By James St. Aubin, CIO of Ocean Park Asset Management
It seems that nobody wants to discuss the extraordinary concentration of prosperity and market value in a narrow cohort of artificial intelligence-focused technology companies. Since ChatGPT’s launch in November 2022, the “Magnificent Seven” — Nvidia, Microsoft, Meta, Google, Amazon, Tesla, and Apple — have appreciated more than 260%, while the average S&P 500 stock gained just 35%.
This concentration has reached extremes reminiscent of the Dot Com era. Nvidia alone, which has appreciated over 1,000% since ChatGPT’s launch, now represents approximately 7.3% of the entire S&P 500 Index and commands a market value exceeding $4.4 trillion. The Magnificent Seven collectively account for over a third of total S&P 500 market capitalization, a level of concentration that creates downside risk that shouldn’t be ignored.
The scale of AI infrastructure spending is undeniably massive. UBS projects approximately $375 billion in AI-related capital expenditures for 2025, rising to roughly $500 billion in 2026. McKinsey estimates that AI-focused data centers will require about $5.2 trillion of capital through 2030. These eye-popping forecasts have justified the valuation multiples: the S&P 500’s forward price-to-earnings ratio now sits at approximately 23x, Dot Com territory.
But here’s the uncomfortable truth: much of this infrastructure spending has been tech giants spending on each other. Nvidia invests billions in AI firms like OpenAI while simultaneously selling them chips and services, creating complex webs of interconnected deals. This circularity raises legitimate questions about the sustainability and ultimate return on investment of the entire ecosystem. If the demand for AI products and services fails to justify the massive infrastructure investments currently underway, the financial consequences could prove severe given the size and interconnectedness of these deals.
- Has Michael Burry got it right about AI sector earnings?
- Artificial Intelligence: boom, bust or just beginning?
- Seeing Machines lands breakthrough North American deal
The Bear Case nobody’s talking about
Consider three critical vulnerabilities that remain largely underappreciated by mainstream investors.
First, there’s the energy problem. Data centers consume enormous amounts of electricity, with the United States alone needing to add 80 gigawatts of electricity generation annually to meet expected demand. Currently, only 65 gigawatts are in development, suggesting a significant shortfall. Without adequate power infrastructure, the AI buildout may face physical constraints that limit growth regardless of demand or competitive positioning. These infrastructure limitations could emerge as binding constraints that derail even well-conceived investment theses.
Second, there’s no guarantee that today’s market leaders will remain market leaders. Technology history is littered with former dominants that faded into irrelevance. Netscape and AOL were once considered invulnerable. Another chip manufacturer could out displace Nvidia. Models might not need as much computing power. A new competitor could emerge with a superior approach. The question of whether current winners will maintain their competitive advantages amid inevitable waves of new entrants remains unanswered.
Third, and perhaps most troubling, is AI becoming a victim of its own success. If AI ultimately automates jobs faster than the workforce can adapt, the economic and sociopolitical consequences could be severe. Concentrated wealth, disrupted labor markets, and political backlash could create the conditions for significant economic upheaval and market volatility.
Balancing promise against prudence
The transformative potential of AI appears genuine and may well justify a market premium valuation of technology leaders over the long term. The major AI players possess strong balance sheets, robust earnings, and substantial free cash flow that has funded capital expenditures without heavy reliance on the debt and equity issuance common during the Dot Com era. This financial strength may provide a margin of safety largely absent in previous technology manias.
However, current valuations leave minimal room for disappointment. Even OpenAI’s CEO Sam Altman suggested caution when he told reporters, “Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes.” When the chief executive of the company that sparked the current euphoria expresses concerns about excessive enthusiasm, investors should take note.
Revolutionary technologies like radio, railroads, and the internet created both enormous value and spectacular losses for those who overpaid or concentrated within a narrow set of winners. As Mark Twain said, “History does not repeat, but it often rhymes.”
For investors, the takeaway is straightforward: beware of the “this time is different” mindset. Excessive optimism about a promising idea often leads to short-term disappointment, even when long-term success eventually follows.
Related ETFs
Would you like your funds listed here? Contact us
Artificial Intelligence & Robotics UCITS ETF Artificial Intelligence
LON:ARKI / USD 
Artificial Intelligence UCITS ETF Artificial Intelligence
LON:INTL / USD
Comments (0)