Elon Musk: The Singularity is here! 5 years of doctoral work compressed into 1 year, AI rewrites the timeline logic

Midjourney founder Holz states that holiday programming projects are more than those of the past 10 years. Elon Musk replies, “We have entered the singularity,” and declares “2026 is the Year of the Singularity,” igniting the entire internet. Google senior engineer reveals: Google’s one-year project was completed in one hour using Claude Code. Yuchen Jin says that with AI tools, a 5-year PhD could be completed in one year. Time itself is being re-compressed.

Shocking confession from a Google engineer reveals signals of losing control

What truly pushes things into the storm’s eye is a public share from a Google senior engineer. She said a sentence later screenshot and shared by countless people: “A project at Google that took a year to complete was finished in one hour with Claude Code.” She also added, “I’m not joking, this is not funny.” The tweet quickly garnered nearly 7 million views, and even “this is not funny” became a meme.

The weight of this statement lies in the identity of the speaker. This is not from a sensationalist blogger but from a top-tier Google senior engineer, a legendary figure in the Go language community, and a top 1% global tech expert. The feeling is subtle—neither boastful nor marketing (after all, it’s a competitor’s product)—but a professional suddenly realizing that things are out of control, a震動.

Even more interesting, as the whole web discusses this, Claude Code’s creator Boris Cherny personally stepped in, sharing how he uses this tool himself. He didn’t “explain the principles” but demonstrated live: “This is how I use it.” This completely halted the discussion because even the developer himself is using AI to assist in developing AI tools. This recursive evolution is precisely the core feature of the singularity theory.

What is truly unsettling is not “faster,” but “emergence.” Rohan Anil, co-author of Google AI and now at Anthropic, said: “When you piece these components together, what ultimately emerges is fascinating to me.” Lucas Beyer, who has worked at DeepMind and OpenAI and now at Meta, said more directly: “I understand every step it builds, but the result still amazes me.”

The significance of these two statements is: they are not bystanders but are involved in building the model themselves. Emergence means the system as a whole begins to exhibit capabilities that are not explicitly programmed. What is truly alarming is no longer “how fast it is,” but that it begins to act in ways not entirely aligned with human expectations.

Three paradigm shifts brought by time compression

Collapse of the time scale in the education system

Yuchen Jin straightforwardly states that if he had Claude Code and ChatGPT during his PhD, the original 5+ year doctoral cycle could have been completed in about a year. The reason is simple: in his time allocation, coding, writing papers, and reading literature almost occupy all his energy, and AI can accelerate all three by orders of magnitude. More critically, this judgment is not “personal illusion”; his advisor Quanquan Gu also explicitly agrees.

In their view, the reason the PhD training cycle is so long is essentially because these “slow tasks” are artificially prolonged. When these steps are systematically accelerated, “10x speed” is no longer an extreme case but becomes the default setting. Our training pathways from K12, undergraduate to PhD are fundamentally designed for the “pre-AI era.” Large amounts of time are spent on steps that can now be systematically accelerated.

The marginal cost revolution in the software industry

More and more perspectives believe that as AI penetrates more software production processes, the marginal cost of software development is rapidly approaching zero. This means not only lowering the startup barrier but potentially reshuffling the entire value distribution in the tech industry. When coding itself is no longer a bottleneck, the focus of competition shifts to creativity, business models, and execution speed.

Famous investor Paul Graham pointed out that a very real role of AI is “cutting through bureaucratic cycles.” When organizations fall into endless discussions and cannot reach consensus, AI can directly produce a workable 1.0 version. This version may not be optimal, but it forces everyone to start discussing practical solutions rather than remaining in endless talk.

Actionability becomes the only scarce resource

In this era where “thinking” and “doing” are almost instantaneous, hesitation is the biggest cost. In the past, we were used to “think twice before acting” because trial and error was too expensive. But now, AI has shortened the distance from “idea” to “product” to just a few prompts. In this era, the value of “actionability” is being amplified infinitely.

Got an idea? Let AI help you produce a 1.0 version. Inspired? Run it immediately. The world indeed no longer waits for you, but as long as you dare to jump on this fast train, you will find that time is still your friend. Stop clinging to perfect plans and stubbornly sticking to them, because when tasks that once took years or over a decade are compressed into days or even hours, you can no longer describe it as “efficiency improvement.” It’s more like: time itself has been re-compressed once again.