A recent study by Accenture found that firms using advanced AI models such as LLMs and generative AI can increase their revenue by 10 percent, or 2.6 times higher than enterprises which don’t apply the technology

In an era of AI and LLMs, data science and AI are increasingly integrating into the work process. However, applying AI models in business activities is challenging.

Nguyen Van Tuan, CEO of Hyratek, which provides support and AI infrastructure for a martyrs’ photo restoration project, said the demand for equipment systems used in training AI is more than the supply. Buyers have to place orders with suppliers half a year in advance.

AI systems are mostly trained centrally, with very high costs, which is a barrier to the use of AI in enterprises’ operations.

In Vietnam, many businesses are using cloud services to deploy AI models. However, this is costly in the context of large-scale operations and the lack of flexibility in the working process.

Nguyen Van Giap, CEO of Lenovo Vietnam, said at a recent event that in order to apply AI more frequently, enterprises tend to use AI integrated workstations.

Many institutions have shifted to store and develop private LLMs and SMLs as this is considered a solution for security and data training cost reduction.

This not only optimizes the working process, but also helps enterprises’ owners make timely decisions, and at the same time, accelerates innovation in many fields.

With high-performance CPUs and GPUs, workstations are designed to promote the development of AI models, and fine tuning and training on a smaller scale and at a lower cost than on the cloud.

Using data on the premises is not only more secure, but it also allows data scientists to train AI models in a closed and faster loop, thereby reducing the time to get to the final results.

The diversity of LLMs has increased all over the globe. Robert Hallock from Intel said to speed up digital transformation, countries can develop LLMs of their own, including a Vietnamese LLM. Vietnam and China could localize LLMs by incorporating local language factors.

A survey by Finastra found that Vietnam is leading markets in terms of interest in generative AI. Around 91 percent of Vietnamese responded that generative AI can bring positive results.

Many institutions are developing LLMs (large language model) and SLM (small language model) of their own through AI integrated workstations to protect security and limit data training costs.

Trong Dat