The collaboration merges Asus’s high‑performance server platforms with Poesis.ai’s proprietary financial‑modeling AI, and both parties plan to rely on Nvidia technology to power the end‑to‑end trading system.
Poesis.ai, founded by Alex Popa, describes its mission as building a “world model for finance” – an artificial intelligence that not only reads market data but learns how markets function, enabling real‑time decision‑making. The firm’s chief scientist, Charles Elkan, brings experience from senior roles at Goldman Sachs and Amazon to the venture.
Unlike monolithic AI solutions, Poesis.ai employs a modular architecture that separates trading intelligence into interchangeable components. This design allows individual modules – such as price‑trend analysis, risk management or sentiment parsing – to be
On the hardware side, Asus has been expanding its AI infrastructure portfolio through a deepening partnership with Nvidia. The company’s XA P2N‑E2 server, built on Nvidia’s Vera CPUs, is specifically optimised for intensive AI workloads and will serve as the computational backbone for the autonomous agents.
Nvidia itself is pushing agentic AI for finance. At its GPU Technology Conference (GTC) in 2026, the chipmaker showcased a multi‑agent trading platform that integrates diverse data streams, including news sentiment and price movements, with coordinated agents that collectively manage trade execution and risk parameters.
Regulators are watching the development closely, as autonomous trading raises questions of accountability. Authorities are probing who would be held responsible if an AI‑driven trader makes a costly error – the hardware manufacturer, the software developer, or the financial firm that deploys the system.