Semiconductor industry revenue to increase 18% in 2024, Samsung regains number one position

Thanks to the rapid development of the artificial intelligence (AI) field, coupled with the huge demand for graphics cards (GPUs) and neural processing units (NPUs), the semiconductor industry recorded an impressive revenue growth of 18.1% in 2024, which is a significant improvement from the total revenue of 626 billion USD in 2023. This figure is expected to be even higher in 2025, reaching 705 billion USD.

This growth has significantly changed the market share rankings of semiconductor suppliers worldwide. Of the top 25 companies in the industry, 11 companies recorded double-digit revenue growth, while only 8 companies recorded declines. Samsung surpassed Intel to become the number one supplier, while Nvidia rose from 5th to 3rd place. Micron Technology recorded the most impressive jump, from 12th to 6th place in the list of the world's largest semiconductor manufacturers.

Notably, a series of big names such as Intel, Qualcomm, Broadcom, AMD and Apple all dropped in rank in 2024 compared to 2023. In terms of actual revenue growth, SK hynix is the leading name with revenue growth of up to 86%.

Mr. George Brocklehurst, Vice President of market analysis company Gartner, said:

Graphics processing units (GPUs) and AI processors used in data center applications (servers and accelerators) are the key drivers of the chip industry in 2024. Growing demand for AI and generative AI (GenAI) tasks has made the data center the second-largest semiconductor market in 2024, behind smartphones. Data center semiconductor revenue is expected to reach $112 billion in 2024, up from $64.8 billion in 2023. Memory and AI semiconductors will drive near-term growth, with high-bandwidth memory (HBM) expected to account for an increasing share of DRAM revenue, reaching 19.2% in 2025. HBM revenue is estimated to grow 66.3% in 2025, to $19.8 billion.

The future of this market promises some interesting twists. New reasoning models, like the DeepSeek R1, are becoming more accessible and could drive demand for more processing power. However, this demand may not be so great if companies like DeepSeek find ways to optimize the performance of AI models to make them more efficient.