KPMG’s Global Tech survey indicates that corporate investments in artificial intelligence can sometimes yield returns of up to twice the initial outlay, yet a large share of AI initiatives never reach live production. The study interviewed 2,500 managers across 27 countries about corporate AI adoption.
Returns and realities
Kórász Tamás, the newly appointed head of KPMG’s advisory business in Hungary, told Portfolio that while some implementations have produced revenue equal to twice the investment within a year (when taking both one‑time implementation and ongoing operating costs into account), multiple studies show a less optimistic picture overall. A Gartner study found that only about 40% of started AI projects reach production, and another prominent study concluded that only one in twenty AI projects is truly successful.
What drives success?
According to Kórász, success hinges on selecting the right business problem for AI. AI creates the most value where solutions rely on pattern recognition and learning rather than deterministic algorithmic rules. Typical areas of effective AI use include fraud detection in banking, customer service automation (for example in contact centers), recommendation systems on streaming platforms, and personalized product recommendations in consumer sectors. By contrast, attempts to use AI to paper over deficiencies in data systems or operational processes tend to fail.
Actual returns depend on the chosen use case, data availability and quality, operating costs, and how the solution is implemented.
Security and risks
On cybersecurity, Kórász emphasized the human factor as the primary vulnerability: people remain the weakest link, and AI increases the risk of deception because high‑quality synthetic voice and video can now be produced. He also pointed to the rise of AI agents—digital representatives that act on users’ behalf—and stressed the need to define their powers, authorizations and controls.
Another major concern is that AI usage often moves large amounts of data into cloud‑based and distributed environments, raising questions about how AI tools are permitted to access corporate data, trade secrets or patents. At the same time, some AI tools are already used to run vulnerability scans in IT systems, making certain weaknesses easier to detect with automated tools.
Impact on the workforce
KPMG’s report estimates that AI could free up roughly 30% of employees’ time in an average office role within the next two years. Kórász noted this figure varies widely by role: repetitive tasks are most affected, while jobs requiring creativity or complex decision‑making are less so. He cited the example of the shared services sector, where the share of highly repetitive roles fell by 12% over the past five years as automation and AI were introduced.
Kórász expects some job losses, but he believes reskilling, training and organizational redesign can enable many employees to move into higher‑value tasks. He also expects entirely new job types to emerge.
Hungary’s AI readiness
Kórász referenced domestic survey data: a 2024 Központi Statisztikai Hivatal (KSH) survey estimated about 8% of companies were using AI in live production, while a smaller‑sample GKI survey in 2025 reported AI use at 32% of firms. KPMG’s observations point to rapid growth in AI adoption, especially among large and multinational companies, and he noted firms that build AI into their business models can pull ahead of competitors. Even small firms can achieve significant results—digital marketing is an example where small companies can leverage AI successfully.
He argued the principal barrier in Hungary is not technology but mindset: many organizations did not previously manage their data assets deliberately. Successful AI deployment requires sufficient quantities of high‑quality, accessible data, data structuring and governance processes—tasks that often demand significant resources and have thus lagged behind.
Practical advice for SMEs
For small and medium‑sized enterprises starting an AI strategy, Kórász recommended focusing on clear, practical use cases such as automated processing of customer requests, preparation of quotations, and automated answers to frequent questions. Before choosing AI, companies should verify data availability and quality, reassess and adapt processes and documentation, prepare staff through targeted training, and allow room for experimentation because first attempts may not yield the best outcome.
He also stressed the importance of governance: AI use must be properly regulated, controlled and where necessary technologically supervised to protect corporate data assets and prevent sensitive information from reaching unauthorized parties or competitors.
Publication notes
The article’s publication was supported by KPMG. The Hungarian text production was assisted by the Alrite online dictation and video subtitling application optimized for Hungarian. Portfolio’s AI & Digital Transformation conference, scheduled for November 26, was mentioned in the original piece as a related event.
Tags: employee, SME, cybersecurity, artificial intelligence, digitalization, return on investment, KPMG Hungary, AI, data assets, investment
