CLOSE LOOK
Advanced microchips- what’s the big story?

The ‘Magnificent Seven’ big tech stocks continue to dominate the S&P 500 index. Apple, Microsoft, Alphabet, Amazon, Nvidia, Tesla and Meta now represent around 30% of the index. The standout performer among them has been Nvidia, the superchip manufacturer, whose share price has more than tripled so far this year. What has propelled the stratospheric rise of a manufacturer of semiconductor chips to a market cap above $1 trillion?

Let’s first place Nvidia into its sector context. The strong performance of tech this year is based on hopes for a new driver of future profits. Each of the Magnificent Seven is forecast to benefit to a greater or lesser degree from the transformative potential of artificial intelligence or AI. More specifically generative AI, which can create text, images and code and which powers large language models such as ChatGPT.

Nvidia has outperformed the pack as its growth forecasts are the most directly linked to AI. The company engineers the superchips which power AI, adding value with its software expertise. But every one of the high-end 5 nanometre microchips required by Nvidia is supplied by Taiwanese manufacturer TSMC. For all the hype, AI is still a small part of the chip market, perhaps only 5%, while 90% of TSMC’s chip production is used in consumer electronics and cars.

The superchips demanded by generative AI are thousands of times faster than these general purpose chips. As ever greater processing speed is demanded, more and more superchips will be required. When Elon Musk announced the launch of his own generative AI platform earlier this year, as part of the X app, he took pains to assure the market that he had stockpiled the necessary Nvidia superchips. Without them further development of generative AI would be held back.

As often seen in the microchip industry, production capacity has been ramped up to meet a spike in demand. Governments in the US and Europe have recognised the vulnerability implicit in overseas suppliers and invested billions in subsidising new capacity at home. But building a new fabrication plant takes time and can cost up to $20 billion. In the meantime, the pattern of demand might change. A question remains in more cautious investors’ minds. Will the bright prospects for generative AI allow chipmakers to escape the boom/bust cycle this time around?