The next wave of artificial intelligence (AI) has arrived — and it’s moving faster than most companies are ready for. According to S&P Global, nearly six in 10 enterprises are now actively pursuing “agentic AI,” a new generation of autonomous systems that don’t just respond to prompts but act on their own, coordinating workflows, calling other models and making decisions in real time.
That enthusiasm comes with a price, new data indicates. As organizations shift from chatbots to fully fledged digital agents, S&P Global warns that the underlying technology stack will come under far greater strain. That includes data centers, networks, security frameworks and governance policies.
“Companies that can harness agentic AI’s power early can have a significant competitive advantage,” said Eric Hanselman, chief analyst at S&P Global Market Intelligence’s 451 Research division. “But success depends on having the right data foundation and skilled teams in place.”
S&P: Agentic AI’s ‘significant competitive advantage’
The data suggests the rush is already on. S&P Global’s Voice of the Enterprise survey found that 58% of respondents are actively seeking opportunities to deploy AI agents, with only 1% dismissing the concept altogether. GPU shipment projections from major suppliers such as Nvidia have grown quickly since early 2023, rising more than fivefold for 2025 and 2026 as vendors scramble to meet escalating compute demand.
“AI’s rapid growth is outstripping many organizations’ ability to understand how much they should spend and where they should invest,” the report notes.
Agentic systems mark a dramatic departure from the chat-based tools that defined the early years of generative AI. Instead of humans steering each interaction, agents:
- Build and update their own context
- Pull information from multiple models
- Trigger chains of actions without waiting for user input
That independence, S&P Global says, turns AI from a conversation partner into a network of autonomous workers. It also radically changes how companies must think about capacity, data and risk.
Whereas a chatbot may rely on a single model and a limited set of inputs, an agentic system orchestrates several at once. In doing so, it consumes far more computing power. The report describes this as a “bursty” workload pattern. In it, agents generate multiple prompts, spawn additional agents and cascade through large data sets in seconds.
That intensity doesn’t just raise costs. It compounds existing weaknesses in infrastructure and security. Any gaps in data governance, model oversight or identity control, S&P Global warns, “can become chasms into which unwary organizations may tumble.”
How companies are approaching security when using agentic AI
Security is one of the most pressing concerns. Agents are, by definition, non-human identities capable of executing actions across systems. That sometimes means broader permission than their human counterparts. Traditional access controls that mirror user credentials, the report argues, are no longer sufficient.
Organizations will need new policies to monitor agent activity, prevent privilege escalation and verify the authenticity of the data sources that agents consume. Emerging standards like Anthropic’s Model Context Protocol (MCP) and Google’s Agent-to-Agent framework (A2A) are early attempts to solve those challenges by creating common ways for agents to exchange information and authenticate data.
S&P Global’s own operations provide a glimpse of how this technology can be put to work responsibly. The firm’s market intelligence unit has redesigned its know-your-customer process around agentic principles, replacing labor-intensive checks with orchestrated AI agents that analyze documents, verify identities, and perform financial assessments in parallel. The system now manages more than 100,000 cases each year for over 250 financial institutions, pulling from more than 600 data sources in multiple languages. S&P Global says the new approach has reduced analyst workloads by about 40% while maintaining what it calls “zero suppression of true positives” in sanctions and compliance screening.
Even with those gains, the report strikes a cautious tone. As agents multiply, so will their environmental footprint. Data-center power consumption is already climbing as AI workloads scale, and agentic systems are likely to accelerate that trend. Sustainability, once a secondary concern in enterprise AI, now moves to the forefront of strategy.
Agentic AI represents the next great leap in automation—software that doesn’t just think but acts. But it also demands architectural reckoning. Enterprises eager to join the race must strengthen their data governance, modernize identity frameworks, and prepare for a world in which machines are not just tools, but collaborators. “Agentic functionality promises huge potential for improvement,” the report concludes, “but it can require a redoubling of AI efforts to achieve.”
Sign up
Sign up for a complimentary subscription to Digital Commerce 360 B2B News. It covers technology and business trends in the growing B2B ecommerce industry. Contact Mark Brohan, senior vice president of B2B and Market Research, at mark@digitalcommerce360.com. Follow him on Twitter @markbrohan. Follow us on LinkedIn, X (formerly Twitter), Facebook and YouTube.
FavoriteThe post Companies rush toward agentic AI, but the infrastructure bill is coming due appeared first on Digital Commerce 360.
from Digital Commerce 360 https://ift.tt/9xka0DA
0 Comments