Surely one of the silliest things that happened in tech stocks in 2024 was the sudden tumble in Nvidia Corp. shares moments after its fiscal second-quarter earnings release in August. Chief Executive Officer Jensen Huang, who otherwise walked on water this year, mentioned a minor — and resolved — blip in production of its new chip, yet investors panicked anyway.
They soon sobered up. The world, it turned out, continued to spin. But what it highlighted were the deep anxieties lurking just beneath the thin surface of optimism on artificial intelligence. The market is on high alert for signs of the peak and will react disproportionately whenever it thinks it has found one. This doesn’t bode well for 2025, when cooler heads must prevail as the progress of AI development seems almost certain to slow, possibly to a crawl.
Over the past several weeks, AI leaders have been choosing their words carefully. Google CEO Sundar Pichai, speaking at a New York Times event, said he felt the “low hanging fruit” of AI had now been picked. Expanding on the point, he told Semafor: “As we go to this next level, you need more insightful breakthroughs.”
Sam Altman, the co-founder and CEO of OpenAI, talked about how he still felt his company would reach “artificial general intelligence” but that it would “matter much less” than some observers might have previously thought. Superintelligence would be the great disruptor, he said — but it’s further away.
Behind the scenes, several reports have suggested that OpenAI is struggling to conjure the great leaps in capability that had been expected. The Microsoft-backed company’s long-awaited Sora video generation model can still be considered a hugely expensive parlor trick. Recent model releases, boasting “reasoning” capabilities and other bells and whistles, have been plainly iterative but no less expensive to create than previous models. The “wow” factor of ChatGPT is ebbing away.
Apple, meanwhile, has yet to post any evidence of the iPhone “supercycle” that some had hoped would be spurred by the introduction of Apple Intelligence. In fact, the company’s tentative AI incursions have been something of an embarrassment. Its AI-generated summaries have ranged from the comical to the severely inaccurate. With past innovations, the company told itself it would be the best if not the first. With AI it hasn’t managed either.
The first indication of the impact of Apple Intelligence on the company’s bottom line will come from 2024’s holiday quarter sales. These will most likely be record breaking, as ever, but consumers are still largely making buying decisions based on better cameras and long-lasting batteries rather than AI. And it’s not as if there’s a sure solution on horizon — Cook said in June that he would “never” claim that Apple’s AI was 100% safe from hallucinations.
That speaks to one significant concern as we head into the new year, that shoveling more and more data into larger models will only go so far when it comes to creating “intelligent” capabilities, and we’ve just about arrived at that point. Even if more data were the answer, those companies that indiscriminately vacuumed up material from any source they could find are starting to struggle to acquire enough new information to feed the machine. “We have but one internet,” lamented leading engineer Ilya Sutskever, who left OpenAI to launch his own AI company. Those who are in the business of producing new information are now, awkwardly, demanding to be paid.
Some commentators say it is obvious that AI — or generative AI, to be more specific — has been overhyped by those with a strong financial interest in fostering giddy excitement over its prospects. Others say the jury’s still very much out on what AI will become. Another view is that current capabilities, when sensibly deployed, already represent a sea change. I don’t know who is right. In my defense, neither do the companies that have spent tens of billions of dollars trying to find out and seem set to spend even more before the most lucrative use cases are yet to be figured out. In the meantime, efforts to balance the books, by charging a lot more for AI’s use or introducing advertising, could be poorly received.
None of this is an argument for selling AI stocks. It’s an argument for holding them, even if the next year pans out to be a far less thrilling ride than the past two. Jefferies analyst Brent Thill published a recent report comparing the trajectory of AI software revenue to a slow runway take-off — not the rocket ship launch enjoyed by the companies selling semiconductors or access to computing power or both.
More meaningful AI software revenue may not come until 2026, Thill wrote. That will come as a disappointment to many. If August’s Nvidia wobble is anything to go by, investors will be easily spooked by any hint of bad news. Wall Street’s New Year’s resolution should be to rein in its expectations on AI.
A message from Advisor Perspectives and VettaFi: To learn more about this and other topics, check out some of our webcasts.
Bloomberg News provided this article. For more articles like this please visit
bloomberg.com.
Read more articles by Dave Lee
