Executive leadership in global enterprises is facing a sharp reality check regarding artificial intelligence. The KPMG Global AI Pulse study for the second quarter of 2026 reveals that corporate enthusiasm is clashing directly with operational expenses. While spending remains high, spiraling costs related to autonomous AI agents are forcing executives to rethink their scaling strategies.
According to KPMG, 49 percent of surveyed leaders have slowed down or scaled back the deployment of autonomous AI agents. The primary drivers for this rollback are operational expenses, particularly token and metering costs associated with complex tasks. In many cases, these ongoing variable costs exceeded the immediate financial benefits generated by the agentic systems.
The survey of 2,145 C-level executives across 20 countries highlights a growing gap between perceived value and measurable returns. Although 76 percent of leaders state that artificial intelligence delivers substantial business value, a mere 7 percent can mathematically prove a positive return on investment. Furthermore, 42 percent of companies acknowledge incomplete visibility into where their operational AI budgets are actually being consumed.
Despite these cost concerns, investment volumes remain massive across large corporations. Participating companies spend an average of 188 million US dollars on artificial intelligence, with 79 percent continuing to treat the technology as a top strategic priority. Interestingly, organizational accountability plays a decisive role: when the chief executive officer takes direct responsibility for AI outcomes, the likelihood of demonstrating a positive ROI increases fivefold from 4 percent to 14 percent.
Parallel findings from DXC Technology's Digital Future Monitor highlight an additional dilemma around control and governance in the DACH region. In Germany, Austria, and Switzerland, 80 percent of companies strictly require a human in the loop for final decision-making processes. At the same time, 58 percent of these executives consider it likely that AI systems will eventually make major strategic corporate decisions completely autonomously in the future.
This tension between strict human oversight and expected future autonomy creates complex compliance challenges. DXC reports that 66 percent of executives face new audit and compliance tasks due to the black-box nature of advanced AI models. As 71 percent anticipate a rapid increase in virtual AI agents over the next three years, establishing precise financial controlling and clear operational governance is becoming essential for enterprise AI adoption.