Singapore tech stocks slide as chip sell-off spreads in Asia despite CXMT surge

[SINGAPORE] Technology and semiconductor stocks in Singapore slid on Tuesday (Jul 28) as a global memory chip rout that saw Applied Materials shares slide also spread in Asia.

Frencken , an Applied Materials supplier, took the largest hit and fell as much as 8.5 per cent on Tuesday morning. UMS , also a supplier to Applied Materials, declined as much as 8.1 per cent.

The US-listed company’s shares were down as much as 6.7 per cent during the Monday US trading session, as it was caught in a broader sell-off of global semiconductor equipment makers.

AEM shares dropped as much as 5.6 per cent, while those of CSE Global fell as much as 4.9 per cent.

This followed a 5.8 per cent drop in shares of Dutch chipmaking giant ASML, spurred by The Information’s report that an unnamed state-backed company in China is now capable of the mass-production of immersive deep ultraviolet (DUV) lithography machines.

Such tools are considered to be a central workhorse of the semiconductor industry and are a core revenue stream for ASML, which is estimated to have cornered as much as 90 per cent of the total lithography market.

Fears around a threat to existing DUV makers quickly spread to ASML’s peers around the world, as Japan-based lithography equipment makers also took a hit. Shares of Nikon fell as much as 9.2 per cent, while those of Canon fell as much as 6.3 per cent – the biggest drops for both in more than two months.

In the broader artificial intelligence and memory sectors, Monday’s US trading session saw shares of AMD drop about 7 per cent and Micron fall about 5 per cent.

Nvidia, a key buyer of SK Hynix’s high-bandwidth memory chips, also dropped about 5 per cent after The Wall Street Journal reported ​that the GPU and AI giant ​could provide about US$250 billion as a financial backstop for an OpenAI data-centre project.

SEE ALSO

SK Hynix, Samsung Electronics retreat as chip sell-off weighs on Asian stocks

AI finding twice as many cyber flaws in 2026 as it did in 2025

European shares steady as tech weakness offsets US-Iran optimism

US stocks: Wall Street ends mixed as investors focus on tech earnings

More

That overnight weakness carried into Tuesday for South Korean chip stocks, with Samsung Electronics and SK Hynix retreating as much as 9.5 per cent and 10.9 per cent, respectively.

Investors were likely rotating out of AI amid mounting concerns over financing risks tied to infrastructure spending and intensifying competition from China.

This followed the recent launch of Moonshot AI’s Kimi K3 model, which showed that Chinese-made models are closing the gap on American leaders OpenAI and Anthropic.

The Chinese startup’s AI model was said to be able to outperform all competitors in the field, except Anthropic’s Claude Fable 5 and OpenAI’s GPT-5.6. It released an update on the model weights and technical report of the model on Monday, and is poised for public download.

The Tuesday bloodbath in regional technology and chip stocks was opposite to the surge in CXMT shares on Monday.

The memory maker became the most valuable company in China, beating out giants like Industrial and Commercial Bank of China, CATL and Alibaba, after its shares surged 466 per cent during the company’s Shanghai trading debut on Monday.

James Ooi, market strategist at Tiger Brokers, noted how CXMT’s initial public offering, which was reportedly more than 200 times oversubscribed, highlights “strong investor appetite” for China’s semiconductor localisation theme.

However, other analysts were cautious of the meteoric rise in CXMT shares and theorised the surge may be down to speculative investors, a Reuters report showed.

Re-understanding the winners and losers among AI plays

On a whole, amid intensifying competition, the race is no longer about who builds the best model, but who will capture the economic rent of the system, said Blackrock analysts.

“We think cheaper AI changes the winners, not the investment case,” they wrote in a Monday note.

The analysts explained that the cost of the tech is emerging as a key concern for companies deploying it, as worldwide spending on AI models and platforms is expected to reach US$64 billion in 2026, up 63 per cent from 2025.

At the same time, Chinese models are now processing roughly 23 trillion tokens per week, compared with about 12 trillion tokens for US rivals.

“Together, these trends could erode the pricing power of frontier model developers even as AI adoption accelerates,” BlackRock said.

Companies therefore have a stronger incentive to contain costs through model routing and lower-cost models amid rising enterprise AI bills, with the analysts preferring AI infrastructure over the increasingly competitive model layer.