Finance

Memory makers poised to capitalise on AI inference shift

A Motley Fool analysis published on 18 May 2026 argues that the transition of artificial intelligence workloads from training to inference will drive significant demand for memory and storage, positioning chip manufacturers ahead of designers like Broadcom and Intel.

Author
Owen Mercer
Markets and Finance Editor
Published
Draft
Source: Yahoo Finance · original
The AI Inference Supercycle Is Here. These 2 Stocks Will Be the Biggest Winners of This Megatrend (Hint: It's Not Broadcom or Intel)
Micron and SanDisk identified as primary beneficiaries as data centre workloads pivot from training to inference

A shift in artificial intelligence infrastructure spending is underway, with workloads moving from the training of large language models to inference, where trained models are deployed in real-world applications. According to a Motley Fool analysis published on 18 May 2026, this transition is driving substantial demand for memory and storage components, identifying Micron Technology and SanDisk as the primary beneficiaries of the megatrend.

The report cites Deloitte estimates suggesting that inference workloads will account for two-thirds of AI data centre computing power in 2026, an increase from 50% in 2025. While chip designers such as Broadcom and Intel have seen healthy demand for their inference-focused processors, the analysis argues that these companies rely on memory chips to ensure their accelerators perform at capacity. Consequently, memory manufacturers are positioned as the biggest winners of the AI inference era.

Western Digital CEO Irving Tan highlighted the scale of this change during the company’s April earnings call, stating that data generation is at an inflection point as AI workloads extend to large-scale inferencing. The shift increases the volume of data generated, necessitating greater storage capacity and faster data movement between processors and storage to handle the surge in inference requests across data centres, edge devices, and consumer electronics.

Market data supports the bullish outlook for the sector. Gartner predicts the memory industry’s revenue will increase 2.9 times to reach $633.3 billion in 2026, significantly outpacing the 17% growth anticipated for the non-memory semiconductor market. The report notes a severe supply shortage in memory chips, with industry participants warning that demand may outstrip supply until at least next year, likely leading to further price increases in DRAM and NAND flash.

Despite Micron rising 639% and SanDisk gaining nearly 3,400% over the past year, the analysis suggests both stocks remain undervalued. Micron trades at a forward earnings multiple of 7.6 and SanDisk at 24, both below the Nasdaq-100’s multiple of 26.2. Goldman Sachs predicts a 12% increase in S&P 500 earnings per share in 2026, with the report claiming Micron and SanDisk are on track to surpass that growth rate.

Continue reading

More from Finance

Read next: Broadcom shares slip as investors await higher AI chip guidance
Read next: Wall Street AI trade stalls as Broadcom guidance triggers semiconductor sell-off
Read next: Wall Street rebounds as investors return to semiconductor stocks