Most discussions of advanced intelligence skip past physics. Speculative cosmologies (Dyson 1979, Tipler 1994) imagined infinite-time scenarios or singularities that current observation rules out. Discussions of AI capability often treat intelligence as a logical phenomenon abstracted from its physical substrate. The SAIERA Hypothesis fills the gap. Cognition is computation. Computation is physical. Physics has limits. Therefore intelligence has limits, and those limits leave fingerprints.
The framework defines intelligence operationally: a system's ability to reduce local entropy, maintain internal order, and convert available energy into structured outcomes more efficiently than its surroundings. No assumptions about consciousness, agency, or purpose. Just energy flow and information processing. That makes intelligence a continuum with biology, chemistry, and the broader behavior of dissipative structures, rather than an exceptional phenomenon.
Erasing one bit of information has a minimum energy cost (kT ln 2). Any irreversible computation generates entropy that must leave the system as heat. At advanced scales this dictates how much waste heat any computational substrate must dissipate, and the spectral signature of that dissipation when it is done efficiently.
There is a maximum amount of information storable in a finite region with finite energy, set by the region's energy times its radius. No system can pack unlimited information into bounded space. Any sufficiently advanced intelligence has to either approach black-hole-density energy concentrations or distribute itself over very large volumes.
The observable universe has performed at most about 10^120 operations since the Big Bang. That is an absolute computational ceiling for any process within the cosmos. Approaching it forces selectivity: any advanced intelligence must concentrate its computation on high-value targets rather than try to control everything everywhere.
Together these three constraints define a bounded optimization problem. They tell you what computation costs in energy, how dense it can be in space, and what its lifetime ceiling is. From those constraints, the predictions follow.
If a sufficiently advanced intelligence exists and operates near these limits, it would produce four distinct, simultaneous, coherent signatures across the electromagnetic spectrum:
Any single signature can be mimicked by natural processes. Dust extinction produces some infrared excess. Plasmas produce narrow emission lines. Magnetic confinement produces structured regions. But no known natural process produces all four signatures simultaneously, in the same location, with stable coherent properties. The hypothesis hinges on this multi-variable test. A coherent four-signature pattern would be strong evidence for thermodynamic optimization at scale. Anything less leaves the question open.
SAIERA does not claim that advanced intelligence is inevitable, that it must dominate cosmic evolution, that it can violate physical limits, or that its signatures can be hidden. It also makes no claims about purpose, consciousness, or teleology. The hypothesis says only this. If a system exists that approaches the physical limits of energy use, information density, and entropy regulation, it produces predictable signatures, and those signatures can be tested for with current technology.
The hypothesis fails under any of four conditions. No coherent signatures detected in adequately sensitive surveys. All candidate anomalies explained by natural astrophysics without fine-tuning. Cosmological parameters shifting in ways that invalidate the entropy baselines used. Future physics establishing different fundamental limits than the three the framework relies on. Non-detection at current sensitivity does not refute the hypothesis. It constrains its strong form, that detectable optimization is widespread.
SAIERA sits at the intersection of three fields. It contributes to astrobiology by grounding technosignature predictions in physics rather than assumed technological trajectories. It complements AI risk research by giving thermodynamic shape to questions of intelligence ceilings and recursive self-improvement. It extends prior cosmological work by remaining consistent with current observations of accelerating expansion and rejecting unverifiable infinities.
Founder of Cinderpoint Systems LLC. M.S. Artificial Intelligence (MSAI), M.S. Management (MSM). Researches how systems fail under speed, opacity, and scale.