Jane Street, a rising star in Wall Street's quantitative trading industry, announced its astonishing 2025 results: trading revenue reached $39.6 billion, and adjusted EBITDA was approximately $31.2 billion (about RMB 210 billion), surpassing JPMorgan Chase and Goldman Sachs to set a new Wall Street record. The company paid out $9.38 billion in compensation to its 3,500 employees, averaging approximately RMB 18.29 million per person. Founded in 2000, Jane Street is one of the world's leading quantitative market makers, covering over 200 trading platforms in 45 countries, using its own funds to participate in trading across multiple asset classes including ETFs, stocks, and options. The company has a unique management style, using the less common programming language OCaml, and is governed by dozens of partners. Unlike DeepSeek, Jane Street has chosen to heavily invest in the AI industry chain, investing in companies such as Anthropic and CoreWeave, forging a different path in the AI race than Liang Wenfeng.
Article author and source: Investment Community
Some people are quietly making money.
This time, it's Liang Wenfeng's turn to talk about his peer – the rising star in quantitative trading, Jane Street. Jane Street recently delivered an astonishing performance: in 2025, the firm's trading revenue reached $39.6 billion, and its adjusted EBITDA reached $31.2 billion (approximately RMB 210 billion), breaking Wall Street's historical record.
What's astonishing is the compensation: $9.38 billion is paid out to employees annually, averaging 18.29 million yuan per person.
What's even more interesting is that after DeepSeek became popular, Jane Street was often compared to Leung Man-Fung and Magic Square Quant. However, not everyone can become DeepSeek.
Hong Kong is still hiring with an average salary of HK$18 million per person.
The latest data reveals that Jane Street paid out $9.38 billion in compensation to its employees in 2025, more than double the previous year. Based on 3,500 employees, that's an average of $2.68 million per person (approximately 18.29 million RMB).
What does this mean? It's equivalent to about seven times the average salary at Goldman Sachs. In fact, even for newcomers, Jane Street has always offered generous compensation. Their website previously showed that their quantitative trader positions for recent graduates offered a base salary of approximately $300,000 per year, not including bonuses.
Supporting these astronomical salaries is an astonishing track record. Jane Street's trading revenue reached a staggering $39.6 billion in 2025, surpassing JPMorgan Chase and Goldman Sachs, and setting a new Wall Street record for trading revenue. What's truly astonishing to working professionals is that Jane Street distributes approximately one-quarter of its revenue to its employees.
In comparison, JPMorgan Chase's trading revenue was $35.8 billion last year; Citadel Securities' revenue was $12.2 billion; and Blackstone's revenue for the year was $14.45 billion.
Even more astonishing are the profit margins. A filing with creditors shows that Jane Street's EBITDA reached $31.2 billion last year, generating nearly $9 million in profit per employee. Even after adjusting for certain factors, these are still staggering figures.
However, how can a non-bank institution with more than 3,000 employees stand at the forefront of Wall Street?
In short, Jane Street is one of the world's most important quantitative market makers. It doesn't earn money through underwriting, consulting, or management fees; instead, it primarily uses its own capital to participate in trading. Leveraging quantitative models and high-frequency trading systems, it operates in 45 countries and on over 200 trading platforms worldwide, covering a wide range of assets including ETFs, stocks, options, bonds, and foreign exchange. It has offices in Amsterdam, Chicago, Hong Kong, London, New York, and Singapore. To date, its membership equity, or the company's own capital, has grown to $45 billion.
Jane Street has been frequently seen in Hong Kong's capital markets over the past year.
Hong Kong Stock Exchange filings show that Jane Street has appeared on the list of cornerstone investors for the Hong Kong IPOs of companies such as Sanhua Intelligent Control, Leshushi, and IF Coconut Water. In addition, Jane Street has leased six floors of office space in Phase 1 of the flagship Central Harbourfront project in Hong Kong, covering an area of over 220,000 square feet, with monthly rent reaching tens of millions of Hong Kong dollars, setting a new record for the largest leasing transaction in the area.
The expansion continues. Jane Street currently has 45 open positions in Hong Kong, covering areas such as quantitative traders, machine learning engineers, data center engineers, and China business development.
This company, which was so low-profile that it was almost invisible, is gradually coming to the forefront.
The most extravagant financial institution has emerged.
The story of Jane Street begins in 2000.
That year, 28-year-old Rob Granieri resigned from the quantitative trading firm SIG and, together with several former colleagues, founded Jane Street in a small, windowless office in New York.
The company started by trading American Depositary Receipts (ADRs), which are shares of foreign companies listed in the United States. They quickly expanded their business to options and ETFs. At that time, ETFs were a small, niche market, with a market size of only tens of billions of dollars, far from the massive $19 trillion business they are today.
Subsequent events proved that Jane Street had accurately predicted the times.
Especially after the 2008 financial crisis, banking trading activities were subject to stricter regulations, and the resulting space was taken over by some proprietary trading firms. By 2020, the Federal Reserve included Jane Street in its list of qualified counterparties for crisis response tools, placing it alongside established Wall Street institutions like JPMorgan Chase.
Quietly, Jane Street has become one of the most important market makers in the ETF market. In 2025, Jane Street contributed 14% of the trading volume of US ETFs and 20% of the trading volume in Europe. In the bond ETF sector, 41% of creation and redemption transactions were completed through it.
However, Jane Street is not without controversy. It has faced a string of controversies, including allegations of market manipulation by the Securities and Futures Commission of India, insider trading lawsuits related to the Luna crash, and the Bitcoin ETF controversy.
Beyond the controversy, what makes Jane Street truly special is that it didn't follow the traditional path of Wall Street.
Compared to a typical financial institution, Jane Street is more like a math and programming-driven trading lab: it has long used the obscure programming language OCaml to build almost all of its trading systems. There is no star CEO or investment banking titles; it is governed by dozens of shareholder partners. Most of the founders have retired, with only Rob Granieri remaining. He has no official title, and his picture isn't even in the employee directory, so many colleagues don't recognize him.
Every one or two months, the company's official website releases a new puzzle, inviting people around the world to solve it. "The puzzles may seem abstract, but their essence is exactly the same as our work. Only by discovering new problems in the financial markets and finding new solutions can we continue to thrive," Jane Street explains this tradition.
Jane Street is frequently in the spotlight for its almost legendary recruitment methods.
Some have recalled spending an hour understanding the rules of a card game during interviews, and another hour figuring out the optimal winning strategy. Others have been asked questions like, "How would you estimate how many windows there are in New York City?" The final round is known as Super Day. Candidates are reportedly given 100 chips and must continuously bet and participate in market making throughout four to six rounds of technical interviews. If they run out of chips, the offer is gone.
Jane Street employs such a ruthless selection process for only one purpose: to identify true geniuses.
From Finance to AI: Embarking on the Path of Magic Square Quantitative Analysis
I still remember when DeepSeek suddenly appeared, there was a rather imaginative discussion: if a Wall Street quantitative giant turned to AI, would it become the "American version of DeepSeek"?
Such an association is not difficult to understand.
When viewed side-by-side, Jane Street and Magic Square Quant, led by Liang Wenfeng, do share certain similarities: both are extremely low-profile, both started with quantitative trading, and both possess ample computing power. More importantly, they both have a group of people behind them who are accustomed to modeling, calculating, and betting amidst uncertainty.
However, unlike Jane Street, Liang Wenfeng ultimately incubated DeepSeek, while Jane Street chose a different path, heavily investing in the AI industry chain.
The most notable investment is in Anthropic. This large-scale modeling company, founded by former OpenAI employees, is valued at nearly a trillion dollars, and Jane Street was an early investor in it. In addition, cloud computing service provider CoreWeave and AI unicorn Thinking Machines Lab are also among its investments.
Going back to that idea, Jane Street is unlikely to become the next DeepSeek.
What truly shook the world about DeepSeek wasn't just its model capabilities. More importantly, it shattered the dominant narrative in the AI world—the competition for large models isn't necessarily defined by the companies with the most funding and computing power; newcomers can also carve out their own niche through R&D efficiency and different approaches.
Leung Man-fung once admitted, "We didn't intentionally become a catfish; we just accidentally became one."
This statement remains thought-provoking even today.
In fact, the narrative surrounding AI in China has quietly shifted. Open-source ecosystems, low-cost inference, adaptation to domestic computing power, and engineering efficiency are becoming new key terms. Just last month, DeepSeek-V4 listed Huawei Ascend and NVIDIA GPUs side-by-side on its verification platform for the first time, and eight domestic AI chip manufacturers almost simultaneously announced the completion of adaptation.
This is precisely why the significance of DeepSeek becomes even clearer. Its value lies not so much in providing a replicable template, but rather as a reminder: in this long race of AI, giants will approach and capital will pour in, but those who truly rewrite the table will not always be the seemingly stronger party.
