Search

2023 – A time to look back and to look ahead for the global semiconductor industry 

An article written by Pierre Cambou and Amandine Pizzagalli from Yole Group – As Semicon West 2023 ended few weeks ago, it is timely to reflect on the changes in the international semiconductor industry and look ahead five years to what the future holds.

This period is a time of reflection, coinciding as it does with the publication of Overview of the Semiconductor Devices Industry 2023, a compilation of more than 100 publications of Yole Group and capitalizing on the 25 years of semiconductor industry knowledge it gained since its inception in 1998.

The semiconductor industry was born in the USA in 1947 shortly after WWII, with the invention of the transistor. Since then, it played a significant role during the cold war, the US nuclear weapon program and then the space race, becoming the technology backbone of our “advanced” society. Today, it has become central to our human lives governed by digital interactions, smartphones serving as digital catheters and young people now being part of that species: “digital natives”.

As a new cold war is unraveling between China and the U.S., free markets are no more. The availability of cheap labor abroad has created vulnerabilities within the ever-expanding semiconductor supply chain. The semiconductor industry is no more solely market & technology driven, geopolitics, macro-economics and scarce natural resources are creating a new level of competition environment.

semiconductor-overview-2023-scorecard

Changing business models

Key changes in the last five years include Intel losing its dominant position gained in the 70-80’s to Samsung and TSMC. The emergence of the gigantic Taiwanese open foundry player is in part a U.S. success since it mostly serves U.S. fabless players, such as Apple, Qualcomm, AMD, Broadcom, Nvidia, Intel and Amazon.

This new business model is both a blessing and a curse. The U.S. semiconductor device players increased market share up to 53% in 2022 but at the same time have seen a slow erosion of the added value they capture, losing one point every year down to 32% globally in 2022. The entire industry relies almost on a sole player located in Taiwan which is a disputed Chinese territory, and which could be cut off from the rest of the world if a military conflict with mainland China should occur.

Today, there are different efforts happening at once. The first is a U.S. -driven effort of “re” and “friend” shoring semiconductor foundries. At the same time, mainland China is emerging as a significant semiconductor player with its own industry plan worth $143 billion up to 2025.

The third movement is the continuous capital expenditure (capex) to maintain and expand the semiconductor industry’s growth beyond the current 6.4% CAGR of the last 30 years. This requires about 20% of its revenue into new foundries every year.

Pierre Cambou Principal Analyst in the CTO Office of Yole Intelligence, part of Yole Group
The question is not how much is to be invested over the next five years – ~$800 billion – but where these investments will be made. Therefore to steer foundry investment, the U.S. Government and the EU Commission have passed CHIPS Acts valued at $53 billion and $47 billion respectively.

Worldwide activity

Yole has been monitoring announcements made in the last two years and estimate there is ~$800 billion of fab related investments across the globe for the next three to five years, which relates directly to the 20% mentioned earlier.

$205 billion of these are planned in the U.S. with the $40 billion TSMC fab construction in Arizona, which began in December 2022. U.S. players represent 60% of investment in the U.S, and also include projects by Intel, Texas Instruments, Micron and Wolfspeed.

Direct Foreign Investment (DFI) by Samsung, SK Hynix, NXP, Bosch and X-FAB in addition to TSMC, make up the remaining 40%.

In the EU there are $61 billion worth of investments planned. These include a $20 billion Intel fab construction in Magdeburg Germany, with a packaging and test facility in Poland, and a Global Foundries-STMicroelectronics fab in France, with EU IDM players such as Infineon, Bosch and ASML also active and a partnership between TSMC, Bosch, NXP and Infineon under discussion. However, DFI at 85% dominates, leaving EU players representing just 15%.

Calculating announcements made elsewhere, primarily Korea and Taiwan, the U.S. will receive 26% of the total investments and the EU 8% of the total.

U.S. fabless device players represent ~28% market share of semiconductor device today but the EU falls short of its ambition to control 20% of world capacity. It has shifted from the intent of “doubling the footprint” to “securing the existing business”.

Korea represents 30% of the announced investments which would be double its current market share of ~15%. Taiwan represents 15% of today’s announced investments which is also its current market share. Mainland China’s investments will be overshadowed by the Government’s $143 billion plan and represents 18% of announced investments which would imply tripling its current market share.

Semiconductor wafer production

Semiconductor wafer production worldwide is expected to grow 30% by 2028 and reach a total capacity of roughly 12,000kWpm 12’’eq. The semiconductor device industry is expected to face a -7% decrease year on year in 2023 following the peak of $573 billion revenues in 2022. The fab announcements will sustain the +6.4% long term semiconductor growth observed since the industry began but Yole consider that a +4.5% CAGR for the next five years is more realistic.

Will spending solve anything?

It is unlikely that the semiconductor industry will solve its vulnerability with regards to Taiwan quickly. In addition to capital spending, production is affected by human capital and environmental resources. This last is always overlooked, but water and power stability have been scarce lately. It is unlikely that the investments listed will go through smoothly and quickly but at least we can say the investment numbers do add up.

By 2028 the U.S. will have gained more control over semiconductor production, either on its own soil, through “re” shoring, “friend” shoring, mainly in the EU through high subsidies to U.S. players there, but also in Korea, letting Samsung expand quicker than TSMC therefore creating more competition.

The EU will mainly gain secure supply chains but will not own them. Taiwan’s TSMC will be less of a duopoly with Korea’s Samsung, and Intel should become the third man in an open foundry market also populated by UMC, SK Hynix and Global Foundries among others.

Another milestone for 2028 is that Yole Group will celebrate its 30th anniversary; it will continue to provide updates on the state of the semiconductor industry. To assess how the industry’s investment plans come together – stay tuned!

About the author

Pierre Cambou, MSc, MBA, is Principal Analyst in the CTO Office of Yole Intelligence, part of Yole Group. His mission is dedicated to market & technology analyses of the semiconductor industry.

At Yole, Pierre has authored more than 20 Yole Market & Technology reports and 16 Quarterly Monitors. Acknowledged as an expert in the semiconductor industry, Pierre is regularly interviewed and quoted by leading international media.

Previously, Pierre held several positions at Thomson TCS, which became Atmel Grenoble (France) in 2001 and e2v Semiconductors in 2006. In 2012, he founded a semiconductor startup that was later acquired by Wx Solutions.

He holds an Engineering degree from Université de Technologie de Compiègne (France) and a Master of Science from Virginia Tech. (VA, USA). Pierre also graduated with an MBA from Grenoble Ecole de Management (France).

Related products

up