Market researchers: Many companies will scrap projects with generative AI

Contents

At least 30 percent of projects with generative AI (GenAI) in companies will be discontinued by the end of 2025, according to market researchers from Gartner. The analysts see poor data quality, inadequate risk controls, escalating costs or unclear business value as possible reasons for this.

"After last year's hype, executives are impatiently waiting for GenAI investments to pay off, but companies are struggling to prove and realize the value. As the scale of initiatives increases, the financial burden of developing and deploying GenAI models is becoming more tangible," said Gartner analyst Rita Sallam.

Productive in the office in two years?

The considerable investment in GenAI to increase production is also difficult to justify if it cannot be directly translated into financial benefits. The financial outlay varies considerably depending on the use case, i.e. whether you get coding assistants off the shelf or whether you need your own LLM after all. There are no one-size-fits-all solutions and the costs are not as predictable as with other technologies, explained Sallam.

A Gartner survey of around 800 company managers from companies that were early adopters of the technology points to potential: on average, the respondents spoke of a 15.8 percent increase in turnover, cost savings of 15.2 percent and a 22.6 percent increase in productivity. However, the effects are often not immediately recognizable and only become apparent over time, Sallam qualified. According to another Gartner forecast, the use of AI in the office could reach a productive level in around two years, after the valley of disillusionment with the hyped technology has been passed.

The long march to monetization

And it is not only business customers, but also providers who are investing massive amounts of money in AI chips and data centers, who are asking themselves when this will pay off. According to analysts from the rating agency S&P, the cloud giants Microsoft, Alphabet and Meta alone have increased their capital expenditure by around 60 percent year-on-year.

In contrast, the S&P analysts see a rather cautious introduction of the technology in companies, which are still in the process of sorting through the multitude of models and finding use cases. "We believe that these factors indicate that the path to monetization and maturity is longer than previously expected," according to a recent study by the rating agency on the tech industry.

Too little investment, the greater risk?

S&P forecasts that global AI spending will grow by more than 20 percent by 2028. It is then expected to account for around 14% of total global IT spending, compared to 6% in 2023.

There is at least plenty of room for improvement when it comes to attracting business customers: as The Register writes, citing figures from the US Bureau of Census in March, only around 5.4 percent of companies in the USA have reported using AI so far. The statisticians expect this figure to rise to 6.6 percent by the fall of this year.

Meanwhile, the leaders of the tech industry are practicing expectation management and calling for patience. "The risk of underinvestment is dramatically greater than the risk of overinvestment," said Alphabet CEO Sundar Pichai at the recent presentation of the company's financial statements. It remains to be seen whether this bet pays off.

(axk)