AI Capex: The Structural Tailwind and Selective Rotation Ahead

The scale of AI-driven capital expenditure is now a defining feature of the corporate landscape. Wall Street analysts now estimate that the group of major tech companies will spend about $527 billion on capital expenditures this year, a significant increase from the $465 billion forecast just a quarter ago. This spending is not a broad-based corporate trend but a hyper-concentrated cycle, with four tech giants-Amazon, Microsoft, Alphabet, and Meta-accounting for a dominant share of all S&P 500 capex. Their spending has soared 117 percent in the last 10 quarters, dwarfing the 16 percent rise seen across the rest of the index. This creates a powerful structural tailwind for the overall market, as massive, sustained investment in AI infrastructure supports growth and justifies valuations.

Yet the market's initial enthusiasm is giving way to a new phase of selectivity. The core investment thesis has evolved: while the AI capex cycle itself is structural and likely to persist, the returns on that spending are not being rewarded equally. Investors have begun to rotate away from pure infrastructure plays where capex is heavily debt-funded and operating earnings growth is under pressure. This divergence is already visible, with the average stock price correlation among large public AI hyperscalers having collapsed from 80% to just 20% since June. The setup now favors companies that can demonstrate a clear and profitable link between their massive spending and top-line revenue growth.

For institutional allocators, this means the AI trade is entering a more nuanced, conviction-driven stage. The bedrock thesis remains intact-the spending will continue to support economic activity and equity returns. But the path forward demands a sharper focus on capital allocation discipline, credit quality, and the ability to convert capex into sustainable profits. The era of broad-based AI stock rallies is likely over; the next phase will be defined by selective rotation toward the most efficient capital allocators.

Financial Impact: Earnings, Credit, and the Profitability Test

The massive AI capex cycle is now a central pillar of the 2026 earnings outlook, but it introduces a critical tension. Goldman Sachs projects double-digit earnings growth to drive the market, with EPS expected to increase 12% in 2026. This growth must now fund a capex cycle that is projected to start to slow down in growth terms. The implication is clear: the market's ability to justify continued high valuations will increasingly hinge on the profitability of this spending, not just its scale. As analysts note, the slowdown in capex growth will force a more rigorous profitability test, causing traders to pick and choose winners and losers among the largest tech firms.

This divergence is already evident in the broader industrial economy. While tech/AI investment soars, manufacturing construction spending steadily declined in 2025, hampered by trade policy uncertainty and tariffs. This creates a stark contrast between two capital allocation stories. The capital markets are channeling resources toward the highest-return opportunities, as evidenced by the rational redirection of limited resources to higher-return opportunities like data centers. For institutional investors, this signals a structural shift in credit quality and risk. The financial health of the AI hyperscalers is becoming more critical, while other industrial sectors face headwinds that could pressure their balance sheets and limit their ability to invest.

The reallocation is also reshaping adjacent sectors, most notably power and utilities. The explosive growth in data centers is creating a fundamental demand shock. As one analysis notes, the data center buildout is so significant that it is on track to surpass the entire general office construction market this year. . This is not just about new buildings; it is about a massive, sustained demand for electricity. The power sector is thus entering a new growth phase, with transmission and generation projects being re-evaluated for their ability to support this new load. The financial impact here is twofold: it creates a structural tailwind for power infrastructure companies, but also introduces new regulatory and construction risks as the grid strains to keep pace. For portfolio construction, this means the AI story extends beyond the pure-play tech names to include the essential enablers of the infrastructure buildout.

Portfolio Implications: Sector Rotation and Risk-Adjusted Returns

The institutional playbook for the AI cycle is shifting from broad exposure to selective rotation. The next phase of the trade, as outlined by Goldman Sachs, is expected to favor AI platform stocks and productivity beneficiaries over pure infrastructure plays. This is a direct response to the divergence already in motion, where investors have rotated away from companies where capex is heavily debt-funded and operating earnings growth is under pressure. For portfolio construction, this means overweighting the cloud operators and software platforms that can demonstrate a clear, profitable link between their massive spending and top-line revenue growth. The structural tailwind remains, but the risk premium is now tied to execution and profitability, not just scale.

This rotation creates a potential divergence from other market themes. While the AI leaders command high multiples, institutional investors are also actively seeking value stocks and anticipating a rise in IPOs and dealmaking. This search for value introduces a counter-current to the concentrated, high-multiple AI rally. The portfolio implication is one of active management: maintaining a core overweight to the most efficient capital allocators in AI, while deploying capital into sectors and companies that offer better value metrics and are less exposed to the debt-fueled capex cycle. The goal is to capture the productivity boost from AI adoption while mitigating the risk of overpaying for unproven returns.

The primary risk to this setup is a slowdown in the capex growth rate without a corresponding acceleration in earnings. Goldman Sachs notes that AI capex growth will start to slow down in 2026. If this deceleration occurs before the promised productivity and profit gains materialize, it could compress valuation premiums for less efficient spenders. The math is stark: maintaining current returns on capital would require these companies to generate over $1 trillion in annual profit, more than double the current consensus estimate. For institutional flows, this creates a two-way risk. The market's ability to sustain its rally will hinge on whether earnings can meet the elevated expectations set by the spending cycle. Any disappointment could trigger a more pronounced rotation, favoring companies with stronger cash flow generation and lower capital intensity. The bottom line is that the AI trade is entering a higher-conviction, lower-diffusion phase where risk-adjusted returns will be determined by the quality of capital allocation, not the size of the check.

Catalysts and Watchpoints for 2026

The thesis of a sustained, profitable AI capex cycle now faces its first major test in the numbers. The near-term catalysts are clear, and the watchpoints will separate the structural winners from the capital-intensive laggards.

First, the Q4 2025 earnings season is a critical validation point. The market's projected 14.9% earnings growth for calendar year 2026 is a lofty target that must be funded by this quarter's results. Investors will scrutinize the Magnificent 7 for signs of margin pressure directly attributable to their massive capex. The divergence is already visible: while the broader S&P 500 is seeing a beat, the Magnificent 7 underperformed last week. This sets the stage for a detailed look at whether their spending is translating into the promised revenue acceleration or merely diluting profitability in the near term.

Second, watch for further upward revisions to the AI capex consensus. The trend has been consistently upward, with the forecast for 2026 spending climbing from $465 billion to $527 billion in just a few months. Analysts have historically underestimated this spending, and any new revisions would signal that the cycle's scale is even more entrenched than currently priced. However, the market's selective rotation means that higher capex forecasts alone will not be enough to drive all AI-related stocks. The focus will be on which companies can credibly link that spending to top-line growth.

Finally, track the performance divergence between AI infrastructure and AI platform/productivity stocks as a leading indicator of market rotation. The data is already telling: the average stock price correlation among large public AI hyperscalers has collapsed from 80% to just 20% since June. This fragmentation is the market's way of pricing in execution risk. The next phase, as Goldman Sachs outlines, favors AI platform stocks and productivity beneficiaries. A sustained outperformance by these names, relative to pure infrastructure plays, would confirm a shift toward profitability and revenue linkage. Conversely, a continued rally in infrastructure stocks without a corresponding earnings boost would signal that the market is still rewarding scale over substance, a setup that could become vulnerable if capex growth slows.

The bottom line is that 2026 will be defined by selective conviction. The catalysts are in place to validate the cycle's scale, but the watchpoints will determine its quality. For institutional portfolios, the path forward is clear: monitor the earnings beat for margin sustainability, watch capex forecasts for continued upside, and use the stock performance divergence as a real-time gauge of capital allocation discipline.