Sber has completed Russia's first leasing deal for the GigaChat Enterprise software and hardware complex. The contract involved server equipment with a pre-installed neural network model from the bank. SberLeasing acted as the organizer, and Salyut dlya Biznesa as the supplier. The buyer received the complex for 36 months with a minimal down payment and a discount on the equipment – there are no upfront costs.
The first client – a major Russian developer – has already installed the equipment and begun launching GenAI projects. Among the first solutions is an intelligent digital twin of a real estate sales manager. The neural network will generate personalized responses in real time, remember the context of each inquiry, and accompany the client throughout the entire transaction cycle.
According to Alexander Vedyakhin, First Deputy Chairman of Sberbank, there is no need to pay upfront – it is enough to sign a leasing agreement and start working.
In fact, the bank has transformed heavy AI equipment into a service with a predictable monthly payment. Leasing removes the main barrier to the implementation of generative models in the corporate sector – the need for one-time capital expenditures on server infrastructure.
Viktor Ventimilla Alonso, CEO of SberLeasing, called the case indicative and stated that the experience gained would be replicated. For the market, this is a signal: access to powerful neural network models is moving from capital-intensive projects to a tool with revolving financing.
