Jump Trading has hired the entire founding team of Itô, a startup built around trading access to computing power. The team will launch and run a new compute trading desk at the firm.
Itô co-founder Affaan announced the move in a social media post on October 12, 2026.
What Jump is actually buying
Itô’s specialty is trading access to computing power and AI inference capacity. Inference is the work an AI model does after it’s trained, answering prompts and generating output.
The new desk will start with mid-market contracts. These deals mostly run for 3 to 5 years.
Contract sizes in this segment typically span tens of millions to hundreds of millions of dollars, with some reaching the low hundreds of millions.
Early trading activity shows that approximately 30% of these mid-market contracts require upfront payments. Buyers are being asked to put real cash down before the compute is delivered.
Why the Itô team picked Jump
The Itô founders, including co-founder Alejandro, have hands-on experience commercializing computational resources. They previously built a platform aimed at helping buyers secure AI compute supply.
Their background also spans credit products and supercomputing.
According to the research summary of the announcement, the team chose Jump because of its strong credit standing, its risk management capabilities, and its existing infrastructure.
The team’s stated goal is to build market structures that can hedge risks that traditional AI supply chains currently leave unaddressed.
Jump’s long road from the pits to the GPUs
Jump Trading was founded in 1999 by Paul Gurinas and Bill DiSomma, both former pit traders at the CME. The firm has since grown into a prominent name in proprietary trading, with operations in multiple locations worldwide.
Jump runs extensive on-premises CPU and GPU infrastructure to power its quantitative research and trading strategies.
The research summary also notes that Jump has seen a notable rise in trading volumes, particularly in contracts valued in the lower hundreds of millions.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.