PSI Details Data Center Revenue Behind Its AI Research Ambitions

TechCrunch reported on November 10, 2026, that Physical Superintelligence earns its current revenue by using its tools to optimize data centers, while pursuing a longer-term business built on AI-assisted physics discovery.

Chief executive Matt Pines described the objective as converting theoretical-physics breakthroughs into actual technologies. Data centers are the first proving ground for that approach, where PSI’s AI research systems, which it calls virtual physicists, can work on live physical systems with measurable operational constraints.

PSI is a public benefit corporation, with safety, verification, responsible development and broad public benefit established in its charter. On its company page, it reports more than 25 researchers and engineers and more than $58 million raised, led by Breakthrough Energy Ventures.

Data Centers as a Commercial Proving Ground

The current optimization work simulates heat, power, water and sound, connecting facility performance and community impact to a site’s physics. On its vision page, PSI says it builds high-fidelity models of live facilities so virtual physicists can examine the full system, uncover hidden constraints and test interventions, with increased computational density, more efficient energy use, improved thermal management and higher capital efficiency as intended results.

The commercial campaigns also supply research feedback. PSI says live facilities impose constraints that simulations would not propose, and each campaign trains the virtual physicists used in later ones. That gives the data-center business two functions: current revenue and a source of practical constraints for improving the research system.

A Research System Built Around Verification

Pines described an effective research system as a combination of software harnesses around models, tools, memory systems and simulation infrastructure, allowing it to pursue a problem at breadth and depth while considering a wide range of alternatives. PSI formalizes that process as an Autonomous Campaign, a continuous effort running from problem definition through hypothesis, verification, experiment, engineering and deployment. Each campaign has measurable milestones, hard verification gates and a path toward application, while human researchers choose missions and success criteria.

The architecture separates candidate generation from verification. Its checks include conservation laws, symmetry, dimensional consistency, formal proof, simulation, digital twins — virtual counterparts of physical systems — and physical measurement. PSI says the verifiers that certify a result can also become reward functions, or training signals, for later virtual physicists, tying evaluation directly to system improvement.

PSI’s November 6, 2026, research announcement provides an example. The company said its internally developed system, running a then-non-public Anthropic model, produced a complete proof of the Aaronson–Ambainis conjecture during the third day of a run; Lean, a proof assistant that checks every step, accepted it and later accepted a final version after two improvements. Open since 2008, the conjecture concerns when quantum computers can be exponentially faster than ordinary computers on random inputs. David Turturean led the work with early access to Claude Fable 5.2, which Anthropic publicly introduced on November 9, 2026.

Physical Validation Remains the Commercial Test

The proof also illustrates a limit to commercialization. Mathematics and computer science can express results in validated code, but physics and the life sciences still require experimental contact with nature. Pines said PSI is working on efficient ways to validate computational results through experiments, summarizing the distinction with the observation that “nature gets a vote in your models of reality.”

Pines also said theoretical physics is not yet mostly the domain of machines, despite making a stronger conditional pitch to investors six to nine months earlier. That qualification matters for PSI’s business: its present revenue depends on facility optimization, while its larger ambition requires validated discoveries that can become devices or other technologies.

Pines connected a validated fact of nature to the possibility of building a device with capabilities that were previously unavailable. For PSI, the commercial question is therefore whether its generation-and-verification process can repeatedly produce experimentally confirmed results, allowing the same research system to move from optimizing AI infrastructure to creating technologies based on new physics.