Nearly 88% global firms now have dedicated AI budgets, says report

Nearly 88 per cent of global enterprises now have dedicated budgets for Artificial Intelligence (AI), with nearly two-thirds of them spending over 15 per cent of their overall tech budgets on AI projects, a new report said on Monday.

This major investment push marks a clear shift from early experimentation with Generative AI to building intelligent, goal-oriented systems known as AI agents, according to data compiled by Nasscom.

Titled ‘Enterprise Experiments with AI Agents - 2025 Global Trends,’ the report provides a comprehensive look at how companies across the globe are preparing for the next phase of AI adoption.

Sangeeta Gupta, Senior Vice President and Chief Strategy Officer at Nasscom, said that enterprises are at a crucial turning point.

“AI agents represent the next evolution of enterprise AI - one that requires philosophical shifts in how we view work, intelligence, and autonomy,” she said.

However, she also emphasised that scaling AI systems responsibly would require strong trust, data readiness, and continuous human oversight.

Based on responses from over 100 companies across 8-9 global regions and more than 10 industries, the study shows how enterprises are moving beyond passive data analysis to more active AI systems that can perform tasks and make decisions with human oversight.

The report shows that businesses are strengthening their AI foundation through investments in GenAI tools, data infrastructure, and flexible processes.

Many companies have already formed specialised AI teams and are working with advanced platforms, upgrading their tech setups to support the deployment of AI agents.

However, despite high awareness of Generative AI, only half of the surveyed companies are fine-tuning large language models (LLMs) or foundation models for their own needs.

One of the biggest highlights of the report is the growing interest in Agentic AI - systems designed to act independently while still being monitored by humans.

About 62 per cent of companies are experimenting with such AI agents, mainly for internal tasks such as IT operations, HR, and finance.

External uses, like customer service, are still limited, with only 31 per cent of enterprises using Agentic AI in those areas.

However, looking ahead, 88 per cent of companies plan to set aside budgets specifically for Agentic AI systems in 2025.

The report also reveals that most companies are being cautious. Around 77 per cent are designing Agentic AI systems with a ‘human-in-the-loop’ model to ensure oversight and adaptability.

Only 46 per cent are testing fully autonomous agents. Manufacturing companies appear to be ahead in adoption, using AI for robotics, quality control, and other operational areas.

When it comes to benefits, companies believe AI agents can help in making faster decisions and responding better to market changes, the report said.

AI TOOL TO DETECT CANCER: Meanwhile, in a significant achievement for identifying neuro-degenerative diseases early, a team of US researchers has developed a new artificial intelligence (AI) tool that helps clinicians detect brain activity patterns linked to nine types of dementia using a single and widely available scan.

The tool, StateViewer, not only helped in early detection but also provided accurate diagnosis - it identified the dementia type in 88 per cent of cases, including Alzheimer’s disease.

It also enabled clinicians to interpret brain scans nearly twice as fast and with up to three times greater accuracy than standard workflows, according to the research, published online in the journal Neurology.

Researchers from the Mayo Clinic trained and tested the AI on more than 3,600 scans, including images from patients with dementia and people without cognitive impairment.

Currently, diagnosing dementia requires cognitive tests, blood draws, imaging, and clinical interviews, and yet, distinguishing conditions such as Alzheimer’s, Lewy body dementia and frontotemporal dementia remains a challenge.

“Every patient who walks into my clinic carries a unique story shaped by the brain’s complexity,” said David Jones, a Mayo Clinic neurologist.

“StateViewer reflects that commitment -- a step toward earlier understanding, more precise treatment, and, one day, changing the course of these diseases,” added Jones, director of the Mayo Clinic Neurology Artificial Intelligence Programme.

The tool analyses a fluorodeoxyglucose positron emission tomography (FDG-PET) scan, which shows how the brain uses glucose for energy. It then compares the scan to a large database of scans from people with confirmed dementia diagnoses and identifies patterns that match specific types, or combinations, of dementia.

While Alzheimer’s affects memory and processing regions, Lewy body dementia involves areas tied to attention and movement. Frontotemporal dementia alters regions responsible for language and behaviour.

Indo-Asian News Service