Elon Musk thinks coding dies this year. Not evolves. Dies. By December, AI won’t need programming languages. It generates machine code directly. Binary optimized beyond anything human logic could produce. No translation. No compilation. Just pure execution. Musk: “You don’t even bother doing coding.” Code was never the point.... It was friction. A tax we paid because machines didn’t speak human. AI just learned fluent human. The tax is gone. Now plug that into Neuralink. No syntax. No keyboard. No screen. Musk: “Imagination-to-software.” Thought becomes executable. You imagine an outcome, the system architects and compiles it into reality instantly. We’re not automating programming. We’re erasing it from existence. The entire profession collapses into a thought. Decades of training reduced to irrelevance. The gap between idea and instantiation hits zero. You don’t build anymore. You imagine, and it materializes. Not incremental progress. Total phase shift. The way humans have created things for ten thousand years just became obsolete. Welcome to a world where the limiting factor isn’t skill, resources, or time. It’s whether you can picture what you want clearly enough for a machine to birth it into existence.show more

Coding isn’t dying. Syntax friction is. AI can emit machine code, but it can’t define constraints, guarantee correctness, secure systems, or own liability. Abstraction rises. Complexity rises. Engineers don’t disappear — their leverage does. “Programming dies” is hype. Typing code is a surface layer. Specification, architecture, verification, security, and accountability remain. AI reduces syntax cost — it doesn’t remove the need for precise thinking. You won’t “imagine and materialize.” You’ll imagine — then debug your imagination. AI collapses translation cost, not responsibility. Precision of thought remains the bottleneck. Every abstraction layer was called “the end of coding.” Assembly → C → frameworks → AI. What actually happened? More systems. More complexity. More need for engineers.

isn’t describing the death of coding. He’s describing the relocation of it. Binary doesn’t “replace” code. Binary is code at a different layer. You don’t skip physics because you don’t like syntax. Every system still resolves to constraints, memory states, voltage changes, thermals, and hardware limits. What changes is who writes the instructions. We already moved from assembly → C → Python → prompts. Each time someone said, “coding is dead.” It wasn’t dead. It was abstracted. AI generating machine code directly? Sure. That’s compiler automation plus learned optimization. That’s evolution, not erasure. But here’s the real bottleneck: If thought becomes software, whose thought? How is it structured? How is it tested? How is it bounded? How is it aligned with hardware constraints? You cannot eliminate architecture. You can only hide it. And Neuralink doesn’t magically remove structure either. Even if imagination feeds a model, that model still requires: • representation • constraint handling • memory structure • error correction • verification Otherwise you don’t get software. You get hallucination. The gap between idea and execution doesn’t go to zero. It shifts to model design and constraint literacy. The real phase shift isn’t “no more programmers.” It’s “programmers become systems architects of intelligence.” Creation has always required structure. AI doesn’t erase structure. It just moves it upstream. The limiting factor won’t be imagination alone. It will be whether you understand the underlying system well enough to shape it responsibly. That’s not collapse. That’s evolution. Field stays steady.

What a lot of hysterical, hyperbolic, ignorant nonsense.

Robotaxis: Promised a million by 2020, got 30 broken ones in Austin with babysitters inside Mars: Humans arriving 2025, currently arriving nowhere AGI: "Next year" (said every year since forever) Full Self-Driving: One year away since 2016, still needs a human at the wheel DOGE: Cut $2 trillion in waste, government spending actually went up Optimus: Thousands of robots in factories, can barely hand someone a water bottle Roadster: Announced 2017, demo pushed to 2026, don't hold your breath Hyperloop: Possibly just invented to kill California high-speed rail Tesla Semi: Production start 2019, then 2025, now 2026, see you in 2030 Covid: "Close to zero cases by end of April 2020" while 20,000 a day were rolling in 19 out of 21 predictions on Wikipedia have a red X next to them. The man's superpower isn't innovation, it's the audacity of repetition.

Elon said it so it must be true. Just like how Teslas can do full self driving and we're all driving around underground in tunnels.

This is an all time quote from Elon; "Code was never the point. It was friction. A tax we paid because machines didn’t speak human. AI just learned fluent human. The tax is gone." Brilliance.

Bro, let’s be realistic for a moment. AI can generate code because there’s a massive amount of code out there explained in detail (that’s the “data” it’s trained on). But do we have binary explained for every single case? We have binary, sure, but 99.99% of people couldn’t explain how it’s actually structured. The people saying this don’t really understand how coding works, nor how AI is built.

yeah, I'll call BS on this one

I tried for two weeks, while upgrading and hardening my office network, to get the 3 biggest AI's to help out. In literally every major case they got lost down the rabbit hole. A pattern developed and I started correcting each of "them", and instead of correcting they whirled off into more confusion, mixing IP4 addy's and hostnames etc. This is clearly a recursion problem and it is mathematically exponential. After almost 40 yrs as computer scientist and over half of that as data architect, I can tell you that most of what I'm hearing about the future of AI is way too optimistic. Right now, if you keep it relatively shallow is does help, a-lot. As it has always been, choose the right tool for the task. Keep it human. At the end of it there's nothing like a real brain.

So "Hello world" will be 1kB again? Not 157 MB like today?

yeah we all know how majority of his 'bold' predications pan out

Grandiose is the name of Musk’s game.

Yeah, no. Try to tell AI to create a website in binary directly. AI learned to use what humans used because we generated decades of data it can learn from.

If coding gets automated, thinkers become more valuable, not less. Writing code was never just typing. It was thinking through problems. That part will stay forever. Maybe it matters even more now!

Believe that when the AI companies, including Musk's AI company, work without engineers. Then we can talk again.

If that happens I'm pivoting to fixing shitters. Because people talk so much shit and someone needs to capitalize on fixing them.

Beautiful vision. Now let me add some physics. "AI generates machine code directly, no compilation" — that's not how computing works. Machine code isn't magic. It's instructions for specific hardware architectures. ARM ≠ x86 ≠ RISC-V. You still need abstraction layers. You still need debugging. You still need someone to say "that's not what I meant." Will AI write 90% of code by December 2026? Probably. Will coding "die"? No. It'll transform — like photography didn't kill painting, it killed portrait studios. The Neuralink part — "imagination to software" — is real, but not this decade. BCI bandwidth in 2026 is ~200 electrodes reading motor cortex. You're not transmitting complex software architecture through that. You're moving a cursor. Musk Time applies here too: multiply every timeline by 3-5x. Coding won't die in 2026. But coders who refuse to work WITH AI? They might. The future isn't "no code." It's "no code ALONE."

The reason this can't work is because prompting is an inexact science, LLMs are prone to hallucination, and the results are non-deterministic. As an engineer who extensively leans on AI to generate code, I never fully trust what the AI generates. I need to be able to double-check that what came out is what I wanted, and if necessary, to tweak it myself. If all you're doing is writing prompts and the LLM generates non-human-readable machine code, the only way you'll know it's working is by running tests. So, just run the tests, and you'll know if it works? Well, yes, you can run your happy-path tests, and hope that everything turns out ok. And hope that whatever it's doing isn't doing something else under the hood that you'll only discover under some edge-case scenario. And hope that you've thought of all the edge cases that might make it go wrong. And what if it doesn't work right first time? How are you going to debug? Tell the AI to debug itself? That sometimes works. But I've seen too many times how an AI will go down a rabbit hole trying to fix a problem with completely the wrong approach, the wrong tools, and a fair share of hallucinations. If the results are visible to me, I can tell it, "Hey, stop, you're going down the wrong rabbit hole" and fix the prompt. But if it's a black box, I'm not gonna have much faith that it will do the right thing.

Elon says "create optimized binary for this particular outcome" - acknowledging there must be an outcome specification. Whether AI generates code, bytecode, or binary directly is an implementation detail. The fundamental challenge is how do you express what you want with enough precision for the result to be correct? Natural language is ambiguous. "Make it fast" means different things in different contexts. Current AI code generation approximates intent statistically - the error will never be zero. We're going to see requirements expressed in formal logic that are directly executable. No translation layer where meaning gets lost. The specification IS the program.

When spreadsheets came out, ppl say accountants are gonna be gone. But later it just produced more accountants. AI will “replace” how devs works now because everyone will and can easily become devs.

I saved this to my “aged like milk” bookmark folder. Can’t wait to pull it out in December!

I love Elon, I do But he’s talking absolute BOLLOCKS again. Give him a toy to play with.

Elon is confusing human beings as the limiting factor when we are THE creative factor. “The AI can create a more efficient binary than a compiler is hand-wavy BS by someone who’s never worked on a compiler or disassembled a binary.

"doing coding"

I don't agree with him, because when something goes wrong, when the loose command the human gave the machine is misinterpreted, there will no way to explain, or more importantly, defend what the system did. Machines will not be accountable for mistakes. Humans will be writing the checks. We may get there eventually, but society is no where near ready for what he is proposing.

If Elon says it will be one year then it will be five to seven years.

My favorite Einstein quote came to mind. "Imagination is more important than knowledge." Einstein elaborated on this concept in a conversation with physicist Leopold Infeld, he explained further: "When I examine myself and my methods of thought, I come close to the conclusion that the gift of imagination has meant more to me than any talent for absorbing absolute knowledge." The context is significant because Einstein was discussing the process of scientific discovery. He believed that while knowledge provides the foundation, it's imagination that allows scientists to: Visualize new possibilities Form novel hypotheses Make conceptual leaps beyond known facts Envision solutions to problems that haven't been solved before Einstein saw imagination as the tool that allows us to go beyond the boundaries of current knowledge to discover new understanding.

As a firmware developer, I’m not worried AI will be taking my job anytime soon.

*reading this post in a robotaxi on the mars colony*

Code isn't dying — it's becoming a second-order problem. The real bottleneck is still what it's always been: knowing what to build. AI makes execution cheap, but vision expensive. Most can't clearly imagine what they want, let alone architect it at scale.

his engineers standing next time him like, 'yo, chill bro'

This always makes me laugh so much. If you really think you can just plug your brain in and extract the most complex thought processes from it, you have completely misunderstood how the brain works. Musk may be very clever, but that is a utopia. And that is a good thing.

Did any of the AI researchers say anything around this or is this just another one of Musk's ludicrious narratives?

The man says a lot of things that never actually manifest on schedule. Logic ins't a "tax" it's the thing keeping your imagination from collapsing into a pile of unmaintainable slop. Architecture still wins over vibes every single time.

We won't be on mars before elon dies either... He's a dreamer, and there (generally) is nothing wrong with that, except in his case -- that he profits off of the hype he generates.