AI Investments: Boom or Bubble?

Despite massive investment driving stock surges, the AI sector faces warnings of a speculative bubble fueled by “round-tripping” and a critical lack of real enterprise adoption. Experts draw parallels to past market crashes as future commitments far outstrip current profitability and tangible returns.

The spectacle of artificial intelligence has captivated the global economy, painting a future of boundless innovation and unprecedented growth.

It’s a narrative so compelling that AI investments are projected to fuel a staggering 40% of the United States’ GDP growth in 2025. Most US Growth Now Rides on AI

AI companies alone account for an astonishing 80% of the surge in American stocks.

From the polished stages where tech titans like Microsoft’s Kevin Scott and OpenAI’s Sam Altman share the spotlight, the message is clear: AI is not just the next big thing.

It is, by many measures, the only big thing driving economic momentum.

Yet, beneath this gleaming surface of optimism and soaring valuations, a disquieting echo from financial history reverberates.

A closer look at some of the most significant deals shaping this AI-driven landscape reveals a pattern that analyst Rishi Jaluria of RBC Capital Markets warns could be “round-tripping”.

This is an unethical, often illegal, practice of transactions designed to artificially inflate an asset’s value.

It’s a scenario that calls to mind a snake consuming its own tail, where money appears to circulate among a select group of interconnected firms.

This boosts stock prices without necessarily generating commensurate real-world value.

On the surface, these agreements are pitched as mutually beneficial, a necessary lubricant for an emergent industry.

Jaluria himself acknowledges the potential upside: a “less capacity-constrained world” where faster model development leads to higher returns.

This enables the realization of AI use cases currently on hold.

If these breakthroughs materialize into tangible cost savings and new revenue streams for customers, then the grand vision holds, creating net benefits for GDP.

The caveat, however, is critical: “If there is no real enterprise AI adoption, then it’s all round-tripping.” AI Will Transform the Global Economy

And here lies the uncomfortable truth.

While OpenAI’s technological advancements, such as the Sora 2 video generation model, are undeniably impressive, the actual utility for early corporate adopters remains elusive.

A recent report indicates that a staggering 95% of companies attempting to integrate generative AI tools into their operations have yielded precisely zero return on investment.

The real returns, it seems, are primarily confined to the stock market, where investor confidence, rather than proven profitability, is the primary currency.

Consider the curious case of Oracle. Oracle (ORCL) Stock Price & Overview

Despite missing both revenue and earnings projections, and reporting flat net income year-over-year, its stock price recently soared.

This catapulted CEO Larry Ellison into the pantheon of the world’s wealthiest individuals.

The catalyst? A massive 359% increase in “remaining performance obligations” (RPO), a ledger of future revenue from unfulfilled financial agreements, projected to reach $455 billion.

The bulk of this eye-popping figure is attributed to a colossal commitment from OpenAI.

This is a $300 billion purchase of computing power from Oracle over five years, a contract slated to begin in 2027.

The sheer scale of this agreement, the largest in tech history if it materializes, borders on the fantastical.

For Oracle to fulfill its end, it would need to generate 4.5 gigawatts of power capacity – more than two Hoover Dams’ worth of electricity.

OpenAI, a still-private company reporting roughly $3.4 billion in annual revenue through 2025, would be on the hook for approximately $60 billion annually to meet its obligation.

The disparity between current operational realities and these future commitments is stark, raising legitimate questions about feasibility and sustainability.

This circular dynamic isn’t isolated.

OpenAI’s deal with Nvidia rival AMD also saw the chipmaker project $400 billion in AI chip sales over the next half-decade, with OpenAI acquiring a significant stake, including options for up to 10% of AMD.

Conveniently, this announcement coincided with a substantial surge in AMD’s stock price.

These deals, often laden with contingencies (like Nvidia’s $10 billion initial investment in OpenAI contingent on achieving 1 gigawatt of data center capacity, with a potential for $100 billion if 10 gigawatts are reached), are treated by the market as if they are set in stone.

This drives valuations to dizzying heights.

OpenAI itself projects a tenfold increase in revenue in the coming years, aiming for $10 billion by 2025.

Such inflated projections and interconnected financial commitments evoke comparisons to past market frenzies.

While some might recall the dot-com bubble, where companies like Webvan achieved a $21 billion valuation with minimal revenue, Peter Atwater, Adjunct Professor of Economics at William and Mary, sees a more unsettling parallel: the 2007 housing market collapse.

Atwater describes “conveyor belts of capital” where money flowed between interdependent parties, creating a fragile ecosystem.

He notes that “any participant in the system was then dependent on every other conveyor belt in the system working simultaneously to keep the system going.” A new look at the economics of AI

He observes a strikingly similar “developing web of capital flows across the AI space.”

In the fever pitch of a bubble, Atwater explains, “everyone overcommits.”

Buyers overcommit to what they believe will be a scarce future commodity, and sellers agree to overprovide.

The danger, he warns, is that “commitments are among the first obligations to be cut off once conditions change, once confidence begins to fall.”

The numbers underscore this precarious balance.

Over the past two years, tech giants like Microsoft, Meta, Tesla, Amazon, and Google have collectively poured approximately $560 billion into AI infrastructure, yet have brought in a combined $35 billion in AI-related revenue.

OpenAI’s commitments dwarf even these, with even less tangible returns so far.

The massive data centers required to power this AI revolution demand energy equivalent to that of a small country, yet the ultimate revenue generation remains an open question.

“Ultimately, if you do not have a consumer for the product, there will be no AI space,” Atwater asserts, highlighting the critical, yet unanswered, question of monetization.

For now, the AI sector operates in a “forever mindset,” confident it has ample time to figure out how to translate technological prowess into sustainable profits.

As long as confidence remains high, this ecosystem can sustain its “fantasy.”

But when that confidence inevitably falters, the demand for “real-term performance in a very short time frame” will be relentless.

And the consequences, Atwater cautions, will extend far beyond the tech sector itself.

“You have to look at this as a larger ecosystem.

To talk about AI today, it means we have to talk about the credit market.

Wall Street and AI are a single beast.”

A very small number of firms, he warns, currently hold a disproportionate grasp on the American economy, making the stakes incredibly high.

Investors, fueled by a fear of missing out, continue to pile into AI, often without scrutinizing the foundations of these soaring valuations.

The question of whether this colossal flow of capital is genuinely building a new economic bedrock or merely inflating an elaborate, interconnected bubble remains unanswered, with potentially profound implications for us all.