Hudson River picks Lambda for AI

Hudson River Trading has moved a visible piece of its AI stack to an outside provider. Lambda said on May 20 that the quantitative trading firm selected it to support quantitative research and development, with infrastructure aimed at model training and trading simulation. The announcement adds a named trading firm to Lambda’s customer list and gives a public marker for how one large quant shop is sourcing AI capacity. Reuters reported the deal as a cloud contract under which Lambda would supply access to Nvidia chips. ### What exactly did Hudson River Trading buy from Lambda? Lambda said it is providing Hudson River Trading, or HRT, with NVIDIA accelerated computing infrastructure, including HGX B200 systems, advanced networking, storage and orchestration. The companies said the setup is meant to support HRT’s quantitative research and development. DatacenterDynamics reported on May 22 that the workloads named in the agreement are model training and trading simulation. (lambda.ai) Reuters’ May 20 report described the arrangement as a cloud deal for access to Nvidia chips. ### Why does this stand out for a trading firm that already has big internal compute? Hudson River Trading says on its website that it has built “one of the world’s most advanced computing environments” for research, modeling and risk management. (lambda.ai) Its careers page also says the firm’s HPC network engineering team supports large GPU and CPU compute clusters across globally distributed on-premise data centers. (datacenterdynamics.com) Lambda framed the deal as an answer to HRT’s growing demand. In its May 20 post, Lambda said HRT turned to the company as on-premise infrastructure reached its ceiling. DatacenterDynamics separately reported that HRT needed rapid access to compute with high uptime. ### Who at HRT described the decision, and what did he say? Gerard Bernabeu Altayo, compute systems lead at HRT, was quoted in Lambda’s announcement saying the firm needed infrastructure that could “move as fast as our research.” He said Lambda “stood out for its technical depth and operational clarity.” (hudsonrivertrading.com) That wording matters because it ties the purchase to research operations rather than exchange-facing execution systems. (lambda.ai) Neither Lambda’s announcement nor Reuters’ report said the deal covered live trading gateways or latency-critical execution infrastructure. ### How much of HRT’s business already depends on AI-style research? HRT’s current recruiting shows AI and high-performance compute are already central to the firm’s buildout. (lambda.ai) The company is hiring AI Researchers focused on LLMs for its HAIL team, which it says develops powerful AI models used by trading teams to drive a significant fraction of trading. HRT is also advertising for GPU systems engineers, HPC network engineers and a hardware strategy and technology partnerships lead. (lambda.ai) In that latter posting, the firm says its research cluster includes GPU and CPU fleets “larger than many public clouds.” ### What does this say about where AI sits inside trading infrastructure? The public documents point to a split architecture. HRT’s website describes a firm built around automated trading and large internal research systems, while the Lambda deal is specifically attached to research, model training and simulation. (hudsonrivertrading.com) That leaves the execution side largely undescribed in public. What is public is the next step: Lambda’s infrastructure is now part of HRT’s research pipeline, and HRT’s open roles show continued hiring across AI research, storage, networking and trading systems as the firm expands that stack. (hudsonrivertrading.com 1) (hudsonrivertrading.com 2) (hudsonrivertrading.com 3)