Йельский университет: компании с ИИ дают акционерам +0,64% в неделю, но бизнес не готов

Companies that apply AI in business processes deliver an additional 0.64% return per week to shareholders — that is the conclusion of a Yale University study published on July 13, 2026, based on the analysis of 380 trillion AI tokens. Cloud provider Cloud.ru has launched a survey among IT professionals on Habr to find out what is actually preventing Russian companies from adopting AI.

What the Yale Study Found

Yale University analysts collected and processed 380 trillion AI tokens — one of the largest datasets on the use of generative models in the corporate sector. The main finding: companies that have integrated AI into operational processes consistently outperform competitors in stock returns. An additional 0.64% per week — annualized, that is approximately 33% additional growth, which cannot be dismissed as statistical noise.

Importantly, the gap is recorded specifically for AI tools, not for overall IT spending. Companies with high technology budgets but no real use of generative models do not show a similar advantage.

Today, generative models cover a wide range of business tasks:

• writing code, technical documentation, and content
• creating advertising materials and marketing copy
• data analysis, forecasting, and report generation
• automation of contact centers and customer support
• optimization of internal operational processes

Why Is Business Not Ready for AI?

A significant share of companies has either not launched AI projects at all or remains stuck at the pilot stage without transitioning to production deployment. Cloud.ru identifies several systemic barriers.

Lack of formalized processes. A generative model amplifies what already works well, but does not fix chaotic operations. If a business process is not documented, there is nothing to automate: AI will reproduce inefficiency faster, not eliminate it.

Infrastructure gap. Industrial deployment of generative models requires computing power, integration with enterprise systems, and reliable data storage. Many companies have neither a suitable cloud foundation nor an internal team to maintain it.

Shortage of expertise and budgets. AI adoption is not a license purchase — it is a complex project: customization for industry specifics, employee training, continuous refinement, and model maintenance. For mid-sized businesses, the cost of this path often turns out to be unexpectedly high.

"Many businesses are not yet ready for AI — due to the absence of formalized processes, infrastructure, budgets, knowledge, and other reasons,"

Cloud.ru states.

A separate trap is inflated expectations. Management often perceives AI as a savior technology that will fix operational problems on its own. The reality is different: generative models amplify what is already working, but cannot serve as the entry point for digital transformation from scratch.

Why Is Cloud.ru Conducting the Survey?

Cloud.ru launched a survey among developers, architects, CTOs, and product managers on Habr. The goal is to obtain first-hand data from those who have implemented AI, attempted to do so, or consciously declined. The provider wants to understand the real barriers — not to repeat the conclusions of analytical reports from major consulting firms.

Most AI adoption studies rely on data from large corporations. The voice of mid-sized businesses, startups, and individual technical teams is weakly represented in such statistics — yet that is precisely where the bulk of unfinished projects is concentrated. The survey results are planned for open publication.

What This Means

Yale University's data moves AI transformation out of the category of "technology trend" into the category of measurable financial risk. Companies that delay are not losing abstract competitiveness — they are losing specific percentage points of shareholder return. The gap between those already using AI and those still preparing will widen with each quarter. Cloud.ru's study should help businesses honestly assess their readiness and find real unblocking points — rather than simply adding "AI adoption" to their list of strategic priorities.