The artificial intelligence market in Russia is undergoing a tectonic shift. According to my analysis of data from the fintech service "Pay Around the World," in July, the share of cardholders paying for AI service subscriptions reached 51.6%. This means that every second user paying with a card now considers neural networks an integral part of their budget. The average monthly spend is 2,097 rubles, with a typical single payment of 1,905 rubles, which exactly matches the price of a basic subscription to a major chatbot, including taxes.
One service instead of a bundle of subscriptions
The common belief that users assemble a portfolio of dozens of AI products is not supported by the numbers. 74.3% of payers subscribe to only one service, 23.9% to two or three, and only 0.1% to seven or more. On average, each person uses 1.35 services, and this figure has remained stable for six months. People find one tool that fits their needs and work with it, rather than collecting new releases.
Of particular interest is the behavior of the top segment. One in seven users (15.2%) spends more than 10,000 rubles per month on AI, while accounting for 57% of all spending on these services. For them, this is no longer a subscription for convenience, but a full-fledged work tool, often paid for on a usage basis rather than a fixed rate.
Turnover nearly quadrupled in five months
From March to July, turnover on AI services increased 3.85 times, and the number of payers grew 3.45 times. At the same time, average spending per person remained nearly unchanged: 2,178 rubles in March versus 2,097 in July. The market is expanding outward due to an influx of new users, not from increased spending by existing ones. This is a classic sign of a technology transitioning from the "toy for enthusiasts" category to ordinary monthly payments, just as happened with streaming services.
For businesses, the picture is even more telling. On the Yandex AI Studio platform, corporate clients spent 597 billion tokens in the first half of the year—18 times more than a year earlier. Chinese models Qwen, GLM, and Kimi lead the way. At the same time, dependence on foreign GPUs remains critical: authorities are preparing tax incentives for the purchase of AI equipment, and companies MWS AI and Rubytech are already testing systems on eight Chinese GPUs, with commercial launch slated for 2027.
We are witnessing the formation of a dual demand loop: mass consumer subscriptions from below and a regulated corporate-government segment from above. AI has finally become an expense item, not an experiment.
My conclusion: the stability of the average ticket amid explosive growth in the number of payers is a signal that the market is entering a maturity phase. Investors and analysts should keep an eye on the corporate segment and infrastructure solutions, which will determine the price of end-user subscriptions in the coming years.