ISTANBUL

US-based tech giants like Meta, Amazon, Microsoft, and Alphabet are projected to spend around $670 billion on capital investments this year, primarily funding their artificial intelligence (AI) and data center efforts.

Analysts say this massive influx of capital resembles the rapid railroad expansion of the Gilded Age, covering a period from the late 1870s to the late 1990s, when the world’s first billionaires emerged with industrialization.

The rise of railroads and the further advancement of port and mining activities held a key place in the US economy at the time, while agriculture began to lose its former glory.

Such large investments may have a long payback period, and the cash flow and profitability may come under pressure in the short term, analysts warn.

Alphabet, for instance, used very long-term (100-year) bonds to finance its rising AI spending, similarly to IBM’s borrowing in 1996.

This recent and massive shift in technology follows historical milestones, such as the widespread home adoption of personal computers in 1974 and the machine learning model Deep Blue’s victory against chess champion Garry Kasparov in 1997. The digital revolution thus came to the fore after the leap in the Gilded Age.

Nowadays, the world is witnessing another leap. The emergence and the consequent widespread adoption of large language models (LLMs) these days, popularized by the rise of AI chatbots like ChatGPT, pushed the use of AI beyond niche applications into the mainstream.

Companies began to shape their policies and plans around AI, and inevitably, AI spending grew on a massive scale, given that the world’s wealthiest are American tech billionaires.

The US economy is expected to see an economic breakthrough like that of the 1870s with AI, but how these AI spending and investments will affect companies is unclear.

Recent AI spending may be most significant capital investment cycle since railroads

Charles Diebel, an investment advisor and macro strategist, told Anadolu that everything depends on “whether the capital expenditure proves profitable in time.”

“If you look at US railways in the 1870s, the answer is no,” he said. “But this is possibly different—companies loading that much debt on (their) balance sheet need the business model to work to success, (and) as yet, that’s unproven.”

Ethan Feller, equity strategist at Zacks Investment Research, told Anadolu that AI infrastructure spending is “on pace to reach roughly 2% of US GDP (gross domestic product) this year.”

“It's obviously a tailwind for semiconductors and adjacent technology businesses, but the ripple effects extend well beyond tech—industrial companies, power infrastructure, and construction are all direct beneficiaries of this buildout,” he said.

Feller stated that while it is difficult to estimate the long-term effects of AI investments, “they’re likely to be dramatic.”

“Look at what the mega-cap technology companies are signaling with their capital allocation,” he said. “Alphabet and Amazon both meaningfully raised their capex (capital expenditure) guidance at their most recent earnings reports, with the two companies alone expected to spend roughly $400 billion this year.”

He noted that both Amazon and Alphabet are gaining momentum in cloud services, which points to strong demand from businesses for AI and cloud computing.

“I don’t think they would commit hundreds of billions in capital unless they thought they would get a good return on that money,” he said. “What the cloud demand data tells us is that large-scale enterprise AI deployments are already underway.”

Feller stated that OpenAI and Anthropic’s latest models represent a “genuine step change in capability,” and that these products could eventually “displace the majority of entry-level white-collar workers,” citing a recent statement by Anthropic CEO Dario Amodei.

“It’s a bold claim, but if you’ve used the latest models, it’s increasingly difficult to dismiss,” he said.

“Many engineers at leading AI labs openly acknowledge they've delegated the bulk of their technical work to large language models and now primarily direct the work through prompts. The logical next step is turning that same capability toward other knowledge-work professions—accounting, legal, content creation, (and) customer service,” he added.

*Writing by Emir Yildirim