“‘King of the Geeks’: How Alex Gerko Built a British Trading Titan: XTX Markets Conquered Foreign Exchange Trading and Made Its Russian-Born Founder a Multibillion-Pound Fortune”, 2024-10-14 (; similar):
…“What Alex Gerko has done [with XTX] is next generation, next level”, said Paul Rowady, founder of trading consultancy Alphacution Research Conservatory. “There’s no other example of a high frequency style trading firm starting the way that they did, in London, in foreign exchange…and then growing so quickly.”
…XTX made £1.5b in profits across its UK and Singapore entities last year, and Gerko and a small group of quants and developers shared a profit pot of £747m, filings show. Gerko took £413m of that; the remaining £334m was split among 25 other people. In 2022, when the entities reported profits of £1.6b, Gerko earned £417m; 24 others shared £350m.
…In an era when the likes of Virtu Financial have built businesses shaving microseconds off the speed it takes to execute trades, XTX has instead prioritized using reams of market data and artificial intelligence to build its trading models, and advocated against what it calls a “destructive speed race”. Gerko believes “the best way for the market to be brutally efficient is for whoever is smartest to be given a fair crack”, said a person who worked at GSA and XTX. “But if the market is structured in such a way that only the fastest will win, then the market’s monopolized.” Rivals argue this is because XTX entered the speed game too late. “They try to preach as [if] they’re on some moral pedestal…That’s only because they weren’t good at high-frequency trading”, said a competitor.
XTX’s quant researchers use machine learning to scan trillions of market data points, price points, and other factors, in order to build statistical models that learn when and how to price different assets for investors.
That requires vast amounts of computing power. Gerko and XTX’s interest in artificial intelligence is profound, extending to early stage investments in AI start-up Anthropic, processing company Groq and British autonomous vehicle company Wayve, among others.
“We were in the same building as DeepMind. During fire drills I almost thought HR would try to poach them”, quipped a former employee.
But it is the firm’s consumption of Nvidia’s AI chips that underlines the centrality of the technology to its trading strategies.
XTX’s research is handled through a supercomputer in Iceland, where cold weather keeps cooling costs down and geothermal energy is cheap. It can hold 400 petabytes of data, equivalent to about 80tn digital photos. To achieve that, XTX has spent more than £150m on its 25,000 AI chips, according to people familiar with the matter. Most are the last 3 generations of Nvidia hardware, making the low-profile trading firm one of the chipmaker’s biggest corporate customers behind governments, state-backed defence contractors and Tesla, according to a report from Air Street Capital. Nvidia did not respond to a request for comment.
On top of that, XTX is also building its own vast data center in Finland.
“XTX is unique”, said Alvaro Cartea, director of the Oxford-Man Institute of Quantitative Finance. “Their edge is machine learning and neural networks, which combined with their compute power, can outpower pretty much all other firms.” “The fact they have 25,000 GPUs and very few people is an indication that that’s where they think the value is”, Cartea added…It has about 250 employees globally, far fewer than at rivals.
[Interesting GPU:employee ratio; reminds me of Google’s statement that it recently began spending more on compute than labor.]
…That majority shareholder shows no sign of selling. GSA sold its stake back to XTX in 2017, according to people familiar with the matter. XTX has not taken outside investment since. GSA declined to comment…“Alex wants to keep control, it’s his baby. Why would he give up [the] business to then comply with all the obligations of an IPO? There’s no upside for him”, said someone who worked with him.
…Now, Gerko’s focus appears to be on mathematics, outcompeting rivals and taking market share. “They make money hand over fist”, said Jesse Forster, head of equity market structure at Coalition Greenwich. “XTX remind me a bit of Renaissance Technologies”, he added, referring to the hedge fund whose billionaire founder Jim Simons was known for his love of code-breaking and mathematics rather than money. But a person who knows Gerko disagreed: “He wants to take on all of them, he wants to be better than Renaissance, better than Citadel.”