Wall Street's Jane Street Commits $6 Billion to AI Cloud
Jane Street has committed roughly $6 billion to AI cloud capacity. Hudson River Trading has signed a multibillion-dollar deal of its own. The spending extends to in-house data center expansion and stakes in AI chip companies.
Wall Street’s high-speed trading firms are pouring money into securing AI computing power. Jane Street and Hudson River Trading have signed AI cloud deals worth billions of dollars. They are also expanding their own data centers and investing in chip startups. The goal is to analyze markets and run in-house research on AI they can put to work in trading.
How big are these cloud deals
In April 2026, Jane Street committed roughly $6 billion to using CoreWeave’s AI cloud. Separately, it invested $1 billion in CoreWeave stock. The plan is to broaden an existing partnership and use computing resources across multiple facilities, along with Nvidia’s Vera Rubin technology, for research and training of its own AI models.
Hudson River Trading signed a multiyear agreement with CoreWeave in August 2026. Both companies described the deal as worth billions of dollars. Hudson River Trading said it would use Nvidia Vera Rubin NVL72 and HGX B200 systems for AI research and model training.
From six computers to 4,032 GPUs
Jane Street’s in-house research infrastructure has also grown beyond recognition. The firm described its earlier research setup as six Dell computers sitting on an office floor. One Texas data center it recently unveiled houses 4,032 GPUs for AI computing across 56 racks. It also features a fiber-optic network roughly 8,000 km long connecting the equipment, plus liquid cooling systems.
XTX Markets said it is building a large data center in Finland. According to the company, its current research facilities hold more than 25,000 GPUs. Available storage, it said, exceeds 1 exabyte. In other words, firms are scaling up not just computing hardware but also the space to store data.
Direct investments in AI chip companies
The money is flowing into AI chip startups as well. Etched announced it had raised an additional $700 million in a round led by Jane Street. It also said that in August 2026 it delivered its first server rack, used for AI inference, to Jane Street. On top of buying cloud capacity and operating its own data centers, Jane Street has become both an investor in and a hardware customer of a chip company.
Etched says it designed its chips to handle more work while responding faster. The approach involves running computing circuits at lower voltage and combining different types of memory. Beyond simply adding resources to train models, money is now going toward speeding up how quickly trained AI produces answers.
AI investment in finance is widening from model research into data centers, cloud, and chip development. Two things are worth watching from here: what these assembled computing resources actually deliver in research and trading, and whether the speed and throughput Etched touts show up in real-world deployments.