Initiative Builds on Over 15 Years of Collaboration Between Jump and NVIDIA
Jump Trading (“Jump”) today announced it will be one of the first financial services enterprise customers to adopt NVIDIA Vera Rubin NVL72, a next-generation, rack-scale AI supercomputer platform. The deployment expands Jump’s long-standing accelerated computing platform and reinforces the firm’s strategy of building purpose-designed infrastructure to power AI and deep learning research in global financial markets.
As AI research becomes more computationally demanding and experimentation grows more complex, infrastructure decisions increasingly determine research velocity. For Jump, the ability to test, refine, and scale models quickly across constantly evolving market conditions is not optional. It is foundational to performance.
Financial markets generate some of the most complex, dynamic, and data‑intensive environments in the world. Jump’s researchers operate at the intersection of extreme data volumes, rapidly changing conditions, and highly diverse modeling approaches. The NVIDIA Vera Rubin NVL72 platform equips teams with supercomputer‑class compute density, memory bandwidth, and I/O throughput designed to support next-generation AI by increasing performance per watt by 10X.
The deployment builds on more than 15 years of collaboration between Jump and NVIDIA. Over that time, Jump has designed systems around NVIDIA CUDA to enable flexibility, programmability, and sustained research exploration.
“While popular Large Language Model Inference has a large place at Jump Trading, our research spans far beyond mainstream AI applications,” said Joe Stam, Head of Research Technology. “Financial modeling utilizes a wide range of conventional and highly novel model architectures of all sizes for thousands of constantly changing market applications. For more than a decade, NVIDIA CUDA-X has provided the tools and programmability that allow us to explore ideas that would be extremely difficult to replicate on any other platform.”
The Vera Rubin NVL72 rack-scale system provides increased compute density, improved memory bandwidth, enhanced reliability, and rack-scale energy efficiency designed for highly data-intensive workloads. While precise benchmarks will depend on workload, Jump anticipates meaningful performance gains compared to prior systems, enabling researchers to experiment more rapidly and iterate across models at greater scale. In environments where power and throughput are persistent constraints, incremental improvements in efficiency translate directly into research velocity.
“Infrastructure is not an afterthought, it is a strategic capability,” said Alex Davies, Chief Technology Officer at Jump Trading. “We design our AI trading factory as an enduring platform rather than as a series of upgrades. Each generation of accelerated computing allows us to refine how we design for scale, power efficiency, and reliability. The NVIDIA Vera Rubin NVL72 system strengthens that platform and reinforces our commitment to disciplined, high-performance engineering.”
“Not all AI workloads look the same, and financial research is a powerful example of where flexibility and programmability matter most,” said Ioana Boier, Global Head of Capital Markets Strategy at NVIDIA. “Jump Trading’s work spans a wide range of models and constantly evolving use cases, and systems like NVIDIA Vera Rubin NVL72 are built to support that kind of exploration. This deployment underscores how NVIDIA’s full stack platform enables customers to move beyond single‑model inference and build AI trading factories designed for problems that are still being defined.”
Jump’s research model is built on speed, adaptability, and rigorous experimentation. The firm is structured to test diverse AI approaches, respond quickly as market conditions shift, and scale strategies across global markets. That focus on disciplined infrastructure investment continues to shape how Jump builds for the long term.
“We are a deep learning research company in the business of predicting the future”, added Stam. “As the future remains yet unknown, Vera Rubin NVL72 is the ultimate investment that will give us the power and flexibility to tackle the problems we have yet to discover with the solutions we plan to invent.”
Long before the recent surge in generative AI, NVIDIA accelerated computing powered advanced AI deployments inside complex financial environments where reliability and performance are critical. Jump’s expanded AI trading factory continues that trajectory, reinforcing its position at the forefront of AI-driven research in global markets.
These statements are for informational purposes only and do not constitute an offer to sell, a solicitation of an offer to buy, or a recommendation of any security, investment product, or investment strategy, nor do they constitute investment, legal, tax, or other professional advice. The views expressed in these statements represent the opinions of the authors and Jump Trading. Forward-looking statements involve risks and uncertainties, and references to specific companies and their capabilities do not constitute investment recommendations or guarantee future performance.
