Financial companies want to profit from their own AI. Will their investors let them?

This year, at least three financial companies—the neobank Revolut, the credit fintech Affirm, and the payment system Mastercard—announced the launch of their own AI models instead of implementing solutions from third-party developers. AI is no longer merely an operating expense but is becoming a distinct asset class for financial companies. But when—and under what conditions—can this increase the value of their businesses?

In late August of this year, Revolut unveiled its own AI model, PRAGMA, and announced the creation of a dedicated division, Revolut Research, to develop it. The neobank’s AI analyzes sequences of banking transactions for each customer to perform credit scoring, detect fraud, and recommend the most suitable products to customers.

PRAGMA was developed in collaboration with Nvidia. This model was specifically designed to decipher complex financial behavior patterns and provide real-time risk assessments. According to Revolut, in tests using historical data, PRAGMA demonstrated 2.3 times greater accuracy in assessing customer default risk, helped identify 65% more cases of fraud, and increased the relevance of recommendations by 41%.

Revolut co-founder and CEO Nick Storonsky said in an interview with Reuters that the company uses 30–40 million transactions per day to develop its own AI models.

This isn't the only neobank of its kind: In March 2026, the Mastercard payment system announced its new generative AI model, trained on billions of anonymized transactions.

In September 2026, the American credit fintech company Affirm also announced that it had launched its own AI model, which more accurately assesses customers’ creditworthiness (including those with limited credit history) and enables the company to approve loans for them. The previous assessment system had rejected some applications from such customers.

For banks and fintech companies, the primary benefit of using AI lies in reducing the cost of the risk assessment process—and, consequently, the cost of serving customers. This is important for Revolut, as the company plans to go public and, before doing so, increase its valuation to $150–200 billion.

Clint Sukermani, founder of The AI Value Institute, wrote that it is time for financial institutions to stop treating scalable AI initiatives solely as expenses and start viewing them as strategic investments in business growth.

But what does this mean for investors in such companies? How the market reacts to companies’ plans and the actual implementation of AI has not yet been sufficiently studied, notes Zihan Tu, a researcher at the University of Warwick’s Faculty of Economics. In his study, the results of which were published on September 22, 2026, he analyzed announcements by S&P 100 companies regarding the use of AI in 2023–2024 and assessed how they were linked to the market’s short-term reaction. It turned out that announcements of actual AI implementation yielded 4.6 percentage points more abnormal returns on their securities (beyond what would be expected given general market trends and other risk factors) compared to announcements of plans to implement AI.

In an earlier study from 2025, McKinsey estimated that banks that are among the first to adopt AI will be able to increase their return on tangible capital (ROTE) by up to 4 percentage points through business model transformation and the creation of additional economic value. Conversely, banks that delay the adoption of AI are likely to face declining profits in the long term.

Investors are unlikely to be willing to pay a premium simply because a bank or fintech company has its own AI models, wrote Antonis Dragiotis, a Greek expert in risk management, artificial intelligence, and banking technology, and coordinator of the AI Catalyst financial services working group, Antonis Dragiotis, wrote in a commentary for Oninvest.

According to him, they will need evidence that implementing such models will yield better economic results. Moreover, as more and more companies adopt AI models, having them will no longer be a significant competitive advantage in the future.

Vadim Ivanenko, an international expert in financial technology, notes that if a bank or fintech company’s proprietary AI model enables it to reduce the cost of serving a customer—for example, by 30%, reduce fraud losses, improve the quality of its loan portfolio, increase the conversion rate of financial products, or boost customer retention—that is real economic value.

He recommends that investors in companies that have announced the implementation of their own AI models or AI tools analyze the following company metrics before and after implementation:

— the cost-to-income ratio and the cost of serving a single customer or processing a single transaction;

— fraud losses and the false positive rate;

— cost of risk, credit losses, and the quality of scoring models;

— customer lifetime value (CLV), customer retention, conversion rate, and revenue per customer;

— Additional operating profit and return on AI investments, taking into account infrastructure, development, monitoring, and support.

It is important to compare these metrics in order to distinguish the effect of AI implementation from other factors, adds Vadim Ivanenko.

Independent stock market analyst Vadim Kondratyuk adds: a premium to the valuation is justified only if AI increases a company’s ROTE or lowers its cost of equity (that is, if the market begins to view it as more reliable).

He recommends that investors pay attention to three metrics: the trend in the cost-to-income ratio, the cost of risk over the credit cycle (whether it is declining due to improved borrower assessment), and revenue per customer. The latter should grow at a faster rate than that of competitors.

In Revolut’s case, the implementation of its proprietary AI model has not yet been reflected in its financial statements. For the first half of 2026, Revolut Holdings Europe’s net operating revenue rose 36.8% year-over-year to €1.24 billion, while operating expenses increased 36.5% to €1.055 billion. The cost-to-income ratio fell from 82.9% to 80.2%, but credit losses nearly tripled—from €25.7 million to €76.1 million.

Affirm’s stock rose 1.32% the day after the company announced the launch of its own AI model. It is also not yet possible to assess its impact on the company’s metrics, as only a month and a half has passed since the launch. However, Affirm has already stated that with the new scoring system, the number of completed purchases has increased by 3.4% compared to the control group. Furthermore, the additional loans approved were of higher quality: borrowers were better at making payments than customers who received loans using Affirm’s previous models. In the second quarter of 2026 (the most recent data available), Affirm’s operating profit nearly tripled to $147 million, and its operating margin rose from 6.6% to 12.6%. At the same time, operating expenses jumped 24.5%, and provisions for credit losses rose 42.5% to $223 million.

For Mastercard, the situation is somewhat different: the banking metric of “cost of risk” does not apply directly to it. For a payment system, fraud losses and the false positive rate (when a legitimate transaction is flagged as fraudulent). Back in 2024, the company reported that the implementation of third-party generative AI resulted in an average 20% increase in fraud detection, and in some cases up to 300%, and a reduction in false positives by more than 85%. In the second quarter of 2026, Mastercard’s revenue from ancillary services and solutions—including security services and the deployment of its proprietary AI technologies— grew 20% year-over-year to $3.83 billion.

“There’s an interesting paradox: the more accessible AI becomes, the less value there is in the mere fact of using it, and the greater the value of the ability to apply it effectively in a specific business,” says Vadim Ivanenko.

According to him, AI will significantly transform the banking industry, but the market today tends to overestimate the pace of these changes and underestimate the complexity of implementing them in practice.

In addition, the players themselves are beginning to slow down the adoption of AI. For example, in October, the U.S.-based Goldman Sachs banned its traders and bankers from using third-party AI tools—Claude Code from Anthropic and GitHub from Microsoft—after analyzing their performance and risks.

Ivanenko believes that large banks and financial platforms have an advantage over smaller players thanks to their millions of customers, historical data, and massive transaction volumes. On the other hand, traditional banks often have fragmented IT infrastructure, complex approval processes, and high costs associated with making changes. However, a large bank will not necessarily be more successful than a small fintech company. Sometimes technological flexibility, data quality, and speed of implementation are more important than sheer scale.

This article was AI-translated and verified by a human editor