A bubble may be brewing in individual AI stocks, based on the price targets of select analysts.
In the mid-1990s, the advent and proliferation of the internet revolutionized corporate America by opening new sales channels and creating connections that hadn’t previously existed. Since the internet, investors have been patiently waiting for the next-big-thing technology to provide a true leap forward for corporate America. The arrival of artificial intelligence (AI) looks to be the answer.
AI provides a way for empowered software and systems to make split-second decisions without the need for human oversight or intervention. In Sizing the Prize, the analysts at PwC pegged this global game-changing opportunity at $15.7 trillion (with a “t”) by 2030.
Image source: Getty Images.
While sentiment on Wall Street and among analysts has been mostly bullish — as you’d expect with a $15.7 trillion addressable market — not every AI stock is necessarily worth buying. According to select Wall Street analysts, three of the market’s scorching-hot AI stocks could plunge by as much as 72% over the next year.
Palantir Technologies: Implied downside of 72%
Though graphics processing unit (GPU) titan Nvidia is the face of the AI movement, arguably no company has come closer to dethroning it than AI and machine learning-driven data-mining specialist Palantir Technologies (PLTR 4.97%). Shares of Palantir have soared more than 2,100% since 2023 began, equating to an increase in market value of around $320 billion.
The primary reason investors have gravitated to Palantir is its sustainable moat. Its Gotham platform, which secures multiyear contracts from the U.S. government and its immediate allies to collect/analyze data and assist with military mission planning and execution, is irreplaceable. Meanwhile, its Foundry platform, which is designed to help businesses make sense of their data in order to streamline their operations, has no large-scale one-for-one replacement.
However, this sustainable moat isn’t enough to impress longtime bear Rishi Jaluria at RBC Capital Markets. Although Jaluria nearly quadrupled his price target on the company from $11 to $40 earlier this year, a $40 bullseye would represent 72% downside from the $142.10 per share Palantir stock closed at on July 11.
Jaluria’s main issue with Palantir stock is something I’ve harped on repeatedly in recent weeks: its valuation.
Prior to the dot-com bubble, many of Wall Street’s cutting-edge companies topped out at price-to-sales (P/S) ratios of 31 to 43. Palantir ended the previous week at a P/S ratio of almost 114! No megacap stock in history, to my knowledge, has been able to sustain a valuation this aggressive — even those with well-defined competitive advantages. Even the slightest operating slip-up or negative news from the U.S. government could clobber Palantir stock.
Jaluria also cautioned that Foundry takes too tailored of an approach with its clients, which will hamper its ability to scale. The same can also be said for Gotham, which is only available to the U.S. and its immediate allies. In other words, Palantir stock is on shakier ground than its skyrocketing share price implies.
Image source: Getty Images.
Super Micro Computer: Implied downside of 51%
Another red-hot AI stock that has the potential to be pummeled over the next 12 months is customizable rack server and storage solutions specialist Super Micro Computer (SMCI 1.00%).
Shares of Supermicro are up 62% year-to-date (through July 11) and more than 1,100% on a trailing-three-year basis. The reason it’s been a magnet for AI bulls is its role as a provider of customizable rack servers for AI-accelerated data centers. Businesses are aggressively spending on data center infrastructure to gain a competitive edge, and Supermicro’s reliance on Nvidia’s highly popular AI-GPUs in its rack servers has allowed its servers to sell like hotcakes.
Following sales growth of 110% in fiscal 2024 (its fiscal year ends on June 30), Wall Street is forecasting 48% sales growth for fiscal 2025 and another 34% the following year.
None of these figures have been enough to dazzle analyst Michael Ng of Goldman Sachs, who rates Super Micro Computer a sell and expects its shares will fall to $24, equating to 51% downside from where they ended the previous week.
Ng’s skepticism derives from a belief that the AI server market is becoming highly competitive, which is leading less differentiation and, ultimately, weaker pricing power. Ng anticipates Supermicro’s gross profit margin will decline throughout the decade, even as sales potentially climb.
Though not specifically mentioned by Ng, Super Micro Computer must also overcome a loss of trust with the investing community following allegations of wrongdoing last summer. While an independent committee absolved insiders of any wrongdoing and didn’t result in any changes to the company’s reported financial statements, it challenged investors’ trust in the management team and squashed any chance of Supermicro commanding much of a valuation premium.
Even though Supermicro’s stock may appear cheap at just 17 times forward-year earnings, there are reasons investors are leery about giving its shares too much of a premium.
SoundHound AI: Implied downside of 31%
Lastly, AI voice recognition and conversational technologies stock SoundHound AI (SOUN -0.95%) can plunge over the coming year, based on the prognostication of one Wall Street analyst.
Growth has not been an issue for this up-and-coming AI applications company. Sales for the March-ended quarter jumped 151% to $29.1 million from the prior-year period. This speaks to the company’s ability to win new clients in the restaurant, automotive, travel and hospitality, and financial service industries, as well as tie these ecosystems together.
Despite SoundHound AI decisively pointing its revenue needle in the right direction, Northland Securities analyst Michael Latimore foresees its stock plummeting to $8 over the next 12 months, which works out to a decline of 31%.
Whereas the prior two analysts are decisively negative on Palantir and Supermicro, this isn’t the case with Latimore and SoundHound AI. Latimore has a hold rating on the company and is excited about the agentic AI opportunities that lie ahead.
The reason price targets should be kept in check is that SoundHound AI has a long way to go before it demonstrates to Wall Street that its operating model can generate profits. Excluding adjustments to contingent acquisition liabilities during the March-ended quarter, its adjusted loss actually widened from $20.2 million to $22.3 million, in spite of 151% growth in net sales from the prior-year quarter. SoundHound AI also burned through close to $19.1 million in cash from its operating activities. The company isn’t expected to push into the recurring profit column until 2027, at the earliest.
SoundHound AI would also be heavily exposed if an AI bubble formed and burst. Every next-big-thing technology for more than three decades has navigated its way through an early stage bubble, and nothing suggests artificial intelligence is going to be the exception to this unwritten rule. If demand for AI applications even remotely slows, SoundHound AI stock, which is valued at 23 times forward-year sales estimates, will feel the pain.
