AI offerings are a bad joke, and investors aren't laughing

The timely bursting of the AI bubble has seen investors finally wake up and smell the coffee following the latest tech craze. That was briefly reflected by the stock performance of players in the AI arms race during the Aug. 5 market crash, when stock prices for Nvidia, Microsoft, Alphabet, Amazon, Apple and Meta plummeted — even if they did undeservedly recover the next day.

Some of the companies are making massive investments in AI. Meta has indicated through recent earnings reports that it expects to spend up to $40 billion on AI R&D in 2024. Microsoft has spent $56 billion, and the number is rising. Google is projecting an expenditure of $12 billion per quarter.

These are enormous numbers — even for Google — and investors have so far seen nothing for it. According to both Google CEO Sundar Pichai and Meta CEO Mark Zuckerberg, the risk of underinvestment in AI can’t be understated. They argue that building out data stores to train AI models takes time and resources, and being unprepared for the future isn’t a position they want to be in.

This is a sentiment that appeals strongly to investors who were caught out during the first tech boom — those who either lost a lot of money in 2001 when the early internet bubble popped or cronies who didn’t get in until the big money had been made (Warren Buffett himself, for instance). However this FOMO is dangerous and, frankly, this is starting to look like an arms race for a non-existent war.

Let’s take OpenAI — perhaps the most overhyped company in the market since Tesla and the creator of ChatGPT. This is the absolute darling of the sector — really its best hope — and according to some reports, it is eking out an annual revenue run rate of just $3.4 billion. As we can see from the capital being hemorrhaged by Microsoft — which owns 49% of OpenAI — that is absolutely nothing.

What is more, the majority of this is coming from subscriptions to what can only be described as silly jokes. ChatGPT is all but useless for any company wanting to market its products and services as it generates patent garbage that can be spotted a mile away. Meanwhile, kids can have fun morphing works of art into each other on DALL-E, while elsewhere AI is helping to generate deep-fake porn videos of Taylor Swift — great.

Related: Only Congress and DARPA can rein in the dangers of AI

The lawsuits this trash is rightly clocking up is a testament to its shelf-life, along with the growing concerns — again, rightly so — of regulators and governments across the globe regarding the enormous damage this technology can potentially do. Add to this to the lawsuit that Elon Musk is throwing at Sam Altman for “tricking” him into seeding OpenAI, and we have to wonder how long this joke will run.

There are some promising spots in this otherwise desolate landscape, such as AI-linked chip maker Nvidia, whose earnings remain strong. Infrastructure plays like this are always considered a safer bet, not least because AI is not Nvidia’s only revenue source. It has, though, faced recent shipment challenges and has attracted criticism from hedge funds like Elliot Management in London, which has warned that AI technology is nowhere near ready for "prime time," alleging that Nvidia is in a bubble.

Indeed, Microsoft has been at least transparent on timelines, claiming it expects to see a return on its investments within the next 15 years, while Meta has said it expects to see returns from generative AI “over a longer period of time.” Well, 15 years (or “a longer period of time”) are not acceptable timeframes for publicly traded companies — they are barely even tolerated in venture capital.

Right now, we can safely say the AI bubble has burst and Warren Buffett — who offloaded a $90 billion chunk of his Apple stock in the second quarter of 2024 — may be right this time around to be skeptical about the fate of the Magnificent Seven. Betting their fortunes on AI may well be their undoing. If AI does prove to be useful, it won’t be anytime soon.