High-frequency trading firms are competing for the computing power behind their growing use of machine learning. Their approach spans cloud service contracts, purpose-built data centers, and direct backing for companies developing AI chips.
More computing power for market predictions
Firms such as Jane Street and Hudson River Trading increasingly use machine learning to anticipate market moves. That reliance has made access to computing capacity a strategic concern, and the largest traders are spending billions across cloud services, startups, and their own facilities, Bloomberg reports.
One executive estimates that compute spending is now ten to twenty times higher than it was before ChatGPT launched. The estimate points to how quickly the demands of AI have changed the scale of investment these firms consider necessary.
For trading companies, the goal is to have the computing resources needed to develop and run their models. The reported investments suggest that firms are pursuing several routes at once, rather than relying on a single supplier or facility.
Data centers and cloud deals
Some firms are building their own infrastructure. XTX Markets is building a data center in Finland, while Jane Street operates one in Texas with more than 4,000 Nvidia GPUs.
Others are committing heavily to cloud computing. Hudson River Trading spends $1 billion a year on AI, according to its AI chief, Iain Dunning, and has signed a multiyear deal with CoreWeave.
Jane Street has both invested in and purchased services from that provider. It invested $1 billion directly in CoreWeave and committed $6 billion to its computing services. These arrangements show how demand for AI capacity can connect trading firms to companies that supply the infrastructure.
Backing a different kind of chip
Jane Street has also backed a chip startup. The firm led a funding round for Etched and became the first customer for its AI chip, which is supposed to deliver more performance by running at a lower voltage.
That investment adds another layer to the firms’ computing strategies. Alongside building facilities and arranging cloud access, a trader can support the development of hardware intended to meet AI workloads.
The source does not describe the chip’s performance in use or say how broadly it will be adopted. Its significance here is the willingness of a major trading firm to fund a startup and become its first customer while the industry is seeking more AI computing power.
High costs, but hesitation has its own risk
There are clear risks in this race. Bloomberg reports that industry participants see rising costs and fast-aging chips as serious concerns. Large commitments to computing resources can become harder to justify if the hardware loses value quickly or the expense grows faster than expected.
Yet some executives believe standing still carries a greater danger. Jan Boomaars, CEO of Amsterdam-based high-frequency trader Optiver, is among those who see doing nothing as more dangerous.
Together, these pressures help explain why firms are investing across infrastructure, cloud services, and chip development despite the expense. Their decisions reflect a judgment that machine learning’s role in predicting markets makes computing access too important to leave entirely to chance.
The competition is therefore about more than adding servers. It is about securing the capacity and hardware that firms believe they need to keep using AI in a business where predicting market moves matters.