QuantumLight’s $500M Fund Tests Whether AI Can Outpick Human Venture Capitalists

Fund I attracted an undisclosed group of limited partners that included technology founders, family offices and traditional financial institutions. Bloomberg reported that Storonsky and Kondrashov supplied roughly a quarter of that fund’s capital themselves.

Reporting on Fund II says it was oversubscribed, although QuantumLight has not publicly detailed the full list of outside backers or the fund’s allocation terms. The most defensible conclusion is therefore that at least some investors are prepared to commit more capital to a machine-learning-led venture strategy, especially one associated with a well-known fintech founder and a distinctive operating model.

What the fund does not show is equally important. A larger fund is not proof of superior risk-adjusted returns. It demonstrates fundraising momentum and investor willingness to back the thesis; it does not establish that Aleph selects better companies than experienced human investors, or that the strategy will generate stronger net returns after losses, fees and carry.

How the model differs from traditional VC

Traditional venture capital often combines market research with partner judgment, founder references, relationships and pattern recognition. QuantumLight’s approach tries to move the centre of gravity toward structured data.

The potential advantages are clear:

• Broader screening: Software can process far more company and founder information than an individual partner can review manually.
• Less dependence on networks: A systematic process may identify companies outside the investor’s existing social and professional circles.
• More repeatable decisions: A quantitative framework can make it easier to apply the same criteria across opportunities.
• Faster initial selection: AI-assisted screening can reduce the time spent narrowing a large pipeline before human diligence begins.

The risks are also substantial. Historical data can reflect the biases of earlier funding markets. A model may overvalue signals that are easy to measure while missing qualities such as unusual founder insight, market timing or a breakthrough that has no close precedent. And if many investors use similar datasets, the apparent advantage of systematic sourcing could narrow.

QuantumLight is part of a broader movement toward software-assisted venture capital, including AI tools for deal sourcing, underwriting and portfolio analysis. The emergence of dedicated AI-native deal-sourcing companies suggests that venture firms increasingly view data infrastructure as an investment capability in its own right.

The returns question remains open

QuantumLight’s strategy is still too young for a definitive performance judgment. Its reported record of algorithmically recommended investments is a description of process, not a public track record of realised exits and net fund returns.

A serious evaluation would need to compare the firm’s investments with suitable conventional-VC benchmarks while accounting for entry valuations, follow-on decisions, losses, liquidity and the timing of exits. Those outcomes take years to mature, particularly for growth-stage technology investments.

This is the central distinction between an interesting investment experiment and a proven asset-management strategy. QuantumLight may eventually show that a model can identify opportunities human investors systematically miss. It may also show that data-driven selection works best as a complement to human judgment rather than a replacement for it. The available evidence does not yet resolve that question.

Why Storonsky’s dual role draws scrutiny

Storonsky’s involvement gives QuantumLight unusual founder credibility and a potentially powerful source of operating experience. It also creates a governance question: how should responsibility, conflicts and executive attention be managed when the same entrepreneur leads Revolut while co-founding and backing a rapidly expanding venture firm?

That question becomes more consequential as Revolut grows its regulated banking operations. In August 2026, Revolut secured a French banking licence following approval involving France’s ACPR regulator and the European Central Bank. Reuters described the licence as an important step in the company’s European expansion and noted that Storonsky has linked the development to the possibility of pursuing a US banking licence.

The public reporting available here does not establish that a US listing has been formally decided or filed. Nor does it demonstrate a conflict between Revolut and QuantumLight. But the combination of regulated expansion, outside investment activity and a possible future public-market path makes clear separation of duties, transparent conflict procedures and accountable leadership especially important.

The bottom line

QuantumLight’s $500 million Fund II is a meaningful market signal: investors have shown enough interest in its quant-style approach to support a vehicle twice the size of its debut fund. Its use of Aleph also represents one of venture capital’s clearest attempts to make deal selection software-led rather than primarily relationship-led.

But fundraising is not performance. The firm’s model remains an experiment whose decisive evidence will come from realised investments over a much longer period. For now, QuantumLight is best understood not as proof that AI has replaced venture capitalists, but as a high-profile test of whether machine learning can make startup investing broader, faster and more consistently successful.