High-frequency trading (HFT) firms including Jane Street, Hudson River Trading, and XTX Markets are ramping up capital expenditure on artificial intelligence compute infrastructure. According to a Bloomberg report, compute spending among top quantitative trading operations has grown ten to twenty times higher than pre-ChatGPT levels as strategies rely more heavily on machine learning models.

To secure capacity, firms are pursuing direct infrastructure investments alongside massive cloud agreements. XTX Markets is constructing a dedicated data center in Finland, while Jane Street operates a Texas facility equipped with over 4,000 Nvidia GPUs. Meanwhile, Hudson River Trading spends $1 billion annually on AI compute and recently inked a multiyear deal with cloud provider CoreWeave.

Jane Street has taken a direct investment approach to the AI supply chain, putting $1 billion into CoreWeave alongside a $6 billion computing contract. The trading firm also led a funding round for custom chip startup Etched, becoming the launch customer for its low-voltage hardware as firms attempt to control long-term operational costs.

Why it matters

  • Quantitative trading firms are becoming major anchor clients for specialized AI cloud providers and custom silicon startups.

  • Compute expenditure in algorithmic trading has escalated up to 20x, turning infrastructure ownership into a key competitive moat.

Source: the-decoder.com