Can AI's 'iPhone moment' boost these flagging semiconductor stocks?

By Wallace Witkowski

The earnings outlook for equipment maker Lam Research and chip fabricator TSMC is weak in the near term but brighter farther out, analysts say, as data centers expand capacity to support growth from generative-AI products like OpenAI's ChatGPT

After semiconductor companies struggled to manufacture enough chips to address a pandemic-era spike in demand, they now face a supply glut after that demand plummeted. But Wall Street analysts suggest that a boom in demand for artificial-intelligence models can turn that direction around again.

Some of the semiconductor industry's biggest names -- like manufacturing-equipment maker Lam Research Corp. (LRCX) and third-party fabricator Taiwan Semiconductor Manufacturing Co. (2330.TW) -- will kick off semiconductor earnings this week. Analysts hope to hear that the need for more computing power in the coming years to support AI models like OpenAI's ChatGPT will lead to a new boom for chip-equipment makers and fabricators, but they caution investors that it could still take time for those effects to become apparent.

Lam, which reports after the bell Wednesday, makes the very complicated machinery used by foundries like TSMC to etch billions of transistors onto silicon wafers to make chips designed by the likes of Nvidia and Apple Inc. (AAPL), while chip makers like Intel Corp. (INTC) forge their own silicon.

Earlier in the day, ASML Holding NV (ASML.AE) reported a surge in profit and revenue for the quarter that topped analyst estimates, but the prospect that there's still a lot of inventory working its way through the channel pressured shares.

Evercore ISI analyst C.J. Muse said he expects some weakness in the current quarter for Lam but that 2024 looks more promising. Muse, who had forecast a beat and raise from ASML, considers the stock a "top pick" as a "likely relative outperformer amid worries of Samsung/TSMC capex cuts."

"For Lam, we are updating our model to reflect inline [March quarter] results (though MKS issues add volatility) followed by a modest [June quarter] miss (though highly deferred revs dependent)," Muse said, added that he expects Lam to reiterate that the first half of the year is tracking better than the second half.

As for a key positive, Lam will likely show the strongest year-over-year growth in 2024 in earnings per share, making the stock "a must own in into memory pricing inflection likely in [the third quarter.]"

Citi Research analyst Laura Chen maintains that AI is having its "iPhone moment" and that Nvidia is reportedly putting in rush orders with TSMC to handle capacity. However, she cautioned that the outlook for the rest of this year could disappoint when TSMC reports Thursday morning.

Read:ChatGPT is called 'an iPhone moment in AI,' but will it make money like the iPhone?

"While we believe TSMC's leadership in advanced nodes and upsides from AI/HPC [high-performance computing] remain intact in the longer term, TSMC may provide lower [second-quarter 2023] outlook, and lower full year guidance in the upcoming earnings call," Chen said.

TSMC earnings are already set for a worst-case scenario, Susquehanna Financial analyst Medhi Hosseini said in a recent upgrade of the stock. TSMC last week reported that its revenue fell from the year-ago quarter for the first time in nearly four years

Last earnings season was all about the inventory questions in hardware: How much did everyone unload, how big a hit did they take, and how were they managing lead times and order expectations? Now the framing is, How many chips will be needed to scale the massive amounts of data being processed in training and inferencing?

The big customers are "hyperscalers," or cloud-service providers, like Microsoft Corp.'s (MSFT) Azure, Amazon.com Inc.'s (AMZN) AWS, Alphabet Inc.'s (GOOGL)(GOOGL) Google Cloud Platform and Oracle Corp.'s (ORCL) Cloud Infrastructure. And those hyperscalers are already training their own AIs, with Microsoft investing billions in OpenAI and ChatGPT and in Google's generative-AI product Bard.

Generative AI, which has gone mainstream this past year with the release of OpenAI's ChatGPT, uses massive amounts of data drawn from a site's application program interface, or API, to train the AI product and train for inference. Google also has a generative-AI product, Bard, and Adobe Inc. (ADBE) recently released an AI product named Firefly, a "co-pilot" technology aimed at helping create content. Meta Platforms Inc. (META) has its own eponymous product, Meta AI, while Amazon.com Inc.'s (AMZN) AWS has released free machine-learning tools for users

Last quarter, Lam cut its workforce by 7% while increasing R&D spending as a big drop in memory-chip demand dinged its outlook. Analysts polled by FactSet expect Lam to report fiscal third-quarter earnings of $6.52 a share on revenue of $3.82 billion.

Read: How Nvidia plans to fuel the AI surge and a new era of chipmaking

Lam counts TSMC and memory-chip maker Micron Technology Inc. (MU) among its biggest customers, according to FactSet data. Major suppliers for Lam include Ichor Holdings (ICHR), Advanced Energy Industries Inc. (AEIS), Ultra Clean Holdings Inc. (UCTT) and MKS Instruments Inc. (MKSI).

TSMC counts Vanguard International Semiconductor Corp. , Global Unichip Corp. , Entegris Inc. (ENTG) and Tokyo Ohka Kogyo Co. as major suppliers, while Advanced Micro Devices Inc. (AMD), Qualcomm Inc. (QCOM), NXP Semiconductors NV (NXPI) and MediaTek Inc. are major customers. Analysts expect TSMC earnings of $1.22 a share on revenue of $17.02 billion.

Lam shares are down 2% on the week and TSM American depositary receipts (ADRs) are down less than 1%. Over the first calendar quarter, however, Lam shares are up 26.1%, while TSMC has gained 24.9%. Meanwhile, the Dow Jones Industrial Average rose 0.4%, the S&P 500 gained 7%, the tech-heavy Nasdaq Composite rallied 16.8% and the PHLX Semiconductor Index surged 27.6%.

-Wallace Witkowski

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

04-20-23 0754ET

Copyright (c) 2023 Dow Jones & Company, Inc.