You are no longer the smartest type of thing on Earth
Noah Smith
· Noahpinion
· February 13, 2026 at 09:15
· ⏱ 5 min read
| Read on Substack ↗
Summary
=== SUMMARY ===
•The core thesis is that Artificial Intelligence is on the cusp of, or has already surpassed, human-level intelligence in most functional domains. This represents a paradigm shift unprecedented in human history.
•The pace of AI improvement is accelerating dramatically, driven by a massive, exponential increase in compute investment and self-improving AI algorithms. This will lead to profound economic disruption, commoditizing previously complex knowledge work like software engineering.
Summary
Noah Smith argues that AI has already surpassed human intelligence in key domains and will soon do so in all cognitive tasks, driven by a massive and accelerating investment in compute. This means the fundamental dynamic of human-AI power is flipping—humans are no longer the smartest beings—which for markets implies sustained demand for AI infrastructure and compute providers.
•AI can now get gold medals on the International Math Olympiad and solve unsolved math problems, yet most humans cannot.
•The METR curve shows AI agents are rapidly extending the length of time they can work on software engineering tasks, enabling 'vibe coding'.
•Spotify's co-CEO revealed that the company's best developers no longer write code; AI is increasingly writing software end-to-end.
•Global compute investment this year and beyond will 'utterly dwarf' all prior spending, as illustrated by a Bloomberg chart.
•AI's remaining weaknesses—lack of long-term memory and inability to learn on the fly—are expected to be solved soon.
•Massive investment in robotics will give AI more direct control of the physical world, accelerating the trend.
The article argues that AI progress scales with compute and that global compute investment is set to explode (citing a Bloomberg chart of enormous future spending). NVIDIA is the dominant supplier of
The article argues that AI progress scales with compute and that global compute investment is set to explode (citing a Bloomberg chart of enormous future spending). NVIDIA is the dominant supplier of the GPUs that power this compute ramp, making it a direct beneficiary even though not named explicitly.
Risk: Geopolitical export restrictions or a shift in AI architecture (e.g., custom ASICs) could reduce the GPU-centric demand thesis.