The quantitative AI startup SandboxAQ has completed its Series E funding round, with investors including Bridgewater Associates, NVIDIA, and Google.

The quantitative AI startup SandboxAQ has completed its Series E funding round, raising a total of $450 million. The funding comes from Ray Dalio, founder of Bridgewater Associates, Horizon Kinetics, BNP Paribas, Google, and Nvidia. The new investment will help SandboxAQ continue to advance the application scenarios of large quantitative models (LQM) in the intersection of AI and quantitative technology, accelerate innovation in the field of artificial intelligence, and expand its research in finance, biopharmaceuticals, and cybersecurity.

SandboxAQ has raised over $950 million since it spun out from Alphabet in 2022. New investors supporting the development of SandboxAQ include Breyer Capital, Eric Schmidt, Alger, Paladin Capital, S32, TIME Ventures, and T. Rowe Price Associates, Inc. The new funding enhances SandboxAQ's leadership in the fields of artificial intelligence and quantitative technology, driving innovation in the biotech and financial industries.

Olivier Osty, head of Global Markets at BNP Paribas, stated that artificial intelligence and advanced computing power are having a profound impact on financial services. BNP Paribas looks forward to collaborating with SandboxAQ to explore innovative solutions in the fields of artificial intelligence and quantitative technology.

Ray Dalio, the founder of Bridgewater Associates, stated that he is betting on the SandboxAQ team and its large quantitative model (LQM) because he is impressed with the team and the technology.

What are LQMs?

Large Quantitative Models (, abbreviated as LQM), represent the next stage in the development of artificial intelligence. LQM is an AI based on physical, chemical, and biological foundational equations, trained using data generated directly from these equations using validated mathematical methods.

The structure of LQM typically includes several key components. These are special modules for solving equations or conducting statistical analysis through deep learning, traditional numerical methods, or methods that lie between the two, which distinguishes LQM from traditional LLM.

LQM incorporates a curated scientific information database as part of its reasoning process. LQM utilizes LLM to complete language-centric tasks, enabling it to understand and generate text in a manner similar to humans, thus effectively interacting with users and processing text-based data.

How do these components work together? When a problem arises, LQM may first use its language model to understand the issue, after which it can initiate its quantitative module to perform the necessary calculations or simulations. Throughout the process, it can leverage its domain-specific knowledge to guide its reasoning. Finally, it can synthesize all this information to generate a response, which it then communicates to the user through the language model.

SandboxAQ develops a new algorithm for tensor networks.

Quantum computing can certainly directly simulate these systems using the language of quantum mechanics, but its actual realization may still take several decades. This is because current quantum computing still faces a series of challenges, such as controlling the error rate of qubits. Although Google's recently released Willow chip has made significant breakthroughs on this issue, there are still many problems to be solved before large-scale quantum computers can be constructed.

To address these issues, SandboxAQ has developed a new algorithm based on Tensor Networks. This algorithm originally stems from the field of quantum many-body physics and utilizes a fundamental property of nature—locality. In simple terms, locality means that parts of a system that are far apart, such as two distant atoms in a long molecule, do not meaningfully influence each other. By leveraging this property, the tensor network algorithm can efficiently represent quantum states, which is referred to as the "entanglement area law." ( Source: MIT Technology Review )

SandboxAQ and Nvidia CUDA technology collaboration

SandboxAQ has established a deep technical partnership with Nvidia, extending CUDA capabilities so that ordinary GPUs can support quantum computing. This allows them to perform quantum simulations on existing hardware without having to wait for true quantum computers to become available, while also enabling future integration of Quantum Processing Units (QPUs). In a study, SandboxAQ's research team used Google's Tensor Processing Units (TPUs) to complete a complex high-dimensional optimization involving over 600 billion parameters within 24 hours, setting the world record for the largest scale of tensor network computation.

Application Scenarios of LQM

LQM can help scientists analyze complex datasets, formulate hypotheses, and even design experiments, especially useful in the field of biology, where it can predict the three-dimensional structure based on the amino acid sequence of proteins. A specific example is in the development of new drugs. By analyzing molecular structures and predicting interactions, LQM can significantly accelerate the process of identifying potential drugs. In fields such as materials science or structural engineering, LQM can help optimize designs and propose improvements based on specific parameters by running countless simulations.

LQM can also be used to process various data sources to create more accurate climate models, helping us better understand and predict environmental changes. In the financial sector, LQM can handle market data, news, and economic indicators to provide more complex risk assessments and investment strategies.

Introduction to SandboxAQ

SandboxAQ is a B2B company that provides solutions in the fields of artificial intelligence and quantitative modeling. SandboxAQ's large quantitative model (LQM) has made significant advancements in life sciences, financial services, navigation, and other scientific domains. SandboxAQ was spun off from Alphabet Inc. and is an independent company funded by investors and strategic partners, including T. Rowe Price Associates, Inc., Alger, IQT, US Innovative Technology Fund, S32, Paladin Capital, BNP Paribas, Eric Scanmidt, Brehebmidt, Brehebun, Paladin Capital, BNP Paribas, Eric Scanmidt, Breullibmidt, Breulli, and other institutions.

This article reports that the quantitative AI startup SandboxAQ has completed its Series E funding round, with investors including Bridgewater Associates, Nvidia, and Google. It first appeared in Chain News ABMedia.