Meet Jethro the Giraffe (obviously). I hung out with him at Claws 'N' Paws Wild Animal Park in the Catskills this week. Like AI investors in North America, he was sticking his neck out.
For Jethro, that comes with the territory (dude’s 16 feet tall for goodness sake). He’ll be just fine. The same is not true of the North American comms market, which is about to suffer the consequences of massively over-inflating an AI bubble.
To be clear: AI as a whole is not the issue. Artificial intelligence is undoubtedly a transformative technology that will form the DNA of the next global digital economy. The problem is the massive over-investment in one (1) part of the AI ecosystem: the AI factories and data centers that sit in the middle of the AI economy, and massive underinvestment in everything else.
Nvidia was the initial trigger for this over-rotation. Like Jethro, most institutional investors are low-IQ herd animals, and Nvidia’s success triggered a stampede of dumb money into both the company itself--and the core data centers where its GPUs are used.
In the UK, we like to say, ‘no man is an island.’ Equally, no AI factory can exist in a vacuum. These workhorses of the AI revolution need power, water, and a network. Thanks to its ludicrous over-rotation on core technology, the US is miles ahead of the rest of the world in building the factories, but nowhere near delivering the other three essential elements of AI infrastructure, setting up the perfect conditions for economic disappointment accompanied by the loud “pop!” of a bubble bursting.
Nvidia itself presents a classic pre-bubble case study. Its market cap is almost $5 trillion, or equivalent to 20% of US GDP, based on selling a product that will need to be replaced in three years—at the same time as competitors, including AMD, Intel, Huawei, and Meta, catch up to it with their own high-performance GPUs and AI accelerators.
Make its valuation make sense. You cannot. It’s bubbleicious!
Note: Another symptom of the pre-bubble is CEOs saying really stupid things and no one challenging them, like when NVIDIA’s Jensen Huang claimed last month that AI would solve grid shortfalls itself by inventing better energy generation technologies—which is basically like saying “the dog will do my homework,” instead of “the dog ate my homework.”
Opportunity missed
While the US is busy inflating its AI core bubble, the real opportunity is going largely unnoticed (and underinvested) out at the edge of the AI biome, where artificial intelligence is used rather than processed.
Why? One reason is that the opportunity is highly complex. Building edge architecture for AI agentic workflows is an n-dimensional task involving a delicate balance of business case, technology, and industrial capacity. Institutional investors struggle with this kind of nuance --or indeed admitting that there might be anything in the heavens and on earth that they don’t understand. This is why they prefer making prehistoric grunting noises while throwing money at GPUs (“fast chip go fast, ugh, make me big money, ooh”) and AI factories (“fast chip go in big building, ugh ooh, make more fast, more money, ugh”).
But there are also a couple of cultural moats in place: edge networks are telco territory, which investors associate with high cost and low return, especially compared to what they perceive as the almost limitless potential returns on “something to do with AI.”
The second moat is even deeper and broader, separating the investor base from the highly specialized vertical industry environments where AI will be most transformative.
The media and analyst communities also have yet to grasp the nature of the opportunity. A recent report from Huawei was ridiculed for predicting that average monthly mobile data use will increase to 1 terabyte in 2035. Criticism of the forecast – which included words like ‘bonkers’ and ‘barmy’ – was based on comparing Huawei’s prediction to current consumer levels of mobile usage. This is a false equivalency. Huawei’s prediction is based on the arrival of an agentic-AI, IoT, 6G, Industry 4.0-enabled “new world” of ubiquitous digital connectivity where the number of mobile connected devices moves into the trillions.
I don’t know if Huawei’s 1 Terabyte claim is accurate, but I do know that it is absolutely correct about the direction in which the entire planet is heading. It is right to draw people’s attention to the seismic nature of the shift.
So, a bubble is forming in North America, but that doesn't mean the end of AI. That particular journey is just getting started; it will cover the entire planet, and there will be bumps (and bubbles) on the way.
Op-eds from industry experts, analysts or our editorial staff are opinion pieces that do not represent the opinions of Fierce Network.