About Building Alan
The chapters ask what physics allows. This part asks a different question: how close does today’s technology come, and what would close the gap? The answer changes every year, so these pages are living documents. Each one has a review date, and every number in it carries the date it refers to and its source.
Each living review looks at one technology or one bottleneck and measures it against the same yardsticks:
- The Landauer bound: the least heat that erasing one bit can release, \(k_B T \ln 2\) at the temperature the device runs at (Landauer 1961; Bérut et al. 2012).
- The Margolus-Levitin bound: the most operations per second that a given amount of energy allows (Margolus and Levitin 1998; Lloyd 2000).
- The Bekenstein bound: the most information that a given mass and size can hold (Bekenstein 1981).
- Energy at the wall plug: what the whole system draws from the grid, cooling and power delivery included. A device that runs near the Landauer bound at 4 K can still cost more, counted at the wall plug, than one at room temperature.
These pages are written for engineers and researchers who design processors, memory, AI hardware and data centers, and for any reader who wants to know where real machines stand. The numbers behind them are published as data (see the data folder), so they can be reused and checked.
The same method, applied beyond computing, is Escape Velocity: an open atlas of what each technology still needs to reach maturity, from fusion to medicine, and of which technologies wait on which. Its entries on quantum computers, data movement, heat removal and the energy of AI inference link back to these living reviews.
Living reviews
| Living review | Status | Last reviewed | Review due | Curators |
|---|---|---|---|---|
| CMOS: The Baseline | proposed | never | not reviewed yet | none yet |
| The Cost of Moving Data | proposed | never | not reviewed yet | none yet |
| Getting the Heat Out | proposed | never | not reviewed yet | none yet |
| Adiabatic and Reversible Logic | proposed | never | not reviewed yet | none yet |
| Superconducting Logic | proposed | never | not reviewed yet | none yet |
| Quantum Hardware | proposed | never | not reviewed yet | none yet |
| Thermodynamic and Probabilistic Computing | proposed | never | not reviewed yet | none yet |
| Neuromorphic Computing | proposed | never | not reviewed yet | none yet |
| Optical Computing | proposed | never | not reviewed yet | none yet |
| The Energy of a Token | proposed | never | not reviewed yet | none yet |