AI stock correction reflects a more rational market reassessment

Tech-heavy indices like the Nasdaq and South Korea’s Kospi have seen big selloffs in recent days, suggesting that optimism around artificial intelligence-related (AI-related) stocks is fading. The Nasdaq is down around 9 per cent from recent highs. AI-related stocks have seen massive bull runs in the past two years. Investors seem to be taking some profits off the table and reviewing prospects more soberly. Most semiconductor stocks are still up by over 100 per cent from 12 months ago. Caution and profit-taking after massive gains are a rational and healthy process. However, there have also been massive corrections. SpaceX, for example, has seen a selloff, which has sliced 50 per cent off its peak value in just one month. Semiconductor players like AMD, Taiwan Semiconductor Manufacturing Company, SK Hynix and Samsung have lost between 15 per cent and 45 per cent of their respective values. Nvidia, which was the world’s most valuable listed company for a while, has seen a selloff, which pulled its share price down by 17 per cent. (Apple has overtaken Nvidia to become the world’s most valuable company.) 

AI is transformative. People expect it to not only generate gargantuan profits but also to alter the way society functions. It is considered so foundational in impact that it has been compared to the usage of fire or the invention of writing. Trillions have been pumped into the sector. But some investors are now reviewing financial models to apply basic valuation criteria. By traditional metrics, AI stocks may be unbelievably overvalued. AI may eventually yield huge profits. But right now, most AI names like SpaceX, OpenAI, and Anthropic are burning through mountains of cash. AI-related infra stocks such as Nvidia, Hynix, AMD and the hyperscalers running AI-driven data centres are making profits. But they are ultimately dependent on demand from a buildout in a loss-making sector, which is sucking in capital without generating concrete returns. 

Expectations are moderating because investors see losses mounting quarter on quarter while the timelines to profits stretch. Chipmakers and data centres may find their revenues shrinking if demand becomes unsustainable. It’s like a gold rush where the makers of mining equipment are dependent on the production of the precious metal to sustain demand. If gold is not mined in sufficient quantity, the demand for shovels eventually evaporates. This is also the quarterly reporting season. Google and Tesla plunged briefly because management in both committed to continue spending billions on AI despite losses in that segment. Meta, Microsoft and Amazon are expected to declare results in the coming days. Investors will be examining their respective AI exposures with a microscope. 

Another trigger for bearishness is reported breakthroughs in semiconductor design and manufacturing processes by a company in China. If this slices the cost of chips, it will eat into the revenues and profits of Nvidia and others. In that case,  companies such as Meta, Alphabet, Open AI, Anthropic, and Oracle, which have committed hundreds of billions to current chip technology, may find themselves in deep waters as cheaper technology comes into focus. There is also increasing opposition to data centres, with many governments banning data-centre construction due to the vast requirements of water and energy. In addition, there are fears of AI gutting employment, leading to a fall in consumption demand. The AI bubble may not have burst. But the recent profit-booking should lead to more rational evaluations of the pros and cons of the technology. This could also lead to a rotation of capital to a country like India, which is not seen as a big AI play.