Russian-born math doctor turns AI into a ‘money printing machine’

Gerko, born in 1979, earned bachelor’s, master’s and doctorate degrees from Moscow State University, and two master’s degrees from Russian economics schools. He started his financial career at Deutsche Bank, before participating in the field of foreign exchange trading and then moved to live in the UK in 2006 and became a citizen.

In January 2015, Gerko founded and holds 75% of the shares in XTX Markets, a company that uses self-developed AI algorithms to trade market price differences between platforms. Their revenue in the UK last year reached $5.3 billion, up 44% from a year earlier. Profits also increased by 33%, to 2.3 billion USD, becoming one of the most profitable companies in the field of algorithmic trading. This number does not include the company’s branch in Singapore, which has the same revenue as the UK headquarters.

Industry rivals such as Citadel Securities and Jane Street have higher revenues, but require thousands of employees, instead of 250 like XTX. The company also has no outside investors, meaning all profits go to a small group of shareholders within XTX. Gerko is also among the richest billionaires in the UK, with an estimated fortune of $12 billion.

Theo WSJthis year, XTX plans to operate the first of five data center deployments in Finland, expanding its capacity to process large amounts of financial data and train AI models to predict market price movements.

The company built its own complex worth more than $1 billion, instead of renting computing power from technology businesses – a common approach in the financial industry. This helps them keep AI model training costs relatively cheap, amid the AI ​​craze that requires more and more data processing capacity worldwide.

This is a new race in the fiercely competitive capital sector. Many years ago, electronic trading businesses built the fastest wireless networks that could execute trading orders in a few milliseconds. Nowadays, when faced with physical limitations, they gradually turn to developing smarter methods. Gerko is considered a leader in the shift to AI in the financial trading industry.

“Companies try to leapfrog and surpass each other by moving faster and faster. We decided to stay out of that game,” Gerko told students in a conference in 2019.

XTX is headquartered in London’s Knowledge District, near many offices of leading technology corporations such as Google and Meta. The company displays a full-size replica of the Apollo 11 command capsule and an Enigma encryption machine built by Germany in 1943.

No XTX employee works as a traditional trader. They do not decide which assets to buy and resell, but monitor a series of algorithms using deep learning models to predict price fluctuations in the market after a few seconds to a few hours. Many of XTX’s trades revolve around a strategy of willingly buying and selling assets and capturing the difference, others focus on spotting deviations from market rules and betting the price will quickly return to normal.

Every day, this business trades assets, such as stocks, foreign currencies and digital currencies, worth an average of 250 billion USD. Sources familiar with the matter revealed that XTX recently expanded into electricity trading.

To train the prediction model, XTX uses about 25,000 AI chips, most of which are manufactured by Nvidia. The first data center in Finland has an area the size of three football fields, while the second facility is nearly completed and is expected to operate from 2027. Finland was chosen because of its cool weather, which aids in cooling the system.

“Their approach seems to be to go all in on the AI ​​gamble,” said Agustin Lebron, a researcher at trading firm EquiLibre Technologies in the Czech Republic.

Former employees told WSJ that Gerko is a manager who is very close to the company’s operations, always appearing on Slack channels with an avatar as a cartoon character. Rick and Morty. The image was recently changed to Thanos, the villain in the series Avengers.