Why Foundries Welcome the Artificial Intelligence Revolution



Third-party foundries

In the previous part of this series, we saw that the AI (artificial intelligence) revolution could boost demand for data processors, memory, wireless communications, and sensors. Foundries such as Taiwan Semiconductor Manufacturing (or TSMC) (TSM) and Samsung (SSNLF) could benefit since they manufacture various types of chips ranging from processors to communication chips to GPUs (graphics processing units).

Article continues below advertisement

These two foundries manufacture chips for fabless companies Nvidia (NVDA) and Qualcomm (QCOM). TSMC is the sole manufacturer of Apple’s (AAPL) A11 chips that have AI capabilities. Samsung could also benefit from the growing demand for non-volatile memory chips and OLEDs (organic light emitting diodes).

Semiconductor equipment manufacturers

The complexity of AI chips could lead to manufacturing innovation since these chips require larger dimensions and specialized processors. That could boost the demand for new equipment across memory makers and foundries.

The AI deployment is at its early stages, but SME (semiconductor manufacturing equipment) suppliers Applied Materials (AMAT), Lam Research (LRCX), and KLA-Tencor (KLAC) witnessed strong growth in new orders.

Management for these three companies in their September 2017 quarterly earnings calls said they expect to see strong growth in 2018 as memory makers ramp up capacity, as China (FXI) builds new memory plants, and as foundries upgrade their facilities to build next-generation specialized processors.

SME market outlook

Gartner expects the SME market to expand 26% to $46.6 billion in 2017. It also expects semiconductor capital spending to reach $69.9 billion in 2017 and increase to $75.8 billion by 2020.

Applied Materials president Gary Dickerson stated that the cyclicality of the SME business could decline since AI could expand the semiconductor end market beyond PCs (personal computers) and mobile to IoT (Internet of Things) and automotive.

As for manufacturing innovation, 2018 could see Intel transition to its next-generation 10 nm (nanometer) process node and foundries transition to a 7 nm node. Next, we’ll see how the transition to a smaller node could impact investors.


More From Market Realist