A paradigm shift to the 4th industrial revolution is fueling investment in big data, artificial intelligence, and the Internet of things (IoT). As a result, the semiconductor industry, the most important hardware component in the development of these new technologies, entered an uncharted area this year.

In 2017, the global semiconductor market is expected to grow 21% year-on-year to $408.6 billion. Therefore, the Philadelphia Semiconductor Index rose by 43 percent this year, surpassing the 16% growth rate of the S&P 500.

Memory semiconductors are at the center of the company’s record semiconductor growth. The marketsize of the memory semiconductor industry is expected to show record growth this year. Moreover, the memory semiconductor market is expected to run to $123.5 billion this year, which is 61% higher than last year.

The paradigm shift to the Fourth Industrial Revolution will probably be not a game that will end in a year or two. This is because leading companies in the Fourth Industrial Revolution, such as Google, Microsoft, Amazon, Facebook, Alibaba and Tencent, are expected to steadily expand research and development activities for and investment in data centers, big data, cloud and machine learning.

Only a year ago, AlphaGo knocked humankind for a loop by crushing the best go player, Lee Sedol. However, only one year later, an upgraded AlphaGo won by a landslide (100 to 0) in a go game match with the original AlphaGo.

This means that performances in the fields of big data and machine learning including artificial intelligence in the next two to three years can give rise to gaps among global companies in the next 20-30 years. Slowing investment in big data and artificial intelligence technologies right now can mean giving up on the future.

With Moore’s Law reaching its limitations, the growth of semiconductor supply has significantly slowed down, more than ever before. Moore’s Law that the line width is reduced to a certain level each year and, in other words, bit supply increases as many times as the square of an annually reduced line width. However, as Moore’s Law is practically limited, semiconductor technology is seeking its development in a direction to increase the degree of integration through 3D stacking rather than chip size reduction.

However, increasing the number of layers means that bit supply will stay at a linear increase, which means an increase in accordance with the number of increased layers, not the square of the number of layers. In other words, it should be remembered that the supply growth rate of semiconductors is significantly slowing down compared to the past in 20-nano processes.

Demand for memory semiconductors became less volatile as price fuctuations declined and supply elasticity dipped as supply grew closer to technical limitations. In conclusion, it is necessary to revise viewpoints on supply and demand cycles of memory semiconductors. If memory semiconductors are understood as a capex-driven industry as was in the past, it can be said that they entered a proportionreduction stage.

However, even if the capex expanded, supply growth is failing to support it and demand is expected to remain stable despite high prices. Accordingly, stable supply and demand situations for memory semiconductors are likely to continue in 2018 against some concerns.

DRAM demand in 2018 is expected to grow to 13.86 billion gigabytes, up 21% year-on-year, despite burdens from rising prices. On the other hand, DRAM supply in 2018 is expected to run to 13.69 billion gigabytes, a 21% increase, despite a surge in CAPEX.

As a result, in 2018, the DRAM supply / demand ratio is expected to remain steady at 0.988, a small increase or a tight level as was the case this year. In 2018, the DRAM market is expected to reach $90.9 billion, up 26% from $ 72.2 billion in 2017.

NAND flash supply hinges on NAND flash makers’ 3D chip yields. However, unlike DRAMs, which have very low price elasticity, demand for NAND flashes is likely to increase significantly if prices fall.

In other words, a change in supply can be a very important variable in changing demand. So, it can be said that it is quite a challenge to forecast demand and supply. However, our forecast for NAND supply/demand is a 37% increase in demand and a 40% increase in supply, which means that supply will increase slightly more than demand.

In 2018, total demand for NAND flashes is expected to reach 235.7 billion gigabytes while total supply remains at 238.2 billion gigabytes. Their ASP per bit is expected to slide about 12 percent next year. In terms of chips, the NAND flash market is expected to grow 23% to $58.1 billion in 2018 from $47.5 billion in 2017.

Accordingly, Samsung Electronics and SK Hynix, which represent the Korean semiconductor industry, are expected to have their banner year in 2018.

This year, Samsung Electronics dethroned Intel, which had enjoyed the No. 1 ranking in global semiconductor sales for 25 years, to become the world’s top semiconductor maker, The securities industry expects its 2018 earnings to reach 280 trillion won in sales, 69 trillion won in operating profit, and 51.7 trillion won in net profit.

SK Hynix’s 2018 earnings are expected to hit 38 trillion won in sales, 18.3 trillion won in operating profit and 14.7 trillion won in net profit. In the global semiconductor industry, its sales and operating profits are expected to stand just below those of Samsung Electronics, Intel and TSMC, but its market capitalization is quite low, standing out of the top 10.

korean-electronics.com | Blog Magazine of korean electronics, brands and Goods