Many businesses are confused because AI does not help increase revenue

Le Hau, director of an import-export company in Da Nang, said he has been applying AI for more than half a year. With 30 employees, he deploys artificial intelligence tools for the purpose of looking up orders, planning purchases, inventory, production and shipping reservations, in addition to tracking market trends and economic fluctuations to forecast import and export needs. He bought enterprise version AI tools, sent 5 employees to learn AI operations classes, and then instructed the remaining employees.

“Spending hundreds of millions on AI, but the level of effectiveness after half a year of implementation is limited, only reaching 20-30% of requirements, mainly in querying and handling the ‘edges’ of work,” Mr. Hau said. “To be more precise, it stops at the support level and cannot be trusted to handle a specific stage as expected.”

Phuong Thao, manager of a fashion store chain in Ho Chi Minh City, also uses AI chatbots to quickly respond to requests from customers. However, she found that the number of orders was “not much different” compared to when she had not used AI, while it took a lot of effort to hire trainers and operators to “answer as naturally as possible”. She said that in the future, AI will still be applied, but as an additional solution instead of cutting staff.

The generative AI craze has been going on for three years, and a series of companies quickly seized the opportunity, trying to integrate artificial intelligence into as many products as possible. In fact, after the period 2023-2024, 2025 is considered to be the year AI is strongly applied in real life.

However, many studies show that applying AI in businesses is still ineffective. Survey data of 4,454 CEOs, released by auditing and consulting company PwC at the World Economic Forum 2026 (WEF 2026) in Davos on January 19-23, shows that the actual implementation of AI is not happening as expected. Of these, only 12% of businesses achieved benefits by both increasing revenue and cutting costs thanks to AI. Up to 55% of business leaders said they have not seen any benefits after investing in AI. 12% noted that costs soared but revenue remained stagnant, while 1% admitted to being in a bad situation when costs increased but profits decreased.

Mr. Mohamed Kande, Global Chairman of PwC, assessed that the ineffectiveness stems from businesses not being prepared in terms of data and implementation roadmap. “Only 14% of employees use generative AI every day. The gap between expectations and the actual capabilities of the human resources team is a major barrier,” Mr. Kande commented.

Previously, according to a survey from Forrester Research with 1,576 executives and published last month, only 15% of respondents acknowledged that the company improved profit margins thanks to AI. Meanwhile, consulting firm BCG said that only 5% of 1,250 CEOs participating in its survey from May to mid-July saw the widespread value that AI brings. They believe generative AI will eventually transform business, but are reconsidering the pace at which it is happening.

Forrester Research predicts that this year, companies will delay 25% of planned AI spending by another year. “Technology companies are expecting, even pretending that things will transform quickly thanks to AI, but the reality is not that fast,” Forrester analyst Brian Hopkins told Reuters.

In Vietnam, at the AWS Cloud Day event in September 2025, Amazon Web Services (AWS) said that 18% of Vietnamese companies are implementing AI, equivalent to nearly 170,000 businesses, an increase of 39% over the same period last year. In 2024 alone, 47,000 businesses will start applying this technology, meaning an average of 5 more businesses using AI every hour.

AWS’s report shows that 74% still mainly use AI for basic tasks such as optimizing operations and streamlining processes, only 17% reach the intermediate stage and 9% fully transform – making AI a core element in product development and shaping business models.

According to Dr. Le Duy Tan, co-founder of AIoT Lab VN, International University – Vietnam National University, Ho Chi Minh City, to properly assess the level of AI application in businesses, the world’s leading consulting organizations often use the AI ​​Maturity Model assessment framework, helping businesses clearly see their position in the application of artificial intelligence.

Specifically, the evaluation framework has 5 levels including Awareness (curiosity) – AI stops at personal attention; Active (trial by fire) – the stage where the business spends a small budget to test AI in specific departments; Operational (real combat) – turning AI into an official working tool; Systemic (synchronous) – AI goes deep into businesses, transforming from an operational support tool to a decision support tool for leaders; and Transformational – AI becomes the DNA of the business (AI-First).

Based on recently published data, Dr. Tan assesses that Vietnam is in the initial stage of universalization – where large barriers in skills and investment costs cause most businesses to choose the safe solution of testing simple tasks before daring to invest in strategic changes.

According to him, there are many reasons leading to this situation. Many companies try to compete with new technologies on old and fragmented technical platforms, leading to infrastructure that is not capable of handling complex AI tasks when scaling up. In addition, the situation of “flying in the fog” regarding financial efficiency, i.e. pouring money according to trends without specific measurements, makes it difficult for projects to escape the “testing” phase to enter actual profitable operations.

Mr. Bung Tran, co-founder of AI Edu – a fully authorized unit of Google for Education, chose the reference framework AIRI (AI Readiness Index – AI Readiness Index) with four levels: AI Unaware (not aware) – businesses stay out of the game, considering AI as a fiction and has nothing to do with their work; AI Aware (AI awareness) – can start using as an individual; AI Ready (ready) – businesses have prepared their data infrastructure, human resources skills and management processes, and started pilot projects with clear goals; and AI Competent (proficient) – when AI integrates into the flow of work, creating a core competitive advantage.

“Regarding the current situation in Vietnam, most businesses are in the transition zone from AI Unaware to AI Aware. The atmosphere is very exciting, everyone is talking about AI, but the internal strength of data and systems thinking is still very thin. We are excited but lacking readiness,” Mr. Bung commented.

According to him, many Vietnamese businesses think “like technology tourists”, that is, they just go to “AI land” and look around, buy a few souvenirs (buy retail AI tools), take check-in photos (do PR), but have no intention of “settling down”.

“They approach AI with the mindset of playing around. If it’s good then use it, if it’s difficult then give up,” this expert emphasized. “Leading to businesses spending money but only receiving fragmentary tools instead of a solid digital foundation. When the excitement passes, what’s left are expired software accounts and piles of junk data that cannot be reused. AI is not a tourist destination, it is a new living environment that businesses are forced to adapt to in order to survive.”

Citing statistics from research firm Gartner showing that 85% of AI projects fail worldwide, Mr. Bung said that the cause does not lie in technology, but in the lack of fundamental elements: lack of clear strategy with AI; Data lacks availability even though it is the most important “main dish” for artificial intelligence; The “illusion” of AI, seeing this as a miracle, installed and done instead of having to be included in an optimal workflow.

Besides, the risk of handing over “sensitive” data to AI also makes businesses cautious. According to Mr. Tan, in fact this is not an emotional worry, but reflects a clear “security gap” in the management and security capacity of many organizations today, especially in the context of AI increasingly developing towards autonomy and self-decision making.

According to Mr. Bung, in the digital era, data is life. However, being shy does not mean being “secluded”. The solution is a stratified management strategy with a specific regulation, classified into green zones for things the public can share and red zones for business secrets and customer data, requiring maximum confidentiality.

“We have the advantage of being one of the early countries to have a cybersecurity law, a decree on personal data protection as well as the Artificial Intelligence Law – the foundations for having signage to follow. Leaders need to issue clear rules of the game: what employees are allowed to bring to the ‘cloud’ and what is kept absolutely on the ‘ground’,” Mr. Bung said.

Mr. Bung believes that to be successful, businesses need to shift their thinking from “application” to “operating system”, meaning bringing AI deeper into the operating system instead of installing sporadic software. On the leadership side, it is necessary to change the mindset of “know-it-all” (know everything) to “learn-it-all” (self-study and renew knowledge every day), promoting the spirit of “test quickly, make mistakes quickly, fix quickly”.

For Mr. Tan, in order for businesses to no longer hesitate when putting data into AI, the solution does not lie in limiting the use of technology, but in synchronously upgrading security infrastructure, administrative capacity and legal framework. Building control barriers, strengthening encryption, investing in security human resources, and perfecting institutions are mandatory conditions, helping businesses move from fear to proactive trust and effective exploitation of AI in the long term. In addition, the legal framework’s role in transparency and stability will help promote digital trust. When there are clear regulations on data protection, liability and AI governance, businesses will have a “safe zone” to boldly test and expand the application of new technology, instead of worrying about legal risks.

Experts say that 2026 will be a pivotal year for AI to prove its real value. If it cannot soon transform from promises to concrete profits and community benefits, the wave of investment in AI may face a chain collapse like the dot-com bubble period.