The artificial intelligence was over Rush & Bubbels

The artificial intelligence was over Rush & Bubbels

Last week, The Wall Street Journal (WSJ) published a piece that commented on the Infrastructure Frenzy of Artificial Intelligence (AI). As an investor, one can only be impressed and feel worse at the same time and a more than a touch of déjà vu.

According to the Wsj, Ellendale, North Dakota, is a sleepy town of only 1,100 that is now hosted for a half-built AI data center that will be larger than 10 home depots the American equivalent of a large Bunnings warehouse here in Australia.

With a price tag north of US $ 15 billion, that one data center also represents a quarter of the annual gross domestic product (GDP) of the North Dakota.

We have previously commented on whether the expenditure is justified by the income, but every news story in which the dollar size of the AI ​​race is set out forces one to pause and think again.

AI proponents call it the fourth industrial revolution – the continuous transformation of industries and societies driven by artificial intelligence, robotics, internet of things (IoT), blockchain, quantum computing and biotechnology, and build on the digital basis of the third industrial revolution that was aimed at computers and the internet.

Payback time

On the contrary, however, the article points to the big question: how on earth are planning to earn back these investments, and when?

The scale of what the technological equivalent is of a ‘by spec’ build in the office construction industry is amazing.

In the past three years, Tech -Titans have postponed obligations in data centers, chips and energy that historical precedent dwarf. Satya Nadella from Microsoft said that he hopes it will not take 50 years, such as the slow adoption of electricity. Mark Zuckerberg from Meta says they are all investing as if it will not – point to the potential US $ 600 billion from Meta US US until 2028.

But in the end this is a gamble with a high commitment to AI that develops quickly enough to reform the economy and squeeze out profit.

The WSJ draws alarming parallels with the late 90s Dot-Com Manie, where Telecom covered the US with more than US $ 100 billion in fiber optics, only to crash spectacularly. At the time, giants such as WorldCom and Global Crossing belly-up were surrounded by massive overbuning. Nowadays, investors must be reminded that chips and data center-nu an arena almost trillion dollar-up were boring.

The WSJ Journal describes the Dogendale project of CoreWeave as a poster -child for the Gold Rush. Only six years ago it was a small crypto miner worker with fewer than two dozen employees. Now, equal to money in Wall Street, it is appreciated higher than General Motors of Target. Turning Crypto to AI Cloud Computing Post-Chatgpt, and with a culture that calls “Yolo” (you only live once) and “GSD” (Get done) has collected more than US $ 42 billion in contracts, including a reinforced deal with OpenAI. But it is all fed by debts $ 15 billion in loans against rates from more than eight percent, plus $ US56 billion in long-term lease for data centers.

Slightly frightening for anyone who understands the concept of ‘operating a going-chear’ CoreWeave’s technical deals are shorter term (2-5 years). Matching short -term income contracts with long -term conditional obligations allows the company to be exposed if the question dries up. If overbuilding occurs or tenants are a guarantee or switch, CoreWeaves products and services become the dark fiber cables of the 2020s. A Phoenix (AI version 2.0) can stand up from the axle, but you first require a fire to torestor AI 1.0.

According to the report, David Cahn of Sequoia estimates that only AI infrastructure investments 2023-2024 US $ 800 billion in product sales require during the 3-5-year lifetime of the chips for solid returns. Bain & Co. This increases that up to US $ 2 trillion to annual AI turnover by 2030 – more than the combined sale of Amazon, Apple, Alphabet, Microsoft, Meta and Nvidia today, and five times the global market for subscription software.

Where does all this money come from? Which customers are talking about after they have also spent money with Amazon, Apple, Alphabet, Microsoft and Meta?

Morgan Stanley estimated last year’s AI income at only US $ 45 billion, usually from Chatbot subscriptions and cloud access. While consumers love free AI tools, companies are reportedly stingy further than US $ 30/month per user for things like Copilot from Microsoft. Investor Roger McNamee thinks this bubble is larger than all previous technical bubbles together. He says that even success will not justify current expenses.

The same, the same but different

As we have noticed before, this is of course not the DOT-Com Bubble. Today’s hyperscalers (Alphabet, Microsoft, Amazon, Meta) generate cash and much more than those fiber companies from the 90s. Moreover, the Chatgpt of OpenAI has 700 million weekly users (against 500 million in March), with income in 2024 to US $ 13 billion and the pindas are compared to the uses of uses compared to the Pindas compared to comparison compared to comparison compared to comparison compared to the annual committee $ 1 trillion+ data center plans.

And here is an interesting dichotomy. Many proponents note that if AI massively moved White-Collar jobs, the savings can justify all expenses. But if all those white collar lanes are lost, who has the money to buy AI-driven products and services? Henry Ford gave his factory workers a wage increase because he realized that they were his customers. If AI wages the duties of employees, who will the customers be?

A web of guilt and guarantee

Elsewhere, the financing web is complicated, with debts everywhere. The hyperscalers are said to look at US $ 400 billion in 2025 Capex, which is as an aside, more than the Apollo program that is adapted for inflation.

Some players build their own centers; Others rent intermediaries such as CoreWeave, who then pack them with Nvidia chips. Coreweave was rented from Applied Digital in Ellendale, another ex-Crypto player AI was hopeful.

History full of new technology that invested investors

One reason for bubbles is that investors repeatedly make the mistake of believing technology that changes the course of human history, must also be good for investors. From canals and railways to electricity, commercial air travel, the car and television, history is full of examples of technology that rewarded consumers more than investors.

The excitement of the potential of technology stimulates a race that results in overbuilding. That overbuilding is then followed by losses. Ultimately, the technology survives thanks to consumer demand, but not every investor in a provider of that technology earns money. History indeed reveals that most investors lose.

The WSJ quotes the ‘collective hallucinations’ of Andrew Odlyzko in Manias, where hype risks surpasses-as the Dot-Com fiber stoot, where the growth of traffic growth was overestimated, which led to bustes. Execs at companies such as Level 3 did Wild, only to see 95 percent shares and fiber to be inactive for years until streaming came and saved it.

Nowadays, Applied Digital broke the land on another data-less data center without a tenant.

Red flags

We recently described a MIT report that has shown that 95 percent of companies generate zero return on investment (ROI) on their AI companies. Elsewhere, a study in Chicago showed that chatbots do not offer a profit boost in Danish workplaces. In the meantime, the release of Augustus of Chatgpt-5 was rather than revolutionary, and that was despite the escalating training costs of three to five times per model. Chips are also replaced and are therefore quickly written off, unlike permanent fibers.

Of course a frog cannot feel the water temperature rise softly, so the AI ​​race continues, fed by a seemingly endless flow of money. Like the mayor of Ellendale, Don Flaherty, says: “We are now on the Golf and we just have to keep driving on it.”

More from Rogerinvest with Montgomery

Roger Montgomery is the founder and chairman of Montgomery Investment Management. Roger has more than three decades of experience in fund management and related activities, including stock analysis, stock and derivative strategy, trade and effects. Prior to the establishment of Montgomery, Roger positions in Ord Minnett Jardine Fleming, BT (Australia) Limited and Merrill Lynch.

He is also the author of the best -selling investment guide for the stock market, value. Aabel-Hoe to appreciate the best shares and buy them for less than they are worth.

Roger regularly appears on television and radio, and in the press, including ABC Radio and TV, the Australian and Ausbiz. View upcoming media performances.

This message was contributed by a representative of Montgomery Investment Management PTY Limited (AFL No. 354564). The main purpose of this message is to provide factual information and not to provide financial product advice. Moreover, the information provided is not intended to give a recommendation or opinion about a financial product. However, each comments and opinion of opinion can only contain general advice that has been drawn up without taking into account your personal objectives, financial circumstances or needs. Therefore, before acting on the basis of one of the information provided, you must consider the suitability in the light of your personal objectives, financial circumstances and needs and you must consider requesting independent advice from a financial adviser if necessary before you make decisions. This message excludes specific personal advice.