Don’t look now, but some of the technology markets are flashing warning signs for the first time in months. Given a mismatch between growth expectations—many of which have been boosted by AI—and what’s really happening in technology investment, the markets could rationalize the enthusiasm of the AI bubble quite quickly.
AI still has enormous potential. And we have likely started off a new cycle of long-term investment to last a decade or more. But the markets have already reflected excessive optimism, and there are many details to be worked out about AI: the costs, the path of return on investment (ROI), and legal issues about data, just to name a few.
NEW YORK, NEW YORK - SEPTEMBER 13: Traders work on the floor of the New York Stock Exchange during ... [+]
Here’s why we might be ready for a reset on expectations: Technology markets, after selling off steeply in 2022, bounced back in 2023 and early 2024, boosted on the hope that a spending spree in AI would form the next leg in the technology bull market. But so far, AI has been an elitist and speculative play. The best results have gone to only a handful of companies—and already several companies have fallen after failing to deliver the AI bump that investors expected.
Based on an analysis of technology spending patterns unfolding for the year, I expect the hype to abate in the near term. While there is no doubt that AI is a long-term technology trend on par with megatrends such as the Internet and mobile booms of the past, share prices as well as startup valuations have already been bid up too far this early in the cycle.
Elite AI Can’t Prop Up the Whole Market
The first problem for the technology markets: The market has been boosted by a handful of “AI elites.”
This AI elite group includes the largest cloud providers and hyperscale operators, such as Amazon Web Services, Microsoft, Google Cloud, Oracle, and Meta, as well as their largest suppliers. AI infrastructure to build GenAI apps is enormously expensive, so only the richest of companies can afford the billions of dollars in investment in new infrastructure.
These cloud providers have invested billions of dollars in hardware and software infrastructure to build AI tools such as large language models (LLMs). Through the process, an elite group of AI startups has enjoyed huge gains (OpenAI, Anthropic, Mistral, etc.).
These AI spending trends have also translated to the largest AI infrastructure suppliers. These include chipmakers such as Nvidia and AMD, which currently enjoy a nice duopoly on AI processors; server suppliers such as Dell and Supermicro; and networking providers such Nvidia and Arista Networks (Nvidia makes entire AI stacks including networking, compute, and software, which makes it such a formidable player in the market). These groups of AI suppliers have seen the largest gain in share price over the past year.
But here’s the dirty secret of AI spending: It hasn’t surged enough to compensate for slowing capital spending growth in other areas of the technology markets.
Enterprise spending has slowed. The service provider market is a mess. Even with AI investment, the largest cloud service providers have actually scaled back their capital spending (capex) over the past year. As you can see below, a Futuriom analysis of capex at the six largest cloud providers shows a decline in 2023 from 2022 levels, despite the large AI investments.
Hyperscale capital spending declined in 2023 from 2022
What happened is the cloud hyperscalers trimmed other budget items to support their AI goals. In addition, they’ve also been shrinking staff. Microsoft, for instance, eliminated over 10,000 jobs in 2023. Amazon laid off 9,000 last year, and Google about 12,000.
Profit Concentration Skewed the Entire S&P 500
Another worrying factor for the technology markets is that concentration of profits among the elite largest tech companies has been a feature of the entire stock market rally.
A commonly watched basket of elite technology companies, the so-called “magnificent seven (Mag7)” of Apple, Amazon, Alphabet, Meta, Microsoft, Nvidia, and Tesla), have provided most of the returns in the S&P 500 stock index. For example, these top seven companies had average returns of 107% in 2023, while the more balanced and broad MSCI USA Index returned only 27%.
The profit share is even worse. The Mag7 provided nearly all the profit growth in the S&P 500 Index. If you strip out the profits provided by the Mag7, the rest of the companies in the S&P 500 had declining profit growth in 2023.
If AI expectations slip, it could bring down this entire group of stocks. Logic holds that because this group has accounted for an outsized portion of overall market returns in the rally over the past year, that effect will work in reverse: It could accelerate the decline in the market.
AI Results Haven’t Come to Bear
Here’s another reason to be concerned: In earnings results reported in the first quarter (for fourth quarter 2023 results), some of the expectations of immediate business results from AI growth simply haven’t come to pass. Let’s look at a few examples:
Adobe. Shares of the creative software suite company fell 10% in March after it failed to deliver growth that investors had expected to be tied to AI-related products. It also faces competition from AI upstarts such as OpenAI, which are developing competitive products. Adobe shares are 25% below their February highs, indicating investor disappointment in living up to the AI hype.
AMD. Chipmaker Advanced Micro Devices, a rival of AI chip giant Nvidia, surged to record highs on the back of new product announcements in March, but since then it has seen a sharp sell-off and is 25% off the highs, even after projecting strong sales for AI processors.
Nvidia. Shares of Nvidia tripled in 2023 and jumped 27% in January on AI enthusiasm, but they have been weaker since its fourth-quarter results and announcements at its recent GTC conference. Currently Nvidia shares are down 10% from the March highs.
Snowflake. Shares of cloud data lake provider Snowflake show what happens when investors reach for a speculative AI connection. Snowflake was seen as having a natural play in AI by managing data, but shares have been crushed since its fourth-quarter 2023 earnings announcement in February, when shares fell 20% in one day.
Enterprise Trends Are More Sedate
If AI is the elite infrastructure of enterprise technology, more pedestrian enterprise markets have, in fact, slowed and are seeing setbacks.
While the enterprise PC market is seen bouncing back in 2024 after a long slump, many other segments, such as campus networking and service provider, have been in decline. One clear indication of this is the financial results of networking market leader Cisco, which has shown stunted growth over the past year as enterprise orders declined.
Digesting the results of a recent enterprise channel survey, analysts at Raymond James said that many enterprise markets are growing in the low-single digits. And those comparisons will get tougher as the year advances because they are being compared year-over-year to a slow growth period in early 2023.
“Specifically, growth rates remain improved from trends experienced in 2023, but this is largely due to y/y comparisons easing as we begin to lap the rapidly decelerating environment that we flagged last year… “ wrote Raymond James analyst Simon Leopold in a recent research note. “Most partners we spoke to experienced low- to mid-single digit growth (up from low-single digit in 3Q23, but consistent with 4Q23), and consider the current environment stable.”
Stable might seem okay, but it doesn’t justify huge gains or valuations in the technology stock market, largely concentrated in the elite Mag7. So far, a pattern of disappointment has been established early in the year. Technology companies that don’t live up to the lofty expectations have been punished. With the second-quarter earnings season about to begin, investors might look for a continuation of the trends as the great AI reset unfolds. The coming earnings season will shine a spotlight on this.
