One of the most often repeated cautionary tales in the world of technology investment is the story of telecommunications businesses in the late 90s, in the early days of the World Wide Web.
Back then, telcos in the US, predicting that the numbers on internet penetration and revenue would go up sharply — and keep going up in perpetuity – vastly overbuilt fibre network and infrastructure, leading to the collapse of dozens of businesses when the bubble burst.
Now, some investors are sounding a similar alarm on AI.
The current artificial intelligence boom has been sustained mostly by foundational “picks and shovels” work, like building out data centres and training AI models. Actual realised revenue, productivity improvements and end-user applications have mostly taken a backseat, outside of a few standout performers like OpenAI’s ChatGPT.
It’s this environment that has seen the tremendous bull run of Nvidia, which builds the lion’s share of chips that fill those data centres. Investors have treated it almost like an index fund for AI, sure to keep rising as tech companies try to figure out what much-hyped large language models are actually good for. (And how much consumers and businesses will pay for that benefit.)
