The rate of technological progress is not linear but exponential, driven by positive feedback loops where outputs of an innovation cycle become inputs for the next
In his seminal 2001 treatise, Ray Kurzweil formulated the law of accelerating returns to describe the non-linear trajectory of evolutionary and technological systems. The fundamental principle states that an evolutionary process creates more effective capabilities, which are in turn used to facilitate the next stage of evolutionary progress.
Whereas a linear projection assumes an invariant velocity of progress, an exponential model recognizes that velocity itself is an accelerating function of time. Over the course of the twenty-first century, progress will not equal one hundred years at twentieth-century rates; it will compress roughly twenty thousand years of advancement into a single century.
The exponential progression of computing did not begin with the integrated circuit. Moore's law was the fifth paradigm of computing, preceded by electromechanical calculators, relay-based systems, vacuum tube mainframes, and discrete transistors. Each time a dominant physical paradigm approached its thermal or miniaturization limit, an entirely new structural paradigm emerged to maintain the overarching exponential curve.
Today, conventional silicon lithography faces sub-nanometer quantum tunneling barriers and severe thermal dissipation limits. The transition to the sixth paradigm is already underway through heterogeneous chiplet packaging, 3D stacking, optical interconnects, and mathematically grounded deterministic computing architectures.
The primary consequence of exponential compounding is extreme economic bifurcation. Nations and enterprises that base their industrial strategies on linear compounding find themselves unable to absorb or compete with exponential leaps.
As computational capability compounds, the intellectual property governing autonomous decision-making and physical automation captures an overwhelming proportion of global economic surplus. Sovereign states that fail to participate in foundational architectural invention become permanently dependent on external technology providers.
| Epoch and technology paradigm | Period of dominance | Typical calculations per second per 1,000 dollars | Limiting physical bottleneck | Successor paradigm |
|---|---|---|---|---|
| Electromechanical punch-card | 1890 to 1940 | 0.00001 | Mechanical wear and relay inertia | Vacuum tubes |
| Vacuum tube computing | 1940 to 1958 | 0.1 | Filament burnout and heat dissipation | Discrete transistors |
| Discrete transistors | 1958 to 1965 | 100 | Interconnect capacitance and packaging | Integrated circuits |
| Planar silicon monolithic ICs | 1965 to 2015 | 10,000,000 | Gate dielectric leakage, thermal throttling | 3D FinFET and nanosheets |
| Heterogeneous and math-driven | 2015 to 2030+ | 10,000,000,000+ | Power delivery, interconnect latency | Optical and quantum interconnects |