• A new US Semiconductor Industry Association (SIA) report is big on optimism for home-grown chip technology
• Massive subsidies are being pumped into the production of  sub-10nm semiconductors 
• Forecast suggests that by 2032 US will command 28% of the world’s most advanced microchip market, while China will have just 2%
• However the report seriously underestimates the still formidable challenges of Chinese competition

Telecom and IT research reports come in a variety of flavours, running the gamut from the sickly sweet to the bracingly astringent. Others are like chewing through cardboard leavened with a dab of treacle. Pessimism has its place, as does optimism, but sometimes reports are so stuffed to the gills with effervescent buoyancy that they float away into the realms of wishful thinking. 

One report that is leaning towards the latter is Emerging Resilience in the Semiconductor Supply Chain from the US Semiconductor Industry (SIA), which predicts that, by 2032, the US will be manufacturing 28% of the world’s most advanced semiconductor chips (right now its produces none of them) while China will account for just 2%. 

History may well show that China’s contribution as calculated might just turn out to be a gross underestimate.

The SIA report was written in conjunction with the Boston Consulting Group and takes as its starting point the impact of the US Chips and Science Act of 2022. Here “Chips” stands for Creating Helpful Incentives to Produce Semiconductors for America and relates directly to “the creation of incentives to produce semiconductors in the United States.” The federal legislation, signed into law by President Joe Biden, authorises, in total, some $280bn in new funding and incentives to increase domestic US research into, and the manufacture of, state-of-the-art semiconductors. 

Close to $40bn has been dedicated to chip manufacture on American soil, together with 25% investment tax credits for the costs of manufacturing equipment. A further $13bn is set aside for semiconductor R&D and training the workforce that will operate the new fab production facilities. The money will be needed. Last month it was estimated that, to date, incentives provided via the Chips Act have resulted in serious proposals for somewhere between 25 and 50 potential projects valued at between $160bn to $200bn and creating 25,000 to 45,000 new jobs. In itself that is very good but there is sure to be be (a sometimes literal) degree of difficulty relating to the lack of skilled workers because it is expected that, when new fabs are built and operational, 40% of employees will need to be very skilled personnel with at least 2-year technical degree quantifications. To make matters worse, 60% of more senior staff will need to have 4-year undergraduate engineering degrees while many more posts will require master’s-level or doctorate-level qualifications. That’s a long-term and very expensive educational proposition.

Meanwhile, and simultaneously, $174bn has been dedicated to telecom development and the national ecosystem of public sector research into areas of science and technology, including biotechnology, experimental physics, human spaceflight, materials science, quantum computing and systems security. And that’s not all, the other (and equal) intent of the Chips Act is to act as a counterforce to China’s semiconductor efforts and industry whilst greatly strengthening the resilience of the US semiconductor supply chain by massively reducing any reliance on overseas technologies. 

The government of the People’s Republic of China (PRC), smarting and angry at its companies being unceremoniously banned from any meaningful participation in the US technology markets, now has a national “processor autonomy” plan which, it is claimed will see it circumvent US export prohibitions and reach technological self-reliance by 2030, whilst not only keeping pace with the US and other countries but actually overtaking them. The battleground will be in the 10nm node arena of advanced chip fabrication, a fight that, the SIA contends, the US will quickly and easily win. Time will tell just how optimistic that assessment is. 

Just a couple of years ago, almost all advanced sub-10nm semiconductors were made in either South Korea (31%) or Taiwan (69%) but, the report says, thanks to the Chips Act, this will change and, by 2032, the 10th anniversary year of the passing of the legislation, the US will have grown its domestic capacity to make semiconductors by 203% – the biggest chip manufacturing growth rate in the world. 

The SIA report also forecasts that the US will increase its share of advanced logic (below 10nm) manufacturing sector from 0% two years ago to 28% of global capacity by 2032. In other words it will produce well in excess of a quarter of all such chips in the world despite having had a market share of precisely zero in 2022.

Not only that but the SIA reckons the US will carve out 28% of total global chip sector capital expenditures by 2032, running a close second to Taiwan with 31%. According to the report, were it not for the Chips Act, the US figure would have been just 9% of global capex by 2032. The report, which is based on input from “various sources”, also has it that South Korea’s share of the advanced chip market will fall to 9% over the same timeframe. Quite why is not explained and the figures seem to presuppose that both countries will sit back, cease competing and make no advances of their own over the next 8 years. The same applies to Europe and Japan whose advanced semiconductor market shares, according to the report, will sit at 6% and 5% respectively. As for China, we are told that it will command just 2% of the advanced semiconductors market, although it’s just possible it might be a bit more than that by 2032. India, the world’s fastest-growing economy and with wide technological ambitions has its own plans for producing advanced chips, but the sub-continent isn’t mentioned at all. Something of an oversight perhaps?

The thing is that the SIA’s contention that China’s chip manufacturing sector will not be a major force by 2032 is predicated not based on the PRC’s own performance but on how many other countries are able to boost their own production and thus take China down by many a peg. In the decade from 2012 to 2022, China managed a 365% increase in chip manufacturing capacity but, according to the report, the figure for the 10 years between 2022 to 2032 it will grow by a much reduced 86% while the world average for the period will be 106%. 

At 86%, China will be further down the worldwide league table than Europe at 124%, South Korea at 129%, and the US at 203%. Meanwhile, Chinese technology companies, such as Huawei, are already ramping up the scale of their development of 7nm chips and working on building their own EUV (extreme ultraviolet machines) to etch their own high-quality sub-10mn chips in enormous numbers. Maybe the SIA analysis is partially based on the fact that the current generation of smart vehicles, smart fridges, smart kettles, smart toothbrushes and smart arses don’t need to be fitted with sub-10nm chips – but it’s a fair bet they will be by 2032, perhaps even to some captains of the US technology industry.

US leads the way in throwing money at the China problem 

In the end, it’ll all be about money. The SIA says China will spend some $156 bn on its semiconductor industry between now and 2032 while Europe will spend $154bn. These figures fall way short of the planned $300bn set to be invested in South Korea, the $646 to be spent by the US and the $716bn earmarked by Taiwan (always presuming it is still an independent nation in 8 years’ time. 

With the US ploughing vast sums into subsidies for semiconductor companies, the nation isn’t going to fall by the chipside wayside, even if China remains a very significant competitor. To give some indication, here’s a less-than-comprehensive list of initiatives that are already underway or planned. 

In November 2021, Samsung announced it would build a $17bn semiconductor factory to begin operations in the second half this year. It is the largest foreign direct investment ever in the state of Texas. In January 2022, Intel announced an initial $20bn investment that will create 3,000 jobs, the largest investment in Ohio’s history, with plans to grow to that investment to $100bn in eight fabrication plants. In May of that same year, Texas Instruments began construction of a 300mm semiconductor wafer fabrication plant, in Sherman, Texas announcing a $30bn investment that should create 3,000 jobs.

In December 2022, TSMC of Taiwan announced the opening of a second chip plant in Arizona. It brought TSMC’s investments to $40bn. February 2023 saw Texas Instruments invest $11bn investment in a new 300mm wafer fab in Lehi, Utah. In April 2023, Bosch announced the acquisition of TSI Semiconductors and invested $1.5bn in upgrades to further enable the production of silicon carbide chips at the TSI plant in Roseville, California. In June 2023, French company Mersen, a subsidiary of Le Carbone Lorraine, said it would spend $81m on an expansion project in Bay City and Greenville, Michigan as a direct reaction to Michigan’s state implementation of the Chips Act. In March this year, Intelit had been paid $8.5bn from the Chips Act to build four new advanced semiconductor fabs in Chandler, Arizona and New Albany, Ohio and to upgrade plants in Hillsboro, Oregon and Rio Rancho, New Mexico. Last month, TSMC announced it had landed $6.6bn in US subsidies to build a third fabrication facility in Arizona. This will host a 2nm semiconductor process with construction planned for early in 2028. Also in April 2024, Samsung said it received $6.4bn in Chips Act grants to invest in additional capacity at its new site in Taylor, Texas and at an established fab in Austin. And so it goes…

- Martyn Warwick, Editor in Chief, TelecomTV