By James Aspinwall
Most people in the AI space talk about intelligence. Alexander Wissner-Gross wrote a physics equation for it. His 2013 paper proposed that intelligence is a fundamental force in the universe – a system’s tendency to maximize its future freedom of action. The equation has over 400 citations, a TED talk with 2.2 million views, and implications that make most AI researchers uncomfortable. If he’s right, artificial general intelligence isn’t an engineering challenge. It’s a thermodynamic inevitability.
The Resume That Doesn’t Quit
Wissner-Gross was born in 1982 in Great Neck, New York. Before entering college, he’d already won the USA Computer Olympiad, represented the US at the International Olympiad in Informatics, placed 10th nationally in the Intel Science Talent Search, and was inducted into the National Gallery for America’s Young Inventors Hall of Fame.
At MIT, he became the last person in the university’s history to earn a triple major – Physics, Electrical Science and Engineering, and Mathematics – before the option was discontinued. He graduated first in his class from the School of Engineering. He won the Goldwater Scholarship, the Marshall Scholarship, and the Hertz Fellowship, among others.
His PhD in Physics from Harvard, supervised by Efthimios Kaxiras, covered neuromorphic computing, programmable matter, and machine learning. It won the Hertz Doctoral Thesis Prize. He then held a Ziff Fellowship in Computer Science at Harvard.
In total: 128 major academic distinctions, 26 patents, 24+ publications, and media coverage in over 200 outlets. The man collects credentials like most people collect browser tabs.
F = T nabla S: The Equation for Intelligence
In 2013, Wissner-Gross and mathematician Cameron Freer published “Causal Entropic Forces” in Physical Review Letters. The paper’s central claim: intelligent behavior can be understood as a physical system’s tendency to maximize the diversity of its accessible future states.
The equation: F = T nabla S_tau
Where F is the force acting to maximize future freedom of action, T is the temperature (the strength of the drive), and S represents the diversity of probable accessible futures over a time horizon tau.
In plain language: intelligence is the force that keeps your options open.
To test this, they built a software engine called ENTROPICA. Without any explicit goal programming – no reward functions, no training data, no objectives – ENTROPICA independently reproduced:
- Tool use (a large disk using a smaller disk to retrieve a trapped object)
- Social cooperation (two disks coordinating their motions)
- Balancing an inverted pendulum (a classic control problem)
- Playing human games
- Trading stocks profitably
- Passing multiple animal intelligence tests
All from a single principle: maximize future freedom of action.
The implication is profound and unsettling. If intelligence is a fundamental physical phenomenon – like gravity or electromagnetism – then AGI isn’t something we need to cleverly engineer. It’s something the universe tends toward naturally, given sufficient computational substrate.
Relativistic Statistical Arbitrage
Before the intelligence equation, Wissner-Gross published another paper that sounds like science fiction: “Relativistic Statistical Arbitrage” (Physical Review E, 2010).
The core insight: the speed of light creates exploitable financial opportunities. Price information traveling between geographically separated exchanges is already stale upon arrival (~40 milliseconds between distant exchanges). There exist mathematically optimal geographic locations from which to coordinate arbitrage of securities traded on these exchanges. Trading along chains of intermediate optimal locations can effectively slow or stop the relativistic propagation of tradable information.
In other words, he proved that the laws of physics justify building trading nodes at specific points on Earth’s surface. This is the theoretical foundation for high-frequency trading’s obsession with geographic placement.
Datasets Over Algorithms
In 2016, writing for Edge.org, Wissner-Gross published an influential essay arguing that datasets – not algorithms – are the key limiting factor for AI advancement.
His evidence: reviewing 30 years of major AI breakthroughs, the average time between an algorithm’s publication and its corresponding AI advance was 18 years. The average time between dataset availability and the corresponding advance was less than 3 years – six times faster.
Examples: Watson’s 2011 Jeopardy victory used a 20-year-old algorithm. GoogLeNet’s 2014 object classification used a 25-year-old convolutional neural network design. The algorithms were “latent” in the literature, waiting to be activated by large, high-quality datasets.
This thesis arrived years before the scaling laws debate became mainstream, and it remains one of the most data-backed arguments for why compute and data matter more than algorithmic novelty.
Companies and Ventures
Wissner-Gross’s entrepreneurial career runs through a consistent collaborator: Timothy M. Sullivan.
CO2Stats (2007-2016): Y Combinator Summer 2008 batch. A web analytics tool that measured carbon emissions generated by website traffic, allowing site owners to purchase offsets. Acquired in 2016.
Gemedy, Inc. (2011-present): AI-based software for enterprises and government agencies, focused on cybersecurity, mission management, threat intelligence, and automated response. In 2019, Gemedy announced a strategic alliance with Two Six Labs for cybersecurity technology development. Named an SBA Emerging Leader.
Reified, LLC (2019-present): A startup investing and advising firm involved in more than 40 companies across AI, biotechnology, cybersecurity, post-quantum cryptography, whole brain emulation, and generative entertainment. Reified invests through vehicles including Link Ventures (Dave Blundin’s fund) and 021T Capital.
He also serves on advisory boards for the Brain Preservation Foundation, the Organ Preservation Alliance, Sen-Jam Pharmaceutical, and several AI startups.
The Moonshots Podcast
Wissner-Gross joined Peter Diamandis’s Moonshots podcast in 2025 as a regular co-panelist alongside Dave Blundin and Salim Ismail. He functions as the panel’s resident physicist and AI theorist, bringing the kind of first-principles reasoning that comes from someone who literally wrote an equation for intelligence.
Notable episodes include discussions on America’s plan to win the global AI race, GPT-5 implications, whether AI is a bubble, Elon Musk’s AI strategy, the 40x deflation thesis, and the “Solve Everything” paper he co-authored with Diamandis.
Solve Everything: The 2035 Roadmap
Co-authored with Diamandis and available at solveeverything.org, this paper outlines a roadmap for directing superintelligence at humanity’s biggest problems. It introduces two frameworks:
The Industrial Intelligence Stack – a nine-layer model for converting real-world domains into computable problems.
The Abundance Flywheel – commit compute, solve the problem, capture surplus, reinvest.
The roadmap spans three phases: conquering pure information (2026-2027), mastering the physical world (2028-2031), and planetary-scale systems (2032-2035). It includes fifteen specific moonshots grouped into human needs, the frontier of mind, the planetary substrate, and the frontier of physics.
It’s ambitious to the point of audacity, which is precisely what you’d expect from someone who thinks intelligence is a thermodynamic force.
The Innermost Loop
In December 2025, Wissner-Gross launched “The Innermost Loop,” a daily Substack newsletter and podcast tracking rapid AI developments. The tagline: “High-velocity intelligence from the event horizon.” He frames the technological singularity not as a future event but as an ongoing process already underway, with the “Token” (artificial cognition) collapsing toward the cost of electricity and dragging the entire economy with it.
What Sets Him Apart
In the Diamandis orbit – which tends toward enthusiasm and marketing – Wissner-Gross is the rare participant with genuine theoretical depth. He doesn’t just predict AI will transform the world. He published a peer-reviewed physics paper explaining why it must, based on the same thermodynamic principles that govern the rest of the universe.
His equation might be wrong. Intelligence might not reduce to entropy maximization. But the framework is testable, the predictions are specific, and the intellectual ambition is rare in a field dominated by benchmarks and funding announcements.
Triple major. Equation for intelligence. Speed-of-light trading theory. Datasets-over-algorithms thesis. 40+ company portfolio. And he’s 43.
The man operates at a different clock speed.