Memory Seen as a Destination for AI Investment, with Focus on Earnings Durability
After reviewing big tech earnings, Nomura maintained its view that AI investment is unlikely to stop. Its analysis said internal rates of return on AI projects are improving and that, given the risk that AI could cannibalize existing businesses, big tech companies will have little choice but to continue capital spending over the coming years. This could lead to increased funding not only from operating cash flow but also from debt and share issuance.
Memory semiconductor makers were identified as a destination for this investment capital. Nomura said cash was accumulating rapidly at memory companies, while their shares were trading at low valuations that reflected assumptions that earnings had peaked. Samsung Electronics and SK Hynix were valued at a discount of 85–90% to the average for four major big tech companies. This was interpreted as meaning that current prices already reflected concerns that earnings next year could be nearly halved.
Price-to-earnings ratios in the memory sector were cited at around two to three times, while Micron was also mentioned at below six times. By contrast, TSMC was valued in the high teens, which analysts said reflected differences in how long companies had demonstrated the durability of their business cycles. The market is seeking further confirmation of how long earnings and cash flow at memory makers can continue.
Uncertainty over whether profits will grow in proportion to big tech’s substantial investment, and when free-cash-flow deficits will narrow, has contributed to share-price volatility despite continued spending. Operating cash flow and free cash flow at major big tech companies deteriorated as investment expanded. Views also differed on when AI investment would translate into actual returns, with Goldman Sachs citing 2028 as the point when cash flow would improve.