Vals AI Swarm of 90 Claude Opus Agents Discovers Two Room-Temperature Magnetic Semiconductor Candidates

The Three-Question Framework

What Happened?

On October 5, 2026, computational chemistry laboratory Vals AI released a preprint detailing the discovery of two viable room-temperature ferromagnetic semiconductor crystal candidates using an autonomous swarm of 90 Claude Opus 5.5 agents. In an automated closed-loop pipeline running across 72 hours, the multi-agent swarm formulated crystal substitutions, generated VASP and Quantum ESPRESSO density functional theory (DFT) simulation scripts, parsed band structures, and evaluated magnetic exchange couplings (J_ij). The swarm screened over 420,000 crystal candidates from the Materials Project and ICSD, identifying an ordered double perovskite YBaMnFeO_{5+delta} (Tc approx 365 K) and a Prussian blue coordination framework KV[Cr(CN)6] (Tc approx 342 K) with stable direct bandgaps.

Why It Matters

Room-temperature ferromagnetic semiconductors are considered the 'holy grail' of solid-state spintronics. Conventional computing relies on electron charge, which dissipates significant power as heat; spintronics utilizes electron spin, allowing memory and logic to be merged on a single chip with zero standby leakage power. For over three decades, physicists struggled to find materials that simultaneously exhibit robust semiconducting transport and ferromagnetic ordering above 300 Kelvin. By identifying two structurally stable candidates with Curie temperatures well above room temperature, the autonomous swarm demonstrates how frontier reasoning models can break computational bottlenecks in materials discovery.

What Could Happen Next

Vals AI has open-sourced the POSCAR crystallographic files and DFT calculation workflows to materials science repositories and partnered with experimental solid-state chemistry groups at UC Berkeley and the Max Planck Institute for Solid State Research. Physical powder synthesis and thin-film pulsed laser deposition (PLD) trials are scheduled to begin in November 2026 to verify experimentally whether the predicted Curie temperatures and spin-polarized transport hold in laboratory samples.

Executive Key Takeaways

Autonomous 90-Agent Swarm: 90 Claude Opus instances orchestrated end-to-end DFT materials screening without human intervention.

420,000 Structures Evaluated: Automated high-throughput simulation pipeline filtered candidates by thermodynamic stability and band gap.

Two Room-Temperature Candidates: Discovered YBaMnFeO5 (Tc ~365 K) and KV[Cr(CN)6] (Tc ~342 K) with validated semiconducting bandgaps.

Spintronics Breakthrough: Solves a 30-year solid-state physics dilemma to enable non-volatile, zero-leakage spin-logic computation.

Experimental Synthesis Fast-Track: Crystal structures transferred to Berkeley and Max Planck labs for thin-film laboratory validation.

In a major validation of autonomous scientific agents, computational materials startup Vals AI revealed on October 5, 2026, that an orchestrated swarm of 90 Claude Opus 5.5 agents discovered two room-temperature ferromagnetic semiconductor candidates, breaking a multi-decade impasse in condensed matter physics.

For more than thirty years, solid-state physicists have sought materials that integrate semiconducting electronic band gaps with spontaneous ferromagnetic order at ambient temperatures. While silicon microelectronics relies on shuffling electron charge—generating significant resistive Joule heating that limits datacenter scaling—spintronic circuits manipulate the intrinsic spin state of electrons. A true room-temperature magnetic semiconductor would enable non-volatile logic gates that retain memory without power and switch states with minimal energy dissipation.

The Vals AI research initiative tasked 90 autonomous Claude Opus instances with navigating the combinatorial chemical space of 420,000 crystal lattices documented across the Inorganic Crystal Structure Database (ICSD) and the Materials Project. Operating in continuous asynchronous feedback loops, the agent swarm synthesized hypotheses, drafted crystallographic POSCAR geometries, authored Python wrappers for the VASP and Quantum ESPRESSO density functional theory (DFT) simulation suites, and analyzed spin-polarized band structures.

The swarm identified two distinct crystal frameworks that satisfy all rigorous theoretical conditions. The first is an oxygen-ordered double perovskite, YBaMnFeO_{5+delta}, which exhibits strong supersonic superexchange coupling between manganese and iron cations across corner-sharing oxygen octahedra, yielding a calculated Curie temperature (T_C) of 365 Kelvin (92°C) and a direct bandgap of 1.12 eV. The second candidate is a cyanide-bridged coordination polymer, KV[Cr(CN)6], displaying Prussian blue structural symmetry with a T_C of 342 Kelvin (69°C) and 100% spin polarization at the valence band edge.

To ensure mathematical and physical rigor, the agent swarm performed phonon dispersion calculations across the full Brillouin zone, verifying that neither structure exhibits imaginary vibrational modes that would indicate structural instability. Subsequent Monte Carlo simulations verified that magnetic ordering remains robust against thermal fluctuations up to the predicted Curie thresholds.

Vals AI has made all raw crystallographic structures, ab initio input decks, and agent trajectory logs freely available to the global scientific community. Academic consortia at UC Berkeley and the Max Planck Institute for Solid State Research have already initiated solid-state reaction and molecular beam epitaxy (MBE) synthesis protocols to grow thin-film specimens of both compounds, heralding an era where autonomous AI swarms directly resolve long-standing material science grand challenges.

The Three-Question Framework

What Happened?

Why It Matters

What Could Happen Next

Executive Key Takeaways

Primary Sources & Citations