EPAM thought AI was a rare chance. It turned out to be more complicated

The new challenge for the company was not brought by geopolitics or macroeconomics (there were no changes there — the war in the Middle East is still delaying decision-making), but by AI. It turned out that artificial intelligence is changing the structure of demand faster than EPAM can replace the falling demand for traditional services with new AI orders.

North American sales strategy is not coping

In the second quarter, EPAM generated over $160 million from AI-native solutions, which is 11% of total revenue. The new segment is growing rapidly, but the old one is shrinking even faster.

"Somewhere in July, we started to notice that clients are prioritizing AI budgets, reallocating spending from tasks related to traditional services — such as manual testing, UX, JavaScript development, frontend. They are shifting to modernization, and this transition is happening faster than new volumes of work, which should replace previous ones, are ramping up. As a result, we have a growth gap," explained Balázs Fejes, CEO of the company, about the essence of the problem.

This is especially noticeable in the American region, which accounts for 57% of EPAM's revenue. The increase in financial services there was offset by a decline in the software and high-tech sectors, resulting in a revenue growth of only 0.5% year-on-year. EPAM admits that its own sales strategy in North America is not performing at the required level.

"We are seeing that our clients increasingly want to see project feasibility justifications, commercial proposals, and not just engineering solutions, as it was in the past," Fejes explained.

New large deals with existing clients are being postponed, and revenue from them will not start flowing in until the first half of 2027.

Here's another vivid quote from Fejes, describing the essence of the problem: clients don't just want to integrate AI into their business, but to get a return from it.

"This quarter, every client meeting brought up the same question: how to benefit from AI and get a positive return on investment? My answer is: it's a complex question for which AI doesn't offer a simple answer. The reason is that AI does not reduce complexity; on the contrary, it increases it at all levels: talents, architecture, processes, governance, and models. Coding is automated, development is not. The better AI writes code, the more important solution development becomes in the final stage and successful deployment."

The EMEA region (Europe, Middle East, and Africa), which accounts for 41%, on the contrary, leads in growth rates (+10.9% year-on-year). This is largely due to EPAM having already transformed its marketing and sales methods there. The main contribution to growth in this region came from financial services, tourism, consumer goods, and energy.

Money goes not to engineers, but to tokens

The problem is exacerbated by the fact that customers are now forced to divide their budget between EPAM services and AI infrastructure. According to the CEO, clients (primarily from the portfolio of SaaS companies) are redirecting their spending to tokens and graphic processors. This means AI is not only an assistant to engineers and a new area for orders, but also a direct competitor to the company for the client's budget.

And it seems this is a universal problem, not just for EPAM. At least, when asked by an investor whether North American clients are not going to more diversified competitors, Balázs Fejes confidently answers "no".

The conclusion is that clients' budgets are being consumed not by other companies, but by AI itself.

"We haven't seen actual demand really shift to our competitors," says Fejes. "We see some spending shifting to tokens, which we can also offer, but most of our clients buy them directly or allocate funds to infrastructure or GPUs. I think within our market entry concept with larger deals, we want to reinvest the saved funds that our clients will receive through AI-based services into developing their own business. So we are not losing to competitors, if that's what you're asking."

Interestingly, while at the May investor call Fejes eagerly spoke about tokenomics with an emphasis on the cost/sources of token procurement and "dark factories" as an attractive offering for clients, these new words were not heard in the August conversation with investors.

Staff grew, its utilization did too

Meanwhile, contrary to the trend of replacing employees with artificial intelligence, EPAM's own staff only increased in 2026.

At the end of the second quarter, the company employed more than 56,650 production specialists (+150 people compared to the first quarter), and the total number of employees exceeded 62,850 (+100 people compared to the first quarter). Compared to the second quarter of 2025, the production staff grew by 1.5%.

The resource utilization rate also increased – to 78.3% – both compared to the previous quarter (+1.3 percentage points) and year-on-year (+0.2 percentage points). This likely means fewer people were on the bench (waiting for work. — NN) than in previous periods.

Exchange rate differences played negatively

At least 10,000 EPAM employees are in India.

The Indian theme arose in connection with an investor's question about margin sustainability and the impact of currency fluctuations — likely due to the weakening of the rupee in 2026. EPAM Senior Vice President Jason Peterson assured that the company does not benefit much from currency fluctuations. As Devby explains, firstly, Indian employees are already cheap; secondly, in more expensive locations, local currencies are, on the contrary, strong.

"While we have a significant workforce in India, India's share of our overall costs is not as large as many of our competitors'. So we still have very significant costs in countries like Poland, Hungary, Mexico, and even Colombia, where currencies have strengthened. And therefore, currency fluctuations have not actually made a significant contribution. In fact, they even turned out to be somewhat negative," says Peterson.

Quite recently, EPAM management stated that the gap between rapidly evolving AI capabilities and enterprises' ability to reliably implement them into production would ensure the largest technological investments in human history, and this opportunity should be seized. A few months later, they admit that everything is much more complicated.

• 1
• 0
• 0
• 2
• 1
• 1
• 0
• 0