The AI cycle meets the power of predictable income

The AI cycle is entering a new phase, moving away from narrativedriven growth toward an environment defined by rising capital intensity, operational execution, and economic viability. This shift is clearly reflected in market behaviour. AI-related enthusiasm continues to dominate headlines, yet price movements have become increasingly volatile and  differentiated. Rapid rallies, abrupt sector rotations, and sharp reversals point to growing uncertainty around which business models are truly sustainable over the long term. While  some AI frontrunners display price patterns reminiscent of earlier boom phases, only a limited number of companies will ultimately succeed in translating technological leadership into durable earnings.

Dividend investors are necessarily cash centric. In this respect, the current AI cycle contains a number of elements that warrant a more cautious assessment. The unprecedented  scale of capital expenditure by large platform companies is putting free cash flow under visible pressure. In cloud and AI‑infrastructure businesses, investment outlays for data centres, compute capacity and proprietary chip development currently absorb a substantial share of operating cash flows. Depending on definition and period, particularly when looking at free cash flow after growth capex, free cash flow has temporarily turned negative for some major players, as near‑term monetisation lags the pace of investment.

While AI‑related revenues are growing rapidly, they remain modest relative to the scale of capital deployed, and for many AI‑native companies, meaningful profits and self‑sustaining cash flows are still a distant prospect. This is not a call on AI as a systemic bubble, but a reflection of a market phase that is becoming increasingly selective, where capital intensity, execution and monetisation matter more than narrative alone.

Against this backdrop, dividend yield offers a useful historical reference for how markets have rewarded companies with sustained cash generation across cycles. Following this visual support, the methodology of the S&P Global Dividend 100 Index illustrates how high yield can be combined with quality screens to avoid value traps. Dividend distributions remain a clear signal of financial health, operational efficiency and long-term stability.

According to CNBC estimates, AI capital expenditures could reach up to USD 700 billion this year alone.¹ This level of investment already exceeds the estimated USD 636 billion spent by the global oil and gas sector in 2025, according to BMI.² By contrast, revenues across large parts of the AI ecosystem remain materially lower, highlighting a fundamental imbalance between capital deployed and cash generated. This massive capital allocation fundamentally alters the economic logic of the AI cycle. Growth alone is no longer sufficient. What matters is whether companies can monetise these investments efficiently and generate stable cash flows, the necessary destination for all viable businesses.

Across the value chain, this dynamic creates substantial, but highly selective opportunities. For the time being, Nvidia and Broadcom remain key beneficiaries of rising demand for computing power. However, the increasing shift toward in-house chip development at Alphabet and Microsoft is reshaping competitive dynamics. Each newly developed internal TPU or Maia/Cobalt generation redistributes volumes and highlights how quickly dependencies on individual product cycles or customer structures can emerge. This underscores a central insight of the current AI cycle. Sustainable value creation arises where business models are not tied to specific technologies, products, or customers, but instead benefit from broad-based, structural demand.

In this context, companies that serve as infrastructural enablers of the AI ecosystem come into focus. TSMC is a prime example. While chip designers may lose market share as hyperscalers develop proprietary architectures, TSMC remains the indispensable manufacturing backbone of the global semiconductor industry. Every new chip generation, regardless of who designs it, relies on the company’s leadingedge fabrication technologies. As a result, TSMC’s economic value is driven not by the success of individual products, but by the increasing complexity and scale of the entire industry. The same structural pattern applies to equipment suppliers such as Applied Materials. Without their systems, large-scale production of advanced semiconductors would not be possible. Regardless of which chip designer dominates a given cycle, these companies benefit from rising investment levels and growing technological barriers to entry, and are able to translate this structural relevance into predictable cash flows.

As the AI cycle matures, the focus shifts not only within the semiconductor industry but also toward platform companies. As the transition from experimentation to productive deployment accelerates, scale, integration, and monetization become increasingly important. Companies such as Microsoft exemplify this phase, with diversified business models, high levels of recurring revenue, a strong balance sheet, and a track record of steadily rising dividends. Alphabet is gradually moving in a similar direction, with a clearer emphasis on regular capital returns alongside continued strategic investment. The growing importance of predictable income is by no means limited to the technology sector. In more traditional industries, the targeted use of AI is also driving efficiency gains with immediate cash flow effects. Itochu offers a compelling example. As a global trading and logistics conglomerate, the company combines stable business models with disciplined capital allocation and steadily increasing dividends. The use of AI to improve forecasting, optimize supply chains, and reduce waste enhances efficiency without fundamentally altering the company’s risk profile.

Against this backdrop, it becomes clear that dividends in 2026 are not a defensive fringe issue in the AI environment, but rather a reliable indicator of economic substance. Especially in the capital-intensive and selective AI cycle, they show which companies are able to finance growth, absorb volatility, and return capital to their shareholders, regardless of short-term market sentiment. Thus, 2026 offers real opportunities, not for speculation, but for discipline. In an environment characterized by high capital intensity, increasing selectivity, and volatile market movements, predictable returns are gaining importance as a stabilizing factor. They reduce portfolio risks, stabilize return profiles, and enable investors to participate in structural innovations such as artificial intelligence without having to rely on uncertain earnings promises. Against this backdrop, we are focusing in our Allianz Global Dividend strategy on targeted exposure to structural growth themes, combined with a consistent focus on sustainable cash flows and reliable capital returns. Our portfolio alignment follows precisely this approach. We invest in companies that can translate growth in the AI cycle into rising cash flows and dividends. In this way, we achieve a distribution yield above the market without unduly overstretching the risk-return profile.