AI Takes the Wheel: Corporate Lending Moves to Full Autopilot
At Sberbank, the portfolio of loans issued to medium-sized and large businesses with the help of AI has surpassed 7 trillion rubles ($80.6 billion). The algorithms are no longer simply tools for bank employees – they make the decisions themselves in 50% of cases.
Every second deal, from application to disbursement, is now processed fully autonomously, with no human involvement. This amounts to a fundamental redesign of the financial pipeline. AI technologies are shedding their role as auxiliary tools and are being entrusted with decisions that affect the growth of the real economy.
Trillion-Dollar Scale
AI has been used in credit scoring, which assesses borrowers’ ability to repay, for years. But the level of algorithmic autonomy and the amount of money those systems handle continue to grow. They can analyze massive datasets, model cash flows, generate tailored offers and even monitor how borrowed funds are used for their intended purpose. It increasingly looks as if machines can handle nearly every step.
“The portfolio of loans issued by Sberbank to medium-sized and large business customers using AI has surpassed 7 trillion rubles ($80.6 billion), while every second new deal, from application submission to disbursement of the credit product, is processed fully autonomously, without the involvement of bank employees,” said Anatoly Popov, Deputy Chairman of the Management Board at Sberbank.
The industrial deployment of homegrown machine-learning (ML) models and big-data analytics technologies in a critical sector is a major indicator of the maturity of Russia’s IT industry. Financial services are one of its largest markets, and the sector is ahead of many others in deploying AI across business processes.
The technology is reshaping how banks operate internally while also affecting the broader economy. Algorithms can promptly verify borrower-provided information, assess risks and make lending decisions within minutes, allowing businesses to receive funds faster. By changing how companies are financed, this technological shift can accelerate investment projects, the launch of new production facilities, equipment purchases, and expansion of product lines and output.
The Lending Pipeline
The next stage of development will come from expanding the scope of the credit process. Anatoly Popov said Sberbank is deliberately increasing the share of autonomous lending deals. By January 2027, the bank plans to raise that share to 75% of all loans issued to businesses. Meanwhile, the models could be extended to more complex, personalized financial products that require analysis of projects, collateral and large volumes of unstructured data.
Similar developments are underway at other systemically important Russian banks. Alfa-Bank, for example, launched an AI platform for lending to large companies in April this year. Some 65% of new credit deals are now concluded through the platform, while its loan portfolio stands at 1 trillion rubles ($11.5 billion). At T-Bank, AI agents rather than employees make more than 90% of business-lending decisions.
The Bank of Russia supports the trend toward automation but requires algorithmic transparency and consistently high data quality. If a customer is dissatisfied with a lending rejection, the regulator also insists that the AI decision can be reviewed by a human. In any case, the financial institution remains responsible for potential errors made by neural networks. As a result, technological development will proceed alongside the creation of auditing mechanisms.
How AI Took Root in Financial Services
AI has been integrated into banking processes for years, with models gradually becoming more sophisticated and gaining broader decision-making powers. At Sberbank, for example, AI independently made nearly 98% of lending decisions for individuals in 2023 and about 30% of decisions for businesses. In 2024, algorithms were entrusted with 40% of corporate-lending decisions. That same year, Sberbank CEO German Gref said that “people do not make decisions at all” when loans are issued to individuals, with AI taking over the function entirely. Then, in 2025, Sberbank issued a legal entity its first fully automated loan: AI processed the application, assessed the customer’s ability to repay and risk factors, made the lending decision, calculated the interest rate and transferred the funds to the borrower. A human was not involved at any stage. Today, AI models autonomously handle every second corporate-lending deal.
A New Reality for the Digital Economy
By the end of this year, Sberbank plans to raise the share of fully autonomous deals to 75%. Extending autonomous lending to more complex corporate products, such as investment and project finance, is likely to be the next step. AI has effectively become a full-fledged participant in the capital market, determining the speed and direction of money flows. A significant share of the corporate lending pipeline now operates without direct human involvement. This is the new reality of Russia’s digital economy.
