AI Spending to Run Two More Years, Lifting Memory and Power

Artificial intelligence spending will keep expanding over the next one to two years, widening the pool of beneficiaries to memory chips, power infrastructure and data center efficiency, according to a global asset manager. As AI adoption spreads and growth opportunities multiply, investors need to be selective about companies with proven technology and profitability.

"We expect AI-related spending to remain at a substantial level over the next 12 to 24 months," Ray Chiu, chief investment officer of thematic equities at AllianceBernstein, said in an interview with The Seoul Economic Daily on the 5th. "The scope of investment will broaden from training clusters to inference, memory, storage, networking, power and data center efficiency." AB managed about 1,404 trillion won ($1 trillion) in assets as of the end of June.

Chiu said the current AI boom differs from the dot-com bubble because the spending is being led by companies with profitability and scale. The overall level of investment is not excessive, he said, given that it already supports cloud services, advertising and productivity gains. He was more cautious about the possibility of overinvestment at some individual companies. "High valuations can be justified if there is long-term growth and profitability that exceeds the expectations already priced into the shares," he said. "We prefer companies with a sustainable competitive advantage and a clear monetization strategy."

He stressed that Korean chipmakers should be approached by the same standard. "Korean semiconductor companies occupy a strategically important position in memory, particularly high-bandwidth memory and DRAM," Chiu said. Because the supply base is concentrated among a small number of companies and the products are essential to accelerated computing, there is potential on both demand and pricing, he said.

The firm is also looking for opportunities from AI's spread in power and cooling. Companies that resolve constraints on power supply as data centers expand stand to benefit, Chiu said. Citing grid equipment and cables, power management technology, liquid cooling and on-site generation, he said: "We prefer companies that address bottlenecks in grid capacity, power density and the time it takes to deliver power."

On the other hand, he pointed to energy supply shocks as a variable that threatens the growth outlook. If supply instability persists, inflation and tighter monetary policy could follow, raising funding costs for an AI industry that requires large-scale investment. He said investors should monitor capital spending by large cloud companies, AI usage, and chip orders and prices together, watching for whether weakness across multiple indicators coincides with poor monetization.

As an investment principle for responding to those risks, he emphasized cash flow and diversification. The AB International Technology Portfolio holds 60 to 90 stocks, centered on growth companies with established business models and cash flow. "As AI adoption broadens, opportunities may also emerge in areas such as health care and industrials," Chiu said. "Investors need to strike a balance between investing in powerful AI trends and maintaining broad portfolio diversification."