Back to Flash News

Elon Musk: AI Will Surpass Humans in Digital Fields by End of Next Year

Elon Musk stated that by the end of next year, artificial intelligence will reach superhuman levels in all digital tasks that do not require "shaping atoms"; he also mentioned a latest timeframe of 2028.

He distinguished between digital tasks and physical operations: programming, research, documentation, and data analysis are included in the prediction, while physical manipulation still relies on robots. He also mentioned that software will reach Stockfish-level next year, making it impossible for humans to compete in coding; within 12 to 18 months, digital software and tasks will become extremely powerful. At Davos, he said that by the end of this year or at the latest next year, a system smarter than any individual will emerge, and by around 2030 to 2031, it will surpass human intelligence overall.

He also predicted that within ten years, there will be 1 billion humanoid robots, each producing about five times more than a human; the global economy could grow by 20% to 30%, equivalent to an additional $20 trillion to $30 trillion annually. SpaceX aims to expand AI computing power to 10 gigawatts by the end of next year, estimating annual revenue of $300 billion to $500 billion based on $30 to $50 per watt, and mentioned that by 2027, about 15 gigawatts of chips might not be powered due to transformers, liquid cooling, and networks.

These statements come as the United Nations discusses safety standards, with some laboratories calling for a slowdown. Musk himself supports mutual inspections of cutting-edge models while stating that it is difficult to hit the stop button. Tesla plans for Optimus to take on more complex tasks within factories and is discussing a timeline for public sales.

This reflects a timeline-driven expectation repricing. Buyers are cloud and robotics buyers planning capital expenditures based on the 2027 to 2028 digital replacement of human labor, while sellers are policy agendas writing safety covenants on a ten-year scale. Funding is shifting from the "slow down training" narrative to computing power, electricity, and humanoid production lines; beneficiaries include SpaceX, Tesla, and their AI businesses with existing chip and factory plans, while those under pressure are employers and training systems focused on digital positions and writing replacement windows further into the future. The physical world remains excluded from next year's superhuman commitments.

Source: Public Information

ABAB AI Insight

Musk sets the deadline for the end of next year, with 2028 as a buffer, continuing the Davos narrative of "this year or next year". The key limitation is "not shaping atoms": first announcing that digital labor can be replaced, leaving robots for the second phase. The Stockfish analogy targets software engineering pricing, not general chat. The simultaneous appearance of the 10 gigawatt and 15 gigawatt gaps indicates that his capability timeline is faster than the grid's timeline, with bottlenecks framed as transformers and cooling machines, rather than aligning research.

The capital path is to turn predictions into production line orders. SpaceX writes its company value as a proportion of AI, while Tesla positions Optimus as the hand of digital intelligence. The ten-year projection of 1 billion robots shifts labor supply from population to factory rhythm. The 20% to 30% economic growth rate frames replacement as incremental, used to counter the Security Council's slowdown language this week. Mutual inspections of models serve as his proposed safety interface, but do not equate to accepting international licensing.

This aligns with OpenAI's common standards, Anthropic's slowdown, and Armstrong's opportunity cost of mortality. Musk stands on the side of computing power and robots, with a timeline shorter than laboratory safety groups and more specific than the White House's "Department of Justice oversight". The industry is in a period of expanding regulatory windows by competing for deadlines.

The structural change is technological replacement. Digital work is announced as superhuman by next year, while physical work remains on the robot's uphill climb. The mechanism is: first define which tasks do not count as creation, then discount the wages of these tasks to 2027. Those who accept this definition will stop hiring for digital positions early; those who do not will continue to increase personnel according to the ten-year scale of safety covenants.

ABAB News · Cognitive Law

  1. First announce digital can be replaced, then physical will become the next order

  2. The deadline carries "at the latest" to prompt capital to move first

  3. In the year when the grid cannot keep up, the capability timeline will still open at a price