SINGAPORE — Asian investors have grown wary of the sky-high valuations in artificial intelligence stocks. They still want exposure. But they prefer companies that supply the picks and shovels rather than bet on which applications will win.
Temasek aims to lift its allocation to AI companies to as much as 15% over the next five years. That marks a sharp rise from the current 6%. The Singapore state investor already holds stakes in Anthropic and OpenAI. Yet its chief investment officer stresses balance. “You want to ride that trend,” Rohit Sipahimalani said at the Reuters NEXT Asia event. “But the equally big issue is disruption because of AI to many other businesses. We’ve increased our exposure to businesses that are more around hard assets, which are likely to be less disrupted by AI.”
He pointed to the full value chain. Some segments show froth. Others generate real cash flows. “You’ve got to look at the entire value chain,” Sipahimalani added. “We try to play across the entire spectrum.” (Reuters via Yahoo Finance)
Stephanie Hui takes a blunter view. The head of private and growth equity for Asia-Pacific at Goldman Sachs Asset Management admits uncertainty about future winners. “I am not smart enough to tell you today which applications are going to be winning. It’s way too early.” Her firm has instead backed simpler plays. Those include a specialist in liquid cooling and operators of data centers. “We are not going for the front end at this moment,” she said. “We are going for the simple stuff that facilitates an end proxy for AI adoption.”
Global markets hit records on AI enthusiasm. Yet skepticism runs deep here. Investors question whether profit growth can continue at this pace. They wonder if massive infrastructure outlays will deliver adequate returns. Fear of another speculative bubble lingers. Sharp selloffs in semiconductor names have only sharpened that concern.
Fred Hu, chairman of China’s Primavera Capital Group, voiced the worry plainly. “I’m a big believer in the AI revolution but as valuations keep going up, as more and more capital goes into AI… it begs the question, how much is enough.” Satoshi Ueyama of Bain Capital Japan agreed opportunities exist. Still, AI infrastructure needs real end-users to justify the spending. “AI is real but at the same time there’s no denying some parts of the markets are over-excited. Not all AI investment is going to be successful at this stage.” (AOL)
Hardware Dominance Gives Asia Structural Edge
Asia already commands the physical backbone of AI. The region, including Japan, produces roughly 72% of the world’s semiconductors and 95% of the most advanced chips required for AI accelerators such as GPUs. Taiwan, South Korea, Japan and China lead the charge. South Korean firms hold near-monopoly positions in high-bandwidth memory. Japanese companies dominate advanced chemicals and manufacturing equipment. (State Street Global Advisors)
That dominance matters. Western hyperscalers plan to spend $697 billion in 2026 and $873 billion in 2027 on AI infrastructure. Much of that capital flows straight to Asian suppliers. South Korea’s tech giants have responded in force. Samsung Electronics and SK Hynix together plan to invest more than $500 billion in new fabrication plants as part of a national push that could mobilize $880 billion in private capital overall. (Yahoo Finance UK)
Investors have taken notice. Some now hunt opportunities further downstream. Taiwan’s Hon Hai Precision Industry, Quanta Computer and MediaTek sit on watch lists. So do suppliers such as Samsung Electro-Mechanics and Japan’s Ibiden. These firms provide components, advanced packaging, cooling systems, servers and power equipment. Energy names like HD Hyundai and Daewoo also draw interest as data-center electricity demand surges.
Productivity gains reinforce the case. Factories in Taiwan optimize yields with AI. South Korean plants deploy robots at densities five times the global average. In Japan, specialized software writes programs in 15 minutes instead of 16 hours. Singapore reports efficiency jumps of 40% to 70%. Foxconn in China posted a 102% efficiency increase in one deployment. Such results suggest Asia can extract more output from existing capital. That contrasts with the heavy spending seen in the United States.
Demographics add urgency. Japan and South Korea face labor shortages and low unemployment around 2-3%. AI helps offset aging workforces. China and India, meanwhile, grapple with youth unemployment rates above 20%. Retraining programs and targeted incentives aim to channel that talent into tech. Policy support varies but tilts favorable. India offers up to 21-year tax holidays for data centers. Japan and South Korea provide R&D credits ranging from 15% to 50%. These measures make Asian assets look undervalued on price-to-earnings multiples compared with U.S. AI leaders.
But risks remain. Geopolitical tensions could squeeze margins. Over-exuberance in certain segments may lead to losses. And the pace of adoption still depends on tangible returns for end users. So far, the data-center buildout continues. KKR led a consortium that agreed to buy Singapore’s ST Telemedia Global Data Centres in a deal valued at $10.9 billion earlier this year. The transaction underscores how private capital chases the physical side of AI. (The Wall Street Journal)
Temasek’s Sipahimalani captured the prevailing mood. Ride the trend. Hedge the disruption. Focus on hard assets with durable cash flows. Goldman Sachs’ Hui put it more simply. Pick the straightforward enablers. Skip the beauty contest over applications. For now, that formula appears to guide large Asian portfolios.
Markets may keep rewarding the narrative. Valuations could stretch further. Yet these investors have drawn a line. They want participation without blind faith. Hard assets. Real cash flows. Resilience amid disruption. The AI boom has many chapters. Asian capital is positioning for the ones that last.
WebProNews is an iEntry Publication