly characterized much of the AI industry as hype, so his July 2026 remarks represent a notable evolution. [7] The immediate trigger was a mailing-list debate over Sashiko, an AI-based agentic code-review tool, and over whether AI-assisted contributions belong in the kernel workflow at all β a debate that also touches longstanding kernel rules on human accountability, such as the requirement that a human developer sign off (via the Signed-off-by tag) on submitted patches. [1][5] The story was reported across technology outlets including The Register, Ars Technica, Tom's Hardware, Phoronix, Computerworld and Virtualization Review. [2][3][4][5] Some reporting initially framed the statement as a rumor or something Torvalds might do; in fact the LKML post is confirmed and dated July 15, 2026.
Still unresolved
- How, in practice, will the kernel community enforce human accountability rules (such as the Signed-off-by requirement) as agentic tools like Sashiko become more common in the review pipeline?
- Will tools like Sashiko's claimed ~53.6% bug-detection rate hold up under independent scrutiny and real-world kernel workloads?
- Whether other senior maintainers will align with Torvalds' stance or continue to press the skeptical case raised by figures such as Laurent Pinchart.
The same story, argued three ways. Pick an angle β the facts above stay the same.
π§ Cui bono β who benefits?
Beneficiaries
- Microsoft, Google, Amazon (hyperscale cloud providers) β Normalized AI integration in critical infrastructure creates dependency on their GPU/TPU compute and model APIs
via As Linux kernel development adopts AI tooling, the computational substrate shifts from developer workstations to cloud-hosted LLM inference. Kernel contributors needing AI assistance must rent tokens/compute from hyperscalers. Long-term: Linux itself becomes optimized for AI workloads (their primary revenue driver), creating a feedback loop where the OS co-evolves with their business model. - GitHub (Microsoft subsidiary) and AI coding assistant vendors β Linus's endorsement provides top-tier social proof for AI in software engineering workflows
via The Linux kernel is the crown jewel of open-source credibility. Torvalds's rejection of AI bans removes the last principled objection from engineers skeptical of Copilot/Cursor/etc. Enterprise procurement teams can now cite 'even Linus uses it' to mandate AI tooling, expanding GitHub Copilot's enterprise seat sales and locking developers into Microsoft's ecosystem. - Proprietary AI labs (OpenAI, Anthropic, Google DeepMind) β Validation that AI can handle high-complexity, high-stakes engineering tasks
via Kernel development is uniquely unforgivingβmemory safety bugs kill systems, subtle race conditions create security holes. If AI tools prove valuable here, it demolishes the 'AI is only good for boilerplate' critique and justifies premium pricing for o1/Claude-3.5-class models marketed to enterprises for critical infrastructure work. - Junior/offshore developers and code-review-as-a-service platforms β AI that writes acceptable kernel patches compresses the skill gradient, commoditizing mid-tier engineering
via If AI can generate patches that pass Linus's review, the scarcity value of knowing C, the kernel subsystem internals, and LKML norms evaporates. Firms can hire cheaper developers to prompt-engineer patches rather thanεΉε » decade-long kernel expertise. Simultaneously, creates demand for AI-powered code review services to catch what humans can no longer afford to.
Who loses
- Senior kernel developers whose expertise in memory management, scheduler internals, and driver architecture becomes less scarce if AI can replicate their output
- Open-source purists who view AI training on GPL code (without contributor consent) as license-washingβusing copyleft work to build proprietary models
- Hardware vendors (Nvidia competitors like AMD/Intel) if Linux optimizations increasingly favor the GPU architectures used for training the models doing the optimizing
- Universities and bootcamps selling 'learn systems programming' curriculaβif AI handles low-level details, demand for that skillset craters
Rivalry & conflicts of interest
- Independent/self-hosted developer workflows harmed β GitHub/Microsoft and cloud-based IDEs (Cursor, Replit, Gitpod) gains
conflict of interest: Microsoft funds the Linux Foundation and employs numerous kernel maintainers. GitHub (Microsoft) hosts the canonical kernel mirrors and CI infrastructure. Torvalds's acceptance of AI toolingβprimarily delivered via Microsoft's Copilot and OpenAI APIβsteers the kernel's development pipeline toward Microsoft's platforms. - Local/offline development environments harmed β Cloud compute providers (AWS, Azure, GCP) gains
conflict of interest: Linux Foundation's Platinum members include Microsoft, Google, Meta, Huaweiβall cloud or AI hardware vendors. As AI assistance becomes standard, the compute requirements exceed laptop GPUs, forcing reliance on cloud-hosted models. The foundation's funding structure aligns kernel governance with hyperscaler interests. - European/Chinese open-source AI initiatives (Mistral, Baichuan, open-weights models) harmed β U.S. frontier labs (OpenAI, Anthropic, Google) gains
conflict of interest: No direct financial tie known, but U.S. export controls (H100/MI300 chip bans to China) mean only Western labs can train models powerful enough for kernel-level work. Linus's endorsement sets a capability bar that Chinese labs cannot meet under current sanctions, entrenching U.S. AI dominance in critical infrastructure.
Ramifications (follow the chain)
- AI-assisted kernel development β patches authored by models trained on proprietary datasets β Linux becomes dependent on closed AI services β GPL's 'four freedoms' hollowed out (you have source code, but the toolchain to modify it is a rented black box controlled by three U.S. corporations).
- Kernel optimizations guided by AI trained on cloud workloads β Linux increasingly tuned for hyperscale VM/container patterns β bare-metal and edge use cases (IoT, embedded, gaming) deprioritized β those markets either fragment to custom kernels or pay cloud vendors for 'Linux-compatible' managed services.
- Acceptance of AI in kernel β other critical infrastructure (compilers, databases, network stacks) follow β entire software supply chain depends on handful of model providers β single points of failure (API outage, model alignment shift, geopolitical API cutoff) can freeze development of civilization's digital infrastructure.
- AI handles low-level kernel work β new generation of Linux developers never learn the 'why' behind memory ordering, RCU, or lockless algorithms β tribal knowledge evaporates β when AI makes a subtle mistake (race condition visible only under rare hardware timing), no human can debug it β silent reliability degradation in production systems running finance, power grids, hospitals.
intentional reading Microsoft, having spent billions on OpenAI and GitHub Copilot, needs AI coding tools to escape the 'expensive party trick' critique and justify their revenue projections. The Linux kernelβunimpeachable in rigor, led by an uncompromising figureheadβis the perfect Schelling point for legitimacy. Torvalds's dismissive rejection of AI skeptics ('as long as I'm in charge, no ban') reads as orchestrated: give the industry's most credible gatekeeper early access to frontier models, let him experience productivity gains, then amplify his endorsement to override every corporate engineering team's hesitation. The conflict is structural: Linux Foundation funding comes from Microsoft, Google, Metaβthe same vendors selling AI tooling. Torvalds may believe he's making an independent technical judgment, but the incentive landscape ensures that judgment aligns with his funders' strategic needs. The alternativeβLinus rejecting AIβwould cost Microsoft billions in Copilot credibility and force a humiliating pivot. Much cheaper to ensure the kernel's development pipeline becomes a showcase for AI capabilities, even if it means surrendering open-source toolchain independence to proprietary model APIs.
structural reading No conspiracy required: AI coding assistants genuinely reduce toil (boilerplate, API lookups, test scaffolding), so any developer trying them experiences local productivity gains. Torvalds, evaluating submissions on 'does it work and is the code clean,' has no reason to care whether a patch came from a human or LLMβhis incentive is kernel quality, and if AI delivers that, he'll accept it. Meanwhile, hyperscalers have independent reasons to fund Linux development (their infrastructure runs on it) and to push AI adoption (their revenue model). These incentives align without coordination: cloud vendors fund the kernel, kernel developers adopt their AI tools because they work, the kernel optimizes for cloud workloads because that's where the patches come from, and the loop closes. Junior developers adopt AI to compete with seniors, seniors adopt it to keep up with juniors, and within 18 months, not using AI in kernel development is as handicapping as refusing version control. The outcomeβLinux development dependent on proprietary AI servicesβemerges from rational individual choices in a landscape shaped by who controls compute, training data, and foundation funding. Structural lock-in, no smoke-filled room needed.
π Trading signals β winners & losers
Tradeable instruments most exposed to this story, inferred from the analysis above. Not financial advice β informational only, generated by AI from forum discussion and may be wrong.
π Likely winners
- β² MSFTstockMicrosoft$399.067d -0.8%β -0.4% since callOwns GitHub Copilot; Linus endorsement validates AI coding tools
- β² NVDAstockNvidia$197.267d -3.0%β -3.4% since callAI kernel development increases demand for GPU compute infrastructure
- β² GOOGLstockAlphabet$334.867d -4.9%β -3.4% since callCloud TPU/GPU services benefit from normalized AI infra dependency
- β² AMZNstockAmazon$230.867d -7.7%β -6.6% since callAWS GPU instances gain from AI-driven development workflow adoption
π Likely losers
- βΌ AMDstockAMD$454.627d -9.7%β -8.3% since callLinux optimizations may favor Nvidia GPUs used in training
π Call performance β day by day
| date | price | vs entry |
|---|---|---|
| 2026-07-18 | $400.70 | +0.0% |
| 2026-07-21 | $400.70 | +0.0% |
| 2026-07-22 | $397.75 | -0.7% |
| 2026-07-23 | $390.34 | -2.6% |
| 2026-07-25 | $381.58 | -4.8% |
| 2026-07-26 | $381.70 | -4.7% |
| 2026-07-27 | $381.70 | -4.7% |
| 2026-07-28 | $389.10 | -2.9% |
| date | price | vs entry |
|---|---|---|
| 2026-07-18 | $204.12 | +0.0% |
| 2026-07-21 | $204.12 | +0.0% |
| 2026-07-22 | $207.29 | +1.6% |
| 2026-07-23 | $212.06 | +3.9% |
| 2026-07-25 | $208.76 | +2.3% |
| 2026-07-26 | $206.84 | +1.3% |
| 2026-07-27 | $206.84 | +1.3% |
| 2026-07-28 | $196.51 | -3.7% |
| date | price | vs entry |
|---|---|---|
| 2026-07-18 | $346.77 | +0.0% |
| 2026-07-21 | $351.99 | +1.5% |
| 2026-07-22 | $347.15 | +0.1% |
| 2026-07-23 | $342.09 | -1.3% |
| 2026-07-25 | $317.69 | -8.4% |
| 2026-07-26 | $319.74 | -7.8% |
| 2026-07-27 | $319.74 | -7.8% |
| 2026-07-28 | $326.56 | -5.8% |
| date | price | vs entry |
|---|---|---|
| 2026-07-18 | $247.23 | +0.0% |
| 2026-07-21 | $249.99 | +1.1% |
| 2026-07-22 | $247.55 | +0.1% |
| 2026-07-23 | $244.85 | -1.0% |
| 2026-07-25 | $232.11 | -6.1% |
| 2026-07-26 | $232.11 | -6.1% |
| 2026-07-27 | $232.11 | -6.1% |
| 2026-07-28 | $231.39 | -6.4% |
| date | price | vs entry |
|---|---|---|
| 2026-07-18 | $495.76 | +0.0% |
| 2026-07-21 | $503.57 | +1.6% |
| 2026-07-22 | $544.43 | +9.8% |
| 2026-07-23 | $552.33 | +11.4% |
| 2026-07-25 | $521.95 | +5.3% |
| 2026-07-26 | $521.95 | +5.3% |
| 2026-07-27 | $521.95 | +5.3% |
| 2026-07-28 | $494.95 | -0.2% |
π See how every call has performed β the full scoreboard & API β
From the threads
The posts that drew the most replies in the source discussion β shown as posted. Reactions ranged across the spectrum; these are the ones people actually engaged with. Each quote links to its archived source thread so you can verify it; quotes we couldn't tie to a source thread are marked source unverified.
Linus has spoken. As long as he's the maintainer, AI will never be banned from Linux kernel development.
nta I live in the balkans with 1 local job where I pay taxes, and 2 jobs where I literally send an invoice and get money deposited to my international account, no one asks a single fucking question I make 15k euros a month AFTER taxes, but everyone has to be stupid and poor like you, so call me a liar lol
I've got 6 full time jobs and a hot trophy wife
hating on AI is such a reddit karma farming opinion or even worse fuck off r******, its an amazing tool that makes 90% of my job trivial
Continue the discussion
Add your own take β replies are kept on this article and can be upvoted.
π Related Analysis
- NEW FEATURE JUST DROPPED feature shared: ai
- Bitcoin price and AI bubble prediction shared: ai
- AI sector sharp downturn causing panic among investors shared: ai
References
- [1] Linus Torvalds To Critics of AI Coding On Linux: 'Fork It. Or Just Walk Away.' β Slashdot
- [2] Linus Torvalds rebukes anti-AI stances in the Linux kernel code review process β Tom's Hardware
- [3] Linus Torvalds Reaffirms That Linux Is Not 'Anti-AI' & Not A 'Social Warrior' Project β Phoronix
- [4] AI OK in Linux development, says Torvalds β Computerworld
- [5] Linux Creator Plants Foot on AI Open Source Contribution Issue β Virtualization Review
- [6] Torvalds Sets Linux Kernel AI Policy: Use It Responsibly, or Fork It β Windows News
- [7] Linus Torvalds on AI (2024 remarks) β Phoronix
β supportive Β· β critical Β· β neutral wire Β· β partisan Β· β state outlet
βΎ Discussion
Select any text in the article to comment on that passage.