Recent regulatory filings reveal a puzzling trend: even as the S&P 500 reached new highs, institutional investors spent the second quarter trimming their positions in mega-cap technology stocks. This wasn’t a broad market retreat, but a calculated rotation. The data suggests a growing concern not about the promise of artificial intelligence, but about its colossal and escalating cost.
While the AI revenue narrative remains strong, the underlying story is one of capital efficiency. As tech giants commit to an unprecedented spending cycle to build out AI infrastructure, smart money appears to be repositioning. The focus is shifting from the platform owners, who bear the heavy burden of capital expenditure, to the upstream suppliers who profit from the buildout without the same balance sheet risk.
A Puzzling Divergence in a Record Market
The market backdrop makes the institutional selling in big tech particularly noteworthy. The S&P 500 recently closed at a new all-time high of 7,798.99, buoyed by a stable inflation outlook Saxo market quick-take analysis. Yet, aggregate data from Q2 13F filings, which disclose large institutional holdings, showed that 43.8% of managers trimmed their positions in the market’s largest technology companies Reuters institutional 13F report.
This divergence points away from simple profit-taking and toward a more structural re-evaluation of risk. Instead of abandoning the AI theme, investors seem to be changing how they get exposure to it.
The $800 Billion Bill for AI Dominance
The core of this re-evaluation is the staggering cost of the AI arms race. Projections for 2026 show aggregate capital expenditures (capex) for hyperscale cloud providers tracking between $700 billion and $800 billion Opening Bell Daily AI spending analysis.
More critically, this spending is now outpacing cash generation. For several top-tier cloud operators, capex is projected to reach 103% of their operating cash flow Opening Bell Daily AI spending analysis. This means for every dollar of cash earned from operations, more than a dollar is being spent on new data centers, servers, and chips. This dynamic severely constrains free cash flow, the lifeblood of shareholder returns through buybacks and dividends.
Following the Money Up the Supply Chain
The 13F filings also reveal where the capital is flowing. While trimming the well-known tech platform stocks, institutional net allocations expanded toward hardware manufacturers, semiconductor companies, and the broader AI infrastructure ecosystem Reuters institutional 13F report.
This represents a classic rotation up the supply chain. The investment thesis appears to be that companies providing essential inputs—from electricity and cooling systems to specialized chips and networking gear—can capture the upside of the AI buildout with far better capital efficiency. They sell the picks and shovels for the gold rush without having to own the capital-intensive mines themselves.
The Lingering Question: Depreciation Headwinds
A key uncertainty is how this massive capex will impact future earnings. These billions in spending are capitalized on the balance sheet and then expensed over time as depreciation. While this accounting treatment smooths the earnings impact initially, a multi-year wave of depreciation is building. This will act as a significant headwind to reported profits in the coming years, potentially compressing margins even if revenue continues to grow.
Watchpoint: Q3 Earnings and Capital Efficiency Metrics
The next major test of this thesis will arrive with the third-quarter 2026 earnings reports. Investors should look beyond headline revenue growth and focus on capital efficiency. The key areas to watch in company 10-Q filings will be the cash flow statement, any changes to depreciation schedules for property and equipment, and management commentary on return on invested capital (ROIC). These figures will provide the first concrete evidence of whether the returns from AI are justifying the historic investment.
*Disclaimer: This article is for informational purposes only and does not constitute financial, legal, or tax advice. This article was researched and drafted with AI assistance. All investment decisions should be made in consultation with a qualified professional.*
Want the full institutional-style PDF version? Enter your email for the free PDF.