How Circular Deals Are Driving the AI Boom

Key Concepts

• AI Bubble: The potential for an economic bubble driven by inflated expectations and investment in Artificial Intelligence.
• Circular Deals: Investment strategies where money flows between companies, often customers and suppliers, without necessarily generating immediate profit. (e.g., Nvidia investing in OpenAI while OpenAI is a major Nvidia customer).
• Data Center Build-Out: The rapid expansion of data center infrastructure to support AI computing needs.
• Picks and Shovels Play: Investing in companies that provide the essential infrastructure (like construction, power, and hardware) for the AI boom, analogous to the gold rush.
• Infrastructure Arms Race: The competitive rush to build out data center capacity and secure necessary resources (power, water).
• Profitability Concerns: The current lack of profitability in major AI projects despite significant investment.
• Dot-Com Bubble Comparison: Drawing parallels between the current AI boom and the late 1990s dot-com bubble.

The AI Boom: Investment, Infrastructure, and the Bubble Question

The current surge in Artificial Intelligence investment is being likened to a “miracle” by markets, with companies like Microsoft, Meta, and Alphabet committing billions to capital expenditures. This isn’t solely a software revolution; it’s fundamentally reshaping infrastructure, driving massive construction of data centers and straining resources like energy and water. However, a precarious investment strategy is emerging, characterized by multi-billion dollar “circular deals” raising concerns about a potential bubble.

Circular Investment and the Nvidia-OpenAI Relationship

A key concern is the prevalence of circular deals, where funds are exchanged between companies without a clear path to immediate profitability. A prime example is Nvidia’s potential $100 billion investment in OpenAI. This is complicated by the fact that OpenAI is a major customer of Nvidia’s chips, and further interconnected through companies like Oracle, which leases compute power to OpenAI and is itself a customer of Nvidia. This “merry-go-round of money” – hundreds of billions flowing between these tech giants – is fueling anxieties. While not inherently inappropriate, the sheer scale of these transactions raises the risk of overextension. The symbiotic nature of these relationships also creates vulnerability: a stumble by one company could potentially destabilize the entire system. As one expert noted, “if one of those companies stumbles or doesn't do well, does the whole thing fall apart?”

The Infrastructure Build-Out and "Picks and Shovels"

Much of the investment is directly channeled into building data centers across the US, creating an “infrastructure build-out arms race.” Construction spending in 2025 is projected to decline in most sectors except data centers and power stations. This has created a boom for companies providing the “picks and shovels” of the AI industry – those involved in construction, materials, and infrastructure. Morgan Stanley estimates a $3 trillion spend on AI data centers, creating significant demand and opportunity for these supporting businesses. A case study highlighted a repurposed 1 million square foot textile facility being converted into a data center, demonstrating the insatiable demand for space and resources. The speed of deployment is critical; retrofitting existing facilities (6 months) is favored over building from scratch (2 years).

Power Demands and Operational Costs

The expansion of data centers is driving up utility costs, benefiting utility companies supplying energy to these facilities. However, operating these data centers is expensive. The video emphasizes that AI projects are currently operating at a loss, with each use of tools like ChatGPT potentially costing OpenAI money. Sam Altman projects OpenAI breaking even around 2029-2030, a timeline considered ambitious given current spending rates. Concerns exist about the ability of AI startups to meet their financial obligations, making AI data center companies a “canary in the coal mine” – early indicators of potential slowdowns.

Historical Parallels: The Dot-Com Bubble

The current AI boom is being compared to the dot-com bubble of the early 2000s. Similarities include circular deal-making and inflated expectations. The dot-com crash wiped out $5 trillion in value and took years for even strong companies like Amazon (8 years) and Cisco (25 years) to recover. The question is whether the AI bubble will reach a scale that has major economic consequences. The AI boom’s contribution to GDP growth, even amidst tariffs and inflation, underscores the potential impact of a collapse. Everyday investors are exposed to this risk through their 401Ks and other investment accounts.

Too Big to Fail? and the Long-Term Outlook

The sheer size of the companies involved raises the question of whether they are “too big to fail,” potentially requiring government intervention to prevent a wider economic collapse, similar to the 2008 financial crisis. However, despite the risks, many remain optimistic about AI’s long-term potential. The video draws a parallel to the early days of the internet, where initial overinvestment in fiber optic cables ultimately laid the foundation for modern broadband. The argument is that even if there is excess data center capacity, it will eventually be utilized. While some companies may not survive, the underlying technology of AI is considered fundamentally sound and not a bubble itself.

Notable Quotes

• “This is the wager to end them all.” – Describing the scale of investment in AI.
• “If one of those companies stumbles or doesn't do well, does the whole thing fall apart?” – Highlighting the interconnectedness and vulnerability of the AI ecosystem.
• “Time is not your friend” (in AI) – Emphasizing the need for rapid deployment and innovation.

Conclusion

The AI boom represents a significant economic gamble, fueled by massive investment and a rapid build-out of infrastructure. While the potential rewards are substantial, the prevalence of circular deals, current lack of profitability, and historical parallels to the dot-com bubble raise serious concerns about a potential bubble. The long-term success of AI is not in question, but the video suggests that a period of correction and consolidation is likely, with some companies failing along the way. The situation demands careful monitoring, particularly of key indicators like data center demand and the financial health of AI startups.