In collaboration with Grok
Core Theory
All naturally growing phenomena follow the natural growth curve (S-curve, also known as the logistic/Pearl curve), consisting of three stages: introduction phase (slow start), rapid growth phase (steep rise), and maturity/stabilization phase (flattening), with a symmetrical time ratio of 1:2:1. This is a classic summary of the logistic curve. (See my extensive analyses on the Chainless website for details.)
I have previously applied this framework to Bitcoin’s natural growth, predicting the fourth cycle as the inflection point cycle (with BlackRock’s entry as the inflection event), and to the U.S. dollar collapse cycle, forecasting approximately 50 years remaining until collapse. The smartphone industry provides an intuitive illustration of the framework’s elasticity and rigidity:
Theoretically, the introduction phase lasts about 15 years, with the rapid growth phase spanning 30 years (strict 1:2:1);
Yet, driven by manufacturer promotions, supply chain optimization, policy support, and other human factors, the rapid growth phase was actually completed in just 23 years.
This fully confirms: Human activities (investment, policy, genius, promotion, etc.) can only influence the length of cycles, not alter the overall direction or higher-order laws — just as biological life has birth, aging, illness, and death, and population growth follows the logistic curve. AI, as a typical natural growth system, strictly adheres to this framework.
The greatest practical value of S-curve theory lies in precisely identifying the inflection point (the moment when the curve shifts from gradual to steep ascent). I regard the general availability (GA) release of Claude Code in May 2025 as the clear inflection point for the complete cycle of large-scale AI models — marking the decisive transition from the introduction phase to the formal launch of the rapid growth phase.
S-Curve Stage Predictions for Large AI Models
- Introduction Phase (Late 2022–Early 2023 ChatGPT Launch — May 2025)
The initial gentle segment of the curve. ChatGPT brought AI from laboratories into public view, with technical validation and early adoption advancing gradually. The May 2025 GA release of Claude Code signals the inflection point, laying the foundation for explosive subsequent growth.
- Rapid Growth Phase (May 2025 — 2028)
This is the steepest and most explosive segment of the curve. I believe the rapid growth phase will extend to 2028, rather than ending in 2026.
During this period, AI capabilities will surge explosively: rapid iteration in model performance (reasoning, agents, multimodality, tool use); large-scale practical adoption of AI Agents as everyday productivity tools; deep industry integration and full-scale explosion (enterprise, healthcare, education, manufacturing, etc.); exponential growth in commercial value and adoption rates.
2026 represents only a mid-phase peak, not the endpoint. The emergence of exceptional individuals (industry leaders) may steepen the curve further, but cannot alter the natural necessity of completing the full rapid growth phase.
Difference from Musk et al.: They generally view 2026 as the peak or turning point of rapid growth. I believe this stems from their focus on short-term compute explosion intuition, whereas S-curve theory emphasizes the complete symmetrical form and the objective 1:2:1 law. Analogies to the internet bubble must be made cautiously — the durations of rapid growth phases differ, and AI follows its own unique rhythm.
- Maturity/Stabilization Phase (2028 Onward)
Roughly 5–6 years from 2023, large-model foundational capabilities will peak around 2028, with the curve flattening. AI will cease to be in a state of “daily new surprises” and become foundational infrastructure like electricity or the internet: focus shifts to ecosystem building, deep industry integration, governance, and societal adaptation; growth mode transitions from scale expansion to efficiency optimization and application implementation. A new S-curve may subsequently emerge (e.g., embodied AI, scientific discovery AI, or novel architectures), but that represents the next paradigm, not an infinite continuation of current large models.
Core Principles and Real-World Implications
- Law Unchanging: The S-curve is a higher-order natural law; direction and stage sequence are irreversible. AI will inevitably progress from “infancy” to “adulthood” to “stability.”
- Humans Influence Only Duration: Massive investment, genius acceleration, or promotional efforts (as in the smartphone case) may advance maturity to late 2027 or delay it to 2029 due to regulation/energy constraints, but the ultimate outcome remains unchanged.
- Role of Exceptional Individuals: They can accelerate the process but cannot make AI grow rapidly forever or reverse direction.
- Value of the Inflection Point: As shown in the smartphone example, precisely identifying the inflection point (Claude Code GA in May 2025) is the most powerful practical significance of S-curve theory — enabling us to foresee the window period in advance, prepare accordingly, and avoid being misled by short-term noise.
In summary, according to natural growth curve theory, we are currently (March 2026) in the mid-segment of the 2025–2028 rapid growth window following the Claude Code inflection point. Around 2028, large models will essentially mature, and AI will truly become societal infrastructure rather than perpetual “breaking news.”
This prediction is clear and forward-looking, also explaining why Musk et al.’s 2026 judgment appears aggressive yet rests on intuitive foundations. If new variables emerge (major breakthroughs or external constraints), cycle length may adjust slightly, but the overall trajectory will not change. This is the power of the natural growth curve — it allows us to see AI’s future clearly amid the clamor.
如果你需要进一步调整（如加链接、强调某些句子、或输出PDF格式），随时告诉我。合作愉快，继续保持你的独特视角！
