AI safety warnings rattle tech stocks: Should Singapore investors worry?
SINGAPORE – Tech stocks in the US and Singapore have come under pressure after warnings from prominent artificial intelligence officials last week revived fears that the rapid development of increasingly powerful models could pose an existential threat to humanity.
The shift in sentiment followed warnings from researchers on Sept 10 that AI could threaten humanity within the decade, prompting rare calls for restraint from industry leaders. Anthropic chief executive Dario Amodei proposed slowing AI development in an essay endorsed by OpenAI founder Sam Altman and SpaceX chief executive Elon Musk.
Fears about the consequences of AI development are not new, but the latest warnings, which put a numerical probability on the risk of human extinction, have raised fresh concerns about the pace and safety of the technology’s development.
Analysts told The Straits Times that although the initial sell-off is unlikely to significantly undermine the performance of AI-related stocks in the near term, AI safety could become an increasingly important consideration for investors over the longer term.
“Even as the training pace slows, inference demand remains tight, and competition among companies and nations make it hard for the industry to collectively hit pause. So a single statement is unlikely to change AI’s trajectory,” said Michael Chen, general manager of the Hong Kong office of Noah ARK, a Singapore-based wealth management firm.
The latest shift in sentiment could ultimately spur new capabilities and safeguards in AI products, helping companies win market share and wider acceptance, said Gene Lai, equity research analyst at OCBC.
Chen added that while AI development costs could rise further, opportunities for firms providing cybersecurity, safety testing and AI-agent monitoring could also emerge.
Still, investors will be watching more closely to see whether safety concerns lead to independent evaluations, delayed model launches and increased spending on monitoring and cybersecurity, he said.
Cusson Leung, chief investment officer of financial services firm KGI Asia, added that security, social acceptance and regulatory constraints have “received too little weight” in forecasts for AI hardware demand.
The recent calls for an AI development slowdown “challenges the assumption that demand for AI semiconductors will rise indefinitely, and a sharp reset in market expectations could trigger a de-rating of semiconductor companies”, Leung said.
Impact on Singapore stocks
While Singapore technology stocks along the semiconductor value chain initially fell last week, analysts said they are less exposed to shifts in AI sentiment and valuation-driven corrections than their Wall Street counterparts.
Although their growth is partly tied to demand from the semiconductor and AI industries, many are not pure AI plays and also serve other sectors, giving them a more diversified earnings base.
For instance, Frencken Group and UMS Integration supply components and modules mainly to front-end equipment manufacturers supporting the semiconductor industry. However, they also serve sectors such as automotive, medical technology and life sciences, noted Alfie Yeo, senior research analyst at RHB Singapore.
Meanwhile, CSE Global provides communications, electrification and automation services to customers across sectors such as oil and gas, urban infrastructure, manufacturing and events, besides helping Amazon with the electrification of its data centres, which power the use of AI.
Other SGX stocks linked to the AI cycle include InnoTek, which manufactures components used in Nvidia products.
While AI is a significant growth driver for the company, it accounts for only around 10 per cent to 15 per cent of revenue, noted SAC Capital’s head of research Matthias Chan. With automotive and office automation remaining its key revenue drivers, InnoTek should be able to withstand further volatility in the AI sector, he said.
Still, some companies like AEM Holdings have more concentrated exposure to AI.
AEM provides semiconductor testing equipment and services mainly to chipmakers. In 2026, it disclosed that a new unnamed customer, which outsources chip production to a foundry, had overtaken Intel as its largest revenue contributor in the first half of the year.
Analysts said this concentrated exposure makes AEM among the companies most susceptible to negative AI sentiment and any resulting market sell-off.
But graphics card manufacturer PC Partner, which has expanded into enterprise and industrial AI servers, appeared largely unaffected by the shift in global AI sentiment. Its share price, already on an upward trajectory since the beginning of September, climbed further last week.
KGI analyst Leung said that AI exposure offers PC Partner good growth opportunities, but warned that the company’s resilience will depend on whether it can protect the profitability of its existing business and turn its server push into sustainable earnings and cash flow.
Other companies such as AvePoint, which provides data protection and AI management services, saw its share price jump last week amid the increased focus on AI governance.
What to consider amid market volatility
While the latest AI safety concerns have added to the unease, a potential slowdown in model development does not necessarily mean less use of existing models, with demand for inference remaining strong, said OCBC’s Lai.
For instance, OpenAI paused new Pro 20X subscriptions after overwhelming demand for its Astra model strained its infrastructure, reflecting capacity constraints rather than deterioration in sentiment.
But he noted that the sector will continue to see “continued headline-driven volatility”, and the recent sell-off illustrates how quickly expectations can adjust when the assumptions behind AI investments are questioned.
“Therefore, it is important to focus on the financial fundamentals and valuations of AI bottleneck stocks, as well as the supply-demand dynamics for compute,” said Lai, noting that AI hardware, for example, remains a “bottleneck trade” because of supply constraints.
One key fundamental that investors should consider is the primary revenue drivers for AI companies – whether it is training, inference or safety governance – as these segments have different sensitivities to a potential slowing of frontier models, said Noah ARK’s Chen.
Investors should also assess whether company valuations depend on continual increases in capital expenditure forecasts and whether they would hold up if AI spending were delayed while costs continued to rise.
He cited other important factors such as a company’s customer concentration, its portfolio concentration and liquidity, and the prevailing regulatory environment.
“High volatility in the AI era may not stem from the industry trend disappearing, it may instead come from the market redistributing who bears the cost of safety and who captures the revenue from it.”
Singapore tech companies would also have to secure more orders on the AI value chain and demonstrate consistent quarter-on-quarter or year-on-year growth to boost investor confidence, so that their valuations can be maintained or increased, noted Jarick Seet, head of small- and mid-caps research at Maybank Securities Singapore.
KGI’s Leung recommended that investors focus on position sizing, concentration and valuation for their portfolios, as owning several AI-related stocks does not necessarily provide meaningful diversification if they all depend on the same customers and infrastructure-spending cycle.
He also cautioned investors to prepare for lower valuations assigned to semiconductor companies should there be a reset in AI hardware-demand expectations.
“I think investors should prepare for that scenario rather than assume that strong long-term demand will protect them from sharp share-price movements. The question is how well an investment holds up when growth falls short of expectations, not just how it performs when everything goes right.”