The AI Capital Supercycle: Why the Real Costs Are Just Beginning

The current boom in AI capital spending is more than a technological shift. It is a systemic change that hides growing economic weakness. While markets currently reward massive infrastructure investment, this creates a dangerous split: AI-focused firms are pulling in capital, while the bottom half of American households faces a multi-year decline in purchasing power. This situation shows that the common solution of funding the AI build-out creates a hidden, compounding risk: a reliance on liquidity that is already starting to fade. Investors and leaders who choose short-term growth over long-term stability will be caught off guard when the current liquidity cycle resets. The real advantage lies in finding the specific points where the system is overextended rather than simply riding the current wave.

The Illusion of Perpetual Funding

The market treats AI infrastructure spending as a virtuous cycle, but the reality is more fragile. As Mike Wilson of Morgan Stanley notes, the market is absorbing a lot of paper by funding these costs through large equity and credit offerings. This creates a feedback loop where companies are rewarded for high spending relative to sales, which encourages even more spending.

However, this is a classic capital cycle trap. When the marginal cost of compute eventually hits zero, business models based on massive infrastructure spending will face a reckoning. Wilson points to the late dot-com era, where earnings looked strong before dropping 40% once the capital ramp-up ended. We are in the picks and shovels phase, but the shift to actual revenue from productivity is years away.

"There is a 100% chance there is going to be malinvestment here. Just like every major CapEx up... it is no different in railroads, no different electricity cycles. So we are probably not there yet, we do not know."

-- Mike Wilson, Chief US Equity Strategist, Morgan Stanley

The Hidden Divergence in the Real Economy

Robert Kaplan points to a clear split: the AI impact shows up in S&P earnings, but it hides a growing struggle for the median American household. With median incomes flat and purchasing power down 25-30% over the last five years, current economic growth is driven mostly by the affluent.

This creates a systemic weakness. If potential GDP growth remains low due to flat population growth and labor limits, high levels of government debt become harder to manage. The AI boom is acting as a temporary support for an economy facing structural resource limits. The danger is that policymakers and investors may mistake this capital-heavy growth for real, broad productivity gains.

Why the Obvious Fixes Create New Failures

The rush to adopt AI is causing friction in unexpected areas, particularly utility infrastructure. Raj Goyle, a candidate for New York State Comptroller, notes that the push for data centers is creating an affordability crisis for ratepayers. Because utility companies operate with guaranteed rates of return on capital spending, they are encouraged to build aggressively and pass the costs to consumers.

This is a classic example of a system routing around intended efficiencies. By subsidizing AI infrastructure, the public is paying for the competitive advantage of a few large firms. This creates a feedback loop of public frustration and legislative pushback, which will eventually force a re-evaluation of these tax breaks and subsidies.

"The utility industry is the only industry where you are paid to decorate your office because the CAPEX, you get a guaranteed rate of return."

-- Raj Goyle, Candidate for NY State Comptroller

The Value of Idiosyncratic Dislocation

While the broad credit market is in its richest decile of the last 25 years, Greg Peters of PGIM Credit argues that the real opportunity lies in the differences within the market. When index-level spreads are tight, they hide the fact that not all sectors are performing equally.

The most successful investors are not diving headfirst into the unsecured debt of the Magnificent Seven, but are instead looking for dislocations. The market is mispricing risk by treating all AI-related debt as equally safe. Savvy investors are finding better value in assets with structural collateral and wider spreads, betting against the market's uniform enthusiasm for AI-linked paper.

"The markets are separating strong from the weak. And so as a credit investor, you really enjoy the marketplace when companies are not necessarily at top of their game."

-- Greg Peters, Co-CIO at PGIM Credit

Key Action Items

• Audit your liquidity exposure: Over the next quarter, shift focus from growth metrics to liquidity resilience. As net equity issuance rises, the market's ability to absorb capital will tighten.
• Identify malinvestment candidates: In the next 6-12 months, monitor credit spreads for companies where CapEx spending is no longer rewarded by the market. These are your early warning signals for the inevitable cycle reset.
• Shift from unsecured to collateralized: If you are exposed to AI-sector debt, prioritize assets with hard collateral. The current market is overpricing unsecured debt in the Magnificent Seven space; look for structure where others are ignoring it.
• Monitor utility and data center legislation: Pay close attention to state-level moratoriums on data center construction. This creates a regulatory risk that could impact the long-term ROI of AI infrastructure projects over the next 18-24 months.
• Differentiate between infrastructure and adoption: Focus investments on companies that use AI to solve internal productivity problems rather than those just building the picks and shovels infrastructure. The former will survive the inevitable CapEx correction.