On January 20, 2025, Chinese AI startup DeepSeek made waves in the global AI industry. They released a large language model (LLM) called R1, which delivers similar performance to OpenAI’s models at just one-tenth of the cost, dramatically shifting the paradigm of the AI industry. While debates erupted in the West regarding the safety and development methods of Chinese AI models, derivative models based on DeepSeek’s open-source release started spreading, leading to the rise of the “DeepSeek Syndrome.”

Mark Andreessen, a prominent Silicon Valley venture capitalist, called this “AI’s Sputnik moment.” DeepSeek’s achievement, developing the model using only 2,048 GPUs despite US semiconductor export restrictions, has become a symbol of cost innovation in the AI industry.

DeepSeek g토토사이트 확인ned its competitive edge not only by reducing costs but also through efficient model tr토토사이트 확인ning and optimization techniques. What important implications does this shift have for the domestic 토토사이트 확인 market? To discuss this, we spoke with Professor Choi Yong-suk, the President of the Artificial Intelligence Society of the Korean Institute of Information Scientists and Engineers and a faculty member of the School of Computer Science.

The large language models (LLMs) that dominate the existing generative AI ecosystem have models with parameters ranging from several billion to hundreds of billions, offering high accuracy and versatility. OpenAI’s GPT-4 and Meta’s LLaMA are representative examples.

LLMs require a massive scale of graphics processing units (GPUs) and power for both training and inference. It is reported that Meta’s LLaMA 3 used around 16,000 of the latest GPU H100s just for model training.

DeepSeek R1, although it falls under the category of large models with a massive number of parameters, delivers comparable performance with an optimized architecture, offering the same results at a fraction of the cost. Professor Choi emphasized, “The key is that DeepSeek meets both cost reduction and high efficiency, significantly enhancing the practicality and applicability of the AI model.”

The global 토토사이트 확인 industry had previously focused on using expensive, high-performance chips to build high-performance 토토사이트 확인 models. However, the emergence of DeepSeek marked a new trend: creating high-efficiency 토토사이트 확인 models using low-cost, low-performance chips.

Professor Choi noted, “DeepSeek has shaken the existing AI market dominated by US big tech companies, and the US-China AI power struggle has intensified.” He explained, “In the tech field, rapid advancements make it difficult for a single dominant player to remain at the top, and DeepSeek’s emergence is a perfect example of the dynamic nature of the AI industry.”

Countries and smaller companies with limited technical resources are now able to develop high-performance 토토사이트 확인 at low costs, leading to intensified market competition. As a result, big tech companies are shifting from closed models to open-source approaches, making 토토사이트 확인 models accessible to anyone.

Regarding the current state of the domestic AI market, Professor Choi stated, “There are still few successful cases of developing competitive models and commercializing them worldwide.” He added, “Research has mostly focused on localizing and customizing existing large language models (LLMs) or vision models (LVMs) for specific tasks or domains, developing applications or services based on these models.”

After the arrival of DeepSeek, discussions on AI model development strategies have gained momentum in Korea. Professor Choi remarked, “We must continue steady efforts and academic research to develop our own unique models.” He added, “Such research on unique models will ultimately contribute to the advancement of AI applications and services, and these byproducts will have a positive impact on the entire domestic AI ecosystem.”

DeepSeek has proven that low-cost, high-efficiency models are possible in the 토토사이트 확인 industry. While developing independent models in a short time may be challenging, continuous research and experimentation will serve as catalysts for the growth of the 토토사이트 확인 ecosystem.