Analysts look deeply into the earnings details of companies, looking for numbers that can subtly affect a company’s stock performance.
Barclays analysts on Tuesday flagged depreciation expenses related to artificial intelligence that may affect three Magnificent 7 companies’ results in the next year or so.
The companies affected are Amazon.com (AMZN) , Google-parent Alphabet (GOOGL) and Facebook-parent Meta Platforms (META) .
Related: Analyst resets Palantir stock price target ahead of Q2 earnings
Also, the Barclay’s group, led by Ross Sandler, believe earnings will be affected as head counts will start to “normalize” as the companies invest in new AI equipment and facilities in 2025 and 2026. That may require more staff to make the systems work.
Barclays has an overweight rating on Amazon but lowered its price target to $200 from $210.
Barclays also has overweight ratings on Alphabet and Meta with a $200 target on Alphabet and a $520 target on Facebook’s parent company.
The analysts are still very bullish on AI generally and on big-cap tech stocks heading into second-quarter earnings.
Mark Zuckerberg, chief executive officer of Meta Platforms Inc., is among a group of companies investing heavily in artificial intelligence.
AI investment costs will explode
The depreciation question is the larger issue and could be worth as much as 5%-to-10% of reported earnings per share.
Related: Analysts revamp AMD stock price target on AI deal
Depreciation is a non-cash charge designed to account for the working life of an asset over a specific period. Let’s say a server has a 10-year life. If it costs $100,000 to buy and install, company writes off $10,000 a year as an expense.
It affects earnings this way: If a company has $2 in earnings before depreciation and the depreciation works out to 20 cents a share, the net will fall to $1.80. That can depress a stock price.
According to Seeking Alpha, the Barclay’s group believes much of Wall Street is using a standard formula to estimate depreciation for companies — typically a percentage of revenue.
But the huge expenses required to buy, operate and equip AI-related servers and equipment will exceed the typical formulas, Barclays believes.
The basic price for an Nvidia (NVDA) H200 graphical processing unit is about $40,000. The cost for an entire AI installation that includes multiple serves with multiple GPUs can be several hundred thousand dollars or more. And there may be multiple servers working in a data center.
The Mag 7’s showing on Tuesday
Amazon was up slightly at $193.02 on Tuesday. Alphabet was off 1.4% at $183.92, while Meta was down 1.3% to $489.79.
More AI Stocks:
Analyst adjusts Nvidia stock rating on valuationAnalyst revises Facebook parent stock price target in AI arms raceGoogle falling behind climate goals thanks to AI ramp up
In total, three Magnificent 7 stocks were higher on the day, led by Tesla (TSLA) , up 1.6% to $256.56. In addition to Amazon, Apple (AAPL) was up 0.2% to $234.82.
In addition to Alphabet and Meta, Microsoft (MSFT) slipped nearly 1% to $449.52.
Nvidia dropped 1.6% to $126.36. It’s off 6.8% from its peak close of $135.58 on June 18 but is still up 155% this year.
Related: Veteran fund manager sees world of pain coming for stocks
