AI Spending Justified, Hyperscalers Drive Demand

The massive spending on artificial intelligence is not only justified but set to continue its upward trajectory, according to insights shared by Celine Woo, Portfolio Manager at Lazard Asset Management. Speaking on Bloomberg Tech, Woo highlighted the recent earnings reports from hyperscalers, indicating that the hardware and semiconductor sectors will significantly benefit from sustained demand that currently outstrips supply.

Woo noted that the primary catalyst for this growth is the industry's pivot towards inference workloads, a shift that has been confirmed by conversations with industry leaders. While training AI models still demands significant resources, the day-to-day operation and deployment of AI applications through inference are now the key drivers of compute demand.

The full discussion can be found on Bloomberg Technology's YouTube channel.

Hyperscalers Show Strong AI ROI

The recent earnings of major cloud providers, including Alphabet Inc. (NASDAQ:GOOGL), Amazon, and Microsoft, have provided a strong validation of their substantial investments in AI. Woo pointed out that these three companies collectively delivered a 50% growth rate in their cloud segments, a remarkable acceleration compared to just five quarters prior. This performance demonstrates a clear return on investment for their commitment to AI infrastructure and signals continued spending.

When assessing the performance of these tech giants, Woo suggested that Microsoft and Amazon particularly impressed relative to expectations. The ability of these companies to exceed forecasts underscores the tangible benefits flowing from their large-scale AI spending commitments into their core businesses.

Demand Outstrips Supply in AI Hardware

A recurring theme in the AI sector is the persistent imbalance between demand and supply. Woo elaborated that this applies to both cloud capacity and the literal supply of chips powering these advanced systems. While a strong demand environment is positive, it necessitates a synchronized effort across the entire supply chain to ensure timely delivery of components, preventing any potential slowdown in AI adoption.

The supply chain's ability to meet this surging demand is often underestimated. Woo cited the example of the memory space, where long-term agreements are becoming more common, a departure from historical practices. This trend indicates a greater degree of forward visibility and customer commitment, with companies like SK Hynix reporting that customers are increasingly willing to sign five-year deals.

TSMC's Major CapEx Signals Confidence

The commitment from key players in the semiconductor manufacturing chain further bolsters the outlook. Woo highlighted that TSMC, the world's largest foundry, has announced a significant $100 billion capital expenditure plan to expand its capacity, including in the U.S. This substantial investment, made in conjunction with hyperscalers, is not a speculative move but a direct response to clearly visible demand that extends over a five-year horizon. It signifies a collective effort across the supply chain, including memory manufacturers, to align supply with this unprecedented growth in AI demand.