America’s fantasy of home-grown chipmaking

America’s fantasy of home-grown chipmaking

To remain the world’s foremost technological power, the country needs its friends

The Economist, Aug. 21

How low mighty Intel has fallen. Half a century ago the American chipmaker was a byword for the cutting edge; it went on to dominate the market for personal-computer chips and in 2000 briefly became the world’s second-most-valuable company. Yet these days Intel, with a market capitalisation of $100bn, is not even the 15th-most-valuable chip firm, and supplies practically none of the advanced chips used for artificial intelligence (AI). Once an icon of America’s technological and commercial prowess, it has lately been a target for subsidies and protection. As we published this, President Donald Trump was even mulling quasi-nationalisation.

More than ever, semiconductors hold the key to the 21st century. They are increasingly critical for defence; in the ai race between America and China, they could spell the difference between victory and defeat. Even free-traders acknowledge their strategic importance, and worry about the world’s reliance for cutting-edge chips on tsmc and its home of Taiwan, which faces the threat of Chinese invasion. Yet chips also pose a fiendish test for proponents of industrial policy. Their manufacture is a marvel of specialisation, complexity and globalisation. Under those conditions, intervening in markets is prone to fail—as Intel so vividly illustrates.

To see how much can go wrong, consider its woes. Hubris caused the firm to miss both the smartphone and the ai waves, losing out to firms such as Arm, Nvidia and tsmc. Joe Biden’s CHIPS Act, which aimed to spur domestic chipmaking, promised Intel $8bn in grants and up to $12bn in loans. But the company is floundering. A fab in Ohio meant to open this year is now expected to begin operations in the early 2030s. Intel is heavily indebted and generates barely enough cash to keep itself afloat.

 The sums needed to rescue it keep growing. By one estimate Intel will need to invest more than $50bn in the next few years if it is to succeed at making leading-edge chips. Even if the government were to sink that much into the firm, it would have no guarantee of success. The company is said to be struggling with its latest manufacturing process. Its sales are falling and its plight risks becoming even more desperate.

The Biden administration failed with Intel, but Mr Trump could make things worse. He has threatened tariffs on chip imports, and may try to browbeat firms such as Nvidia into using Intel to make semiconductors for them. These measures might buy Intel time but they would be self-defeating for America. Chipmaking is not an end in itself but a critical input America’s tech sector requires to be world-beating. Forcing firms to settle for anything less than the best would blunt their edge.

What should America do? One lesson is not to pin the nation’s hopes on keeping Intel intact. It could sell its fab business to a deep-pocketed investor, such as SoftBank, which has reportedly expressed interest in buying it and this week announced a $2bn investment in Intel. Or it could sell its design arm and pour the proceeds into manufacturing. Intel may fail to catch up with TSMC even then. Either way, the federal government should not throw good money after bad. Taking a stake in Intel would only complicate matters.

That leads to a second lesson: to look beyond Intel and solve other chipmakers’ problems. tsmc is seeking to spread its wings. It is running out of land for giant fabs in Taiwan and its workforce is ageing. It has already pledged to invest $165bn to bring chipmaking to America. A first fab is producing four-nanometre (nm) chips and a second is scheduled to begin making more advanced chips by 2028. Samsung, a South Korean chipmaker that is having more success than Intel, is setting up a fab in Texas. But progress has been slow: Samsung and TSMC have both struggled with a lack of skilled workers and delays in receiving permits.

The last lesson is that, even if domestic chipmaking does make America more resilient, the country cannot shut itself off from the rest of the world. One reason is that the supply chain is highly specialised, with key inputs coming from across the globe, including extreme-ultraviolet lithography machines from the Netherlands and chipmaking tools from Japan. The other is that Taiwan and its security will remain critical. Even by the end of this decade, when tsmc’s third fab in America is due to begin producing 2nm chips, two-thirds of such semiconductors are likely to be made on the island. TSMC’s model is based on innovating at home first, before spreading its advances around the world.

To keep America’s chip supply chains resilient, Mr Trump needs a coherent, thought-through strategy—a tall order for a man who governs by impulse. No wonder he is going in the wrong direction. On Taiwan he has been cavalier, confident that China will not invade on his watch, while failing to offer the island consistent support. His tariffs on all manner of inputs will raise the costs of manufacturing in America; promised duties on chip imports will hurt American customers. He thrives on uncertainty, but chipmakers require stability.

A sensible chip policy would make it attractive to build fabs in America by easing rules over permits and creating programmes to train engineers. Instead of using tariffs as leverage, the government should welcome the imports of machinery and people that support chipmaking. Given the bipartisan consensus on the importance of semiconductors, the administration should seek a policy that has Democratic support—with the promise of continuity from one president to the next.

Economic nationalists should also see the progress of chipmakers in allied countries as a contribution to America’s security. Samsung is aiming to start producing 2nm chips in South Korea later this year. Rapidus, a well-funded chipmaking startup in Japan, is making impressive progress. Both countries have a tradition of manufacturing excellence, and may have a better shot at emulating Taiwan.

The chipmaking industry took decades to evolve. It is built for an age of globalisation. When economic nationalists build their policies on autarky, they are setting themselves a needlessly hard task—if not an impossible one. ■

For subscribers only: to see how we design each week’s cover, sign up to our weekly Cover Story newsletter.