What Radical AI does

Radical AI is a New York-based private company founded in January 2025 that builds AI-driven self-driving laboratories for accelerated materials discovery. The company's core platform pairs a proprietary AI engine called Antimatter—which combines atomistic predictive modeling with generative approaches to screen billions of material compositions and downselect candidates—with a physical autonomous laboratory that executes AI-designed experiments from synthesis through characterization, reportedly producing and characterizing 25+ alloys per day with complete digital trails. Supporting components include the EGIP interatomic potential (top-5 on the Matbench Discovery leaderboard), TorchSim (a PyTorch-native simulation engine open-sourced on GitHub), and the MATRIX/MATRIX-PT and LitXBench benchmarks open-sourced on Hugging Face and GitHub respectively.

The company serves mission-critical industries—aerospace, defense, automotive, energy, semiconductors, and manufacturing—that require advanced metallic materials capable of withstanding extreme environments such as hypersonic flight and spacecraft reentry. Its two named disclosed customers are U.S. government agencies: the Department of the Air Force via an AFWERX Direct-to-Phase II SBIR contract valued at $1,197,902 for high-entropy alloy discovery, and the Department of Energy via a Memorandum of Understanding signed in December 2025 under the Genesis Mission covering clean energy, nuclear, and critical materials.

Radical AI's revenue model combines direct enterprise materials sales to industrial buyers with government contracts, pricing negotiated on a per-engagement basis with no public disclosure. Following a $55M seed round in July 2025 co-led by Alumni Ventures and RTX Ventures—with approximately $65M raised in total—the company is deploying capital into scaling its Alexandria Center for Life Science autonomous facility, expanding into adjacent verticals (fusion energy, advanced batteries, semiconductors), and investing strategically in adjacent AI ventures such as Ricursive Intelligence's January 2026 Series A.

Name

Radical AI

Legal name

Radical AI, Inc.

Website

https://radical-ai.com

Company type

Private

Founded year

2025

Operating status

Operating

Headcount range

11–50 employees

Short description

Radical AI is a New York-based private company that operates a self-driving laboratory combining AI-driven atomistic modeling with robotic experimentation to accelerate materials discovery, serving aerospace, defense, energy, and semiconductor industries with advanced alloys.

Ownership category

akta.pro rank

Product category

AI-Driven Materials Discovery

NAICS

Analytical Laboratory Instrument Manufacturing (334516)

SIC

Special Industry Machinery, Nec (3559)

akta.pro primary industry

Agents & Autonomous Workflows (Tool Use, Planning, Multi-Agent) (HDAAACAF)

akta.pro secondary industry

AI/ML Engineering & Model Development Services (BPAEAHAF)

Keywords

• AI materials discovery
• Autonomous laboratory systems
• High entropy alloys
• Atomistic simulation software
• Generative materials design

Headquarters

HQ city

New York

HQ country

United States

HQ region

North America

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Infrastructure, Operations, Marketing or Sales

Revenue model

1. Materials Sales: Selling advanced materials to critical industries including energy, aerospace, defense, semiconductors, and automotive that require advanced materials for their applications
2. Government Contracts: Direct-to-Phase II SBIR contract with U.S. Air Force for accelerated discovery of novel high entropy alloys with enhanced thermomechanical properties

Go-to-market motion2 records

Distribution channels1 record

Marketing channels6 records

Core offering

Radical AI develops and operates a self-driving laboratory system that combines an AI prediction engine (Antimatter) with autonomous robotic experimentation to discover and produce advanced metallic alloys for aerospace, defense, energy, semiconductor, and manufacturing customers. The company also releases supporting open-source scientific software (TorchSim, EGIP, MATRIX, LitXBench) and sells or licenses novel materials discovered through its closed-loop AI-and-lab workflow.

Product overview

Radical AI operates as an integrated platform combining AI software with an autonomous physical laboratory for accelerated materials discovery. The core system consists of Antimatter (AI prediction engine) and the Autonomous Lab (physical self-driving lab) that execute AI-designed experiments. The portfolio includes open-source tools: TorchSim (atomistic simulation engine), EGIP (interatomic potential), LitXBench/LitXAlloy (LLM extraction benchmarks), and MATRIX/MATRIX-PT (multimodal benchmarks). The closed-loop Materials Flywheel integrates computational predictions with robotic experimentation, targeting advanced alloys for aerospace, defense, energy, and semiconductor industries.

Differentiator

Problem solved

Functional benefit

Brands

• Antimatter: AI engine that combines atomistic predictive modeling with generative approaches to screen billions of material compositions and predict structures and physical properties.

Products and services

• Antimatter AI engine that combines atomistic predictive modeling with generative approaches to screen billions of material compositions and predict structures and physical properties, downselecting candidates from 10 billion to 100 for physical experimentation. Used by enterprise customers and government agencies seeking novel alloys.
• Autonomous Lab Self-driving laboratory that executes AI-designed experiments from chemical synthesis to characterization, enabling creation and characterization of 25+ alloys per day with complete digital trails. Sits at the Alexandria Center for Life Science in NYC.
• TorchSim Next-generation PyTorch-native atomistic simulation engine providing 100x speedup over ASE and 100,000,000x speedup over DFT, supporting NVE, NVT, NPT, gradient descent, and Frechet cell FIRE integrators.
• EGIP (Efficient Geometric Interatomic Potential) First-generation direct force prediction interatomic potential achieving top-five ranking on Matbench Discovery and 3.5x speed improvement over the leading eSEN model for fast and accurate atomistic simulations.
• LitXBench Benchmark framework for evaluating LLM performance on scientific literature extraction, enabling accurate extraction of experimental data from materials science papers with auditable, editable code representation.
• LitXAlloy Dense benchmark derived from 19 experimental alloy papers totaling 1,426 target values, hand-extracted with LLM-assisted scrutiny using approximately 1.1 billion Claude Code tokens.
• MATRIX Multimodal benchmark for scientific reasoning in materials science spanning four task families across SEM, XRD, EDS, and TGA modalities. Released on Hugging Face as open source.
• MATRIX-PT Post-training dataset accompanying the MATRIX benchmark for training vision-language models on materials science experimental data. Released on Hugging Face.

Quantifiable outcome

• Nearly 10x faster materials discovery compared to DARPA/GE MACH program baseline
• +3 more outcomes

Named customers2 records

Segments5 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability10 records

Feature9 records

Partnerships

Two partnerships are on record, tiered core.

• U.S. Department of EnergycoreStrategic or Co-development Partner · 20 December 2025Memorandum of Understanding to advance AI-driven materials discovery in support of the Genesis Mission. Collaboration across closed-loop autonomous research infrastructure, translational research on novel materials, next-generation AI capabilities for scientific discovery, and joint pilot projects. Addresses national security, energy independence, and long-term economic resilience through materials innovation.
• U.S. Air Force (AFWERX)coreStrategic or Co-development Partner · 25 August 2025Direct-to-Phase II SBIR contract for accelerated discovery of novel high entropy alloys (HEAs) with enhanced thermomechanical properties to address challenges in hypersonic flight. Traditional alloys cannot endure extreme conditions, creating urgent need for new advanced materials.

Scale indicators11 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• CuspAI: UK-based AI-for-materials startup building generative and physics-informed models to discover new materials. WEF explicitly groups CuspAI with Radical AI and Lila Sciences as the leading autonomous-lab cohort, making it the closest head-to-head competitor for federal and enterprise materials contracts.
• Lila Sciences: AI-native autonomous science company building self-driving laboratories for scientific discovery across materials and life sciences. Recognized alongside Radical AI by the World Economic Forum as a peer leader in AI-driven materials discovery with comparable platform ambitions.
• Citrine Informatics: AI platform for materials and chemicals R&D, using machine learning on materials data to accelerate formulation and process design. Directly comparable as a commercial AI-for-materials software platform targeting industrial customers in similar verticals to Radical AI.
• Kebotix: AI and robotics company focused on discovering and developing new materials via a closed-loop platform combining generative AI with automated experimentation. Closely comparable architecture to Radical AI's self-driving lab thesis, competing for similar customer segments.
• Orbital Materials: AI-driven materials discovery company building foundation models for materials science, including alloy and battery materials. Direct competitor in the AI-native materials model space, focused on industrial use cases that overlap with Radical AI's alloys and energy applications.

Broad incumbents

• Schrödinger: Public computational chemistry and materials simulation company with a widely deployed physics-based platform across pharma and materials. A broad incumbent whose established ML/atomistic tooling competes with Antimatter/TorchSim for industrial materials R&D budgets.
• Google DeepMind (GNoME / AI for Science): Google's AI-for-Science effort, including the GNoME project for materials discovery and inorganic crystal structure prediction. A well-funded incumbent whose foundation-model approach to materials overlaps with Radical AI's generative atomistic strategy.
• Microsoft AI for Science: Microsoft's initiative pairing foundation models with scientific discovery workflows, including materials and chemistry. Represents a hyperscaler incumbent approach that competes for the same AI-for-materials narrative and partnership opportunities with federal agencies.
• IBM Accelerated Discovery: IBM's AI-for-materials and accelerated discovery effort combining foundation models, quantum/HPC, and lab automation. A broad incumbent alternative for enterprise customers evaluating AI-native materials platforms versus Radical AI's offering.

Others

• Citrine Informatics (alt: Anthropic Claude for Science / Sandia NL): National-laboratory and federally funded AI-for-materials initiatives (e.g., Sandia's autonomous experimentation programs) — adjacent ecosystem players that often partner with or compete against commercial entrants like Radical AI for the same federal R&D funding streams.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights6 records

Customer concentration

• Xx.com/RadicalAI
• GitHubgithub.com/Radical-AI
• LinkedInlinkedin.com/company/radicalai

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 records

Funding overview

Funding rounds1 record

Investors10 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Radical AI