What is changing
The leading high-frequency trading firms are shifting resources toward AI computing infrastructure. According to Bloomberg, as cited by The Decoder, the largest players are committing billions to cloud contracts, startups and owned data centers, competing with one another for available computing power.
The push stems from growing use of machine learning in forecasting market movements: computing is no longer a peripheral support, but a factor of production that these firms regard as decisive for their business.
The moves by individual companies
One industry executive estimates, according to the account reported, that spending on computing is now ten to twenty times higher than before the launch of ChatGPT. Different strategies are taking shape on this basis. XTX Markets is building a data center in Finland; Jane Street manages one in Texas with more than 4,000 Nvidia GPUs.
Hudson River Trading, for its part, spends $1 billion a year on artificial intelligence, according to Iain Dunning, its head of AI, who also announced a multiyear agreement with CoreWeave. Jane Street, by contrast, has chosen a different path, investing $1 billion directly in CoreWeave and committing $6 billion to its computing services.
On the semiconductor front, Jane Street led a funding round for startup Etched and became the first customer of its AI chip, presented as capable of delivering superior performance thanks to lower-voltage operation.
Risks and open questions
The perception of risk is an integral part of this shift: rising costs and the rapid aging of chips are seen by the industry as real dangers, again according to Bloomberg. Jan Boomaars, CEO of Dutch trader Optiver, nonetheless believes that not acting is the most dangerous option.
Several elements remain to be verified: the figures released come largely from Bloomberg's account, while the numbers on spending and commitments are not accompanied by public accounting documents in the available material. Moreover, the expected return on these investments is not quantified, nor is it clear how far the hardware race could alter the competitive balance among the companies involved.
Why it matters
The case shows how demand for AI computing is extending from Big Tech to specialized financial operators, with potential effects on the data center and chip supply chain. The choices described amount to a two-pronged strategy: buying cloud services on the one hand, and directly financing suppliers and becoming their customers on the other, as in the case of Etched.
Competition for computing power thus becomes a positioning factor among intermediaries that traditionally compete on execution speed and forecasting models.
