MV Weekly Market Flash: The...
Like any good complex organism, the AI narrative is splitting into multiple versions of itself, each reacting in different ways to the daily flow of information that feeds its life support systems. Time was when this was a simple, one-celled story. Buy AI! The collective wisdom of the market came up with a catchy name for the trade – the Magnificent Seven, mega-cap companies close enough to this emergent technology to be considered viable proxies. We were always a bit dubious about the logic underpinning the Mag 7. Nvidia – sure, its graphic processing units are essential for powering the large language models that put generative AI capabilities at the fingertips of the human user base. Microsoft, Alphabet and Amazon are the hyperscalers, supplying cloud computing space for the models to run on. We saw a less compelling case for the other three names in the Mag Seven – Apple, Tesla and Meta – to be fundamental cogs in the AI story. But hey, whatever – the market needed a go-to trade, and these are the tech industry’s leading behemoths, so why not?
Memory Is Not What It Used to Be
There were at least three mutating sub-narratives among the Mag Seven during the Thursday trading session this week. After the market close on Wednesday, memory chipmaker Micron released a blowout earnings report showing a 15-times surge in quarterly profits and a year-on-year revenue growth rate of 345 percent. These memory chips, long regarded as among the least sexy bits of tech architecture, are essential parts of the AI infrastructure value chain, and they are in hot, hot demand. So much demand that Micron’s gross profit margin more than doubled from a year ago to around 85 percent. That’s a level more befitting a luxury goods maker than a semiconductor shop.
Good news for Micron, not so good for anyone who has to buy these memory chips. Like, say, Apple, which announced some sizable price increases for its iPads and Macbooks, specifically citing the cost pressures arising from memory chips. Or Microsoft, which, let us remember, also sells products alongside its newer, jazzier business line of cloud computing. Or, let’s be honest, any mega-cap tech company whose sky-high capital expenditure outlays have increasingly been drawing analysts’ scrutiny.
Buy This, Sell That
As investors consider reapportioning their AI investments into new hot-demand names like Micron or South Korea’s SK Hynix – the latter seemingly single-handedly powering the Korean Kospi stock index to triple-digit gains this year – the question arises as to where the funds for these new momentum-chasing investments are going to come from. Mr. Market seems to have an answer – the Mag Seven! So another sub-story that may be going on here is selling pressure on stocks like Nvidia and Amazon as investors redirect funds out of those companies to be on what they think is the right side of the memory trade.
But there is more to this “source of funds” story than memory. Anthropic, which many observers now believe is the leading AI model company, is due to go public sometime in the second half of this year (Anthropic’s main competitor, OpenAI, m ay also go public in this time frame but has expressed some hesitation recently, possibly due in part to the recent rocky post-IPO price path of SpaceX). There’s the pure AI play, a company without non-AI legacy baggage like Microsoft or Amazon. Funds will be needed for these investments as well.
Anthropic’s looming presence is not just a funding redirection story, but also a manifestation of a talent war. Yet another sub-narrative this week was the announcement by Alphabet of some key personnel departures, including members of some of its most high-profile AI projects who have decamped for Anthropic and OpenAI. The competition for talent in the AI space is intense, and Alphabet’s losses in this area are seen as key factors in the decline of about seven percent in the company’s stock this week.
A Pandora’s Box of Open Source
But there are yet more complexities to the story for investors weighing the merits of pure play investments in Anthropic and/or OpenAI. China’s DeepSeek briefly knocked established AI names for six last year when it launched what appeared to be a competing platform on par with ChatGPT but much more cost-effective. Now DeepSeek is ramping up its latest models and benefitting in part from cheaper energy sources powering its data centers in Inner Mongolia. Other model developers in China and Japan are touting the benefits of open source systems, which could ultimately throw a wrench into the business models and price projections of the established US players.
All of which is to say that the AI story is vastly more complex than it was a year ago, and chances are that it will be more complex still a year from now. The complexity is breeding uncertainty, and uncertainty is showing up in the very volatile day-to-day price movements in this sector. The fundamentals remain strong when we consider the traditional metrics of growth, profitability and asset quality. But we can safely say that the days of simply buying the Magnificent Seven and buckling in for the ride are over.