Google Cloud Taps SandboxAQ’s Physics-Based AI to Reshape R&D

PALO ALTO, Calif. – June 29, 2026 – In a significant move to bridge the gap between abstract artificial intelligence and tangible scientific discovery, SandboxAQ announced today that its specialized AI models will become available on Google Cloud's Marketplace. The collaboration aims to place powerful, physics-grounded computational tools directly into the hands of researchers in sectors like biopharma and advanced materials, potentially accelerating innovation cycles that currently span years and cost billions.

Starting in the third quarter, the company will roll out two of its flagship Large Quantitative Models (LQMs). The first, AQCat, is designed for materials and catalyst discovery, followed by AQPotency, which targets the early stages of drug discovery. This integration represents a powerful growth signal, indicating a strategic push by both companies to embed sophisticated, science-aware AI into the mainstream enterprise toolkit, making it as accessible as a simple query in a conversational AI.

“Bringing our LQMs to Google Cloud's Marketplace will put the rigor of first-principles science directly into the hands of every researcher, in the tools they already use,” said Jack D. Hidary, CEO of SandboxAQ. He described the partnership as a “powerful combination” that pairs the reasoning capabilities of a frontier model like Google’s Gemini with the quantitative precision of his company’s specialized models.

The Dawn of the Quantitative Model

While Large Language Models (LLMs) have captured the public imagination with their ability to process and generate human-like text, their application in hard sciences has been hampered by a critical limitation: they don't inherently understand the laws of physics. This can lead to nonsensical or physically impossible suggestions, a problem known as 'hallucination.'

SandboxAQ’s Large Quantitative Models represent a different breed of AI. These models are not just trained on vast datasets but are explicitly “physics-grounded,” meaning they are built upon a foundation of scientific equations and real-world laboratory data. This distinction is crucial. Instead of merely identifying statistical correlations, LQMs are designed to simulate outcomes that adhere to the fundamental principles of chemistry and physics, providing a level of accuracy and reliability that scientific R&D demands.

This approach, which blends AI with quantum techniques, allows the models to tackle computationally-intensive problems that have long been a bottleneck in scientific research. For instance, accurately calculating how a molecule will bind to a surface or a protein is a complex quantum-mechanical problem. By leveraging quantum-inspired algorithms, LQMs can deliver what SandboxAQ calls “gold-standard accuracy” at a fraction of the time and cost of traditional methods, which often involve either expensive supercomputing clusters or slow, iterative lab experiments.

This move also builds on SandboxAQ's strategy of platform agnosticism and accessibility, following a similar recent integration with Anthropic's Claude. The goal is clear: to make high-fidelity scientific simulation an integrated function within the AI-powered workflows researchers are already adopting.

Accelerating the Economics of Innovation

The commercial implications of this partnership are immense, targeting two of the most research-intensive sectors of the global economy. The first model to launch, AQCat, addresses materials and catalyst discovery. Catalysts are unsung heroes of the industrial world, underpinning over 90% of all commercially produced chemical products and playing a vital role in everything from fertilizer production to the development of sustainable aviation fuels.

AQCat focuses on calculating adsorption energy—a key metric that determines how strongly molecules bind to a catalyst's surface. By rapidly screening thousands of potential candidates computationally, it allows researchers to bypass costly and time-consuming physical tests for all but the most promising options. This could fundamentally alter the economics of creating next-generation materials for green hydrogen, plastics recycling, and more.

Following closely behind is AQPotency, a tool aimed squarely at the formidable challenges of drug discovery. The global drug discovery market, valued at over $112 billion in 2025, is projected to swell to nearly $187 billion by 2034. Yet, the process is notoriously inefficient, with development costs for a single new drug often exceeding $2 billion. A major hurdle is identifying 'binders'—molecules that attach tightly to a disease-related protein. AQPotency will allow researchers to computationally screen vast libraries of compounds to find the most promising binders, dramatically accelerating the foundational first step of designing a safe and effective drug.

While the market for AI-driven discovery includes formidable players like Schrödinger and Insilico Medicine, SandboxAQ is differentiating itself through this combination of physics-grounded precision and seamless cloud accessibility.

Google Cloud’s Strategic Play for the Scientific Enterprise

For Google Cloud, this collaboration is far more than a routine addition to its Marketplace. It is a clear strategic signal of its intent to dominate the high-value scientific and engineering verticals. By integrating SandboxAQ’s specialized models, Google Cloud enriches its ecosystem, moving beyond general-purpose AI to offer industry-specific solutions that solve critical business problems.

“Bringing SandboxAQ's Large Quantitative Models to GCP Marketplace is one of the ways we are empowering healthcare researchers to accelerate drug discovery and solve one of the most critical gaps in healthcare today,” commented Brian Goldstein, Vice President of Strategic AI and ISV at Google Cloud. His statement underscores a targeted strategy to bolster Google’s capabilities in life sciences, a sector where massive datasets and complex computational needs make it a natural fit for cloud services.

This move enhances the value proposition of the entire Google Cloud ecosystem. A pharmaceutical company already using Google Cloud for data storage and its Gemini AI models for research analysis can now, within the same environment, access a state-of-the-art simulation engine for molecular modeling. This creates a powerful, integrated R&D platform that is difficult for competitors to replicate. It transforms the cloud from a utility for computing power into an end-to-end workbench for scientific discovery.

By embracing a company operating at the intersection of AI and quantum, Google is also signaling its commitment to next-generation computational paradigms. It positions Google Cloud not just as a provider for today's needs, but as a partner for the future of research. By embedding the rigor of physical science into the accessible architecture of the cloud, this partnership signals a pivotal shift where the pace of digital innovation begins to set the pace for discovery in the physical world.