AI Capex: It’s Not All or Nothing

Recent AI-related volatility across risk markets mixes signal with noise. The more important task is to recognize that we are in a multiyear buildout and to invest with that sequencing in mind.

Bubble or not, AI is already causing profound economic change, and this is only likely to accelerate irrespective of the risks to its buildout over the short and long term. Particularly now that debt is part of the equation in financing capital expenditures, investors increasingly are applying the “dot-com” experience as a point of comparison and reason for caution.

Yet we continue to believe that analog—that AI would play out as a grand boom/bust cycle across a wide swath of tech companies—is too binary. Today’s AI capital cycle is more layered: established, cash-generative businesses are integrating AI into existing offerings; newer, more capital-intensive models are scaling; and a widening set of capital providers will share the claim on returns.

That mix lifts hurdle rates and demands greater investor discipline, but it also makes the system more resilient. The destination—aggregate returns on capital—will be uneven. The journey—investment, jobs, output and productivity—is already visible and underway.

From ROE to ROIC

An important element in this journey is recognizing the shift we are seeing from return on equity (ROE) to return on invested capital (ROIC). Debt issuance tied to AI has ramped significantly—some of the hyperscalers have raised over $90 billion in the past few months alone—pulling creditors into the capital structure at significant size and lifting hurdle rates as interest expense replaces the recent tailwind of interest income. Indeed, capital structures will span unsecured corporate debt, asset-backed vehicles and vendor financing; cost-of-capital engineering therefore becomes a core competency.

For multi-asset investors, the choice is where to participate along that capital stack and what compensation to demand. Equity offers upside, but carries heightened sensitivity to execution, utilization and competitive shocks. Credit can benefit from covenants, collateral and seniority, particularly where large capex is anchored by durable free cash flow. Private markets and hybrids add further dimensions for calibrating exposure to timing, cash flow and downside protection.

A Revolution in Three Acts

Sequence matters. As capacity is being added, data centers are being built, power is being contracted, and specialized hardware is shipping, the rate of adoption will broaden out at different speeds. Company-level differentiation therefore becomes important to identify, an equities market theme we highlighted in our Solving for 2026 thought leadership campaign. In practice, a diversified incumbent firm extending proven products with AI faces a different financing and execution profile than a firm leaning heavily on external capital; investors should expect dispersion in both timing and outcomes.

Against this backdrop, we see three phases. First, the buildout: infrastructure spend quickly converts into revenue for chipmakers, component suppliers and power providers. Second, adoption: monetization shifts toward software and services, with potential revenue streams in the trillions. Third, outcomes: durable returns emerge only when customers can point to tangible gains from AI—lower unit costs, higher quality, faster delivery—and are prepared to pay for them. The first act is well underway, but the execution of acts two and three will take time to observe and quantify.

Macro Tailwinds from the Capital Cycle

Importantly, the quantum of AI capex we are seeing is a genuine growth and productivity driver, a reality we highlighted in our earlier piece AI: Boom? Bust? Or Both? 

The result is a broad-based impulse that touches manufacturing, construction, utilities and professional services, not just the technology sector.

Early readings are consistent with that view. Investment in data centers, power infrastructure, specialized hardware and software deployment is visibly charging economic activity—tech and AI capex contributed 50% of U.S. economic growth in the second quarter—and supporting measured productivity improvement. With announced projects in flight and order books still robust, we expect this momentum to persist over the next 12 to 24 months, even as the mix of beneficiaries evolves from hardware and power toward software and services.

For markets, the near-term effect is expected to be higher nominal revenues and improved operating leverage for the suppliers most directly tied to the buildout, creating a cyclical tailwind for risk assets. That impulse is arriving alongside fiscal support and a gradual easing in financial conditions, reinforcing the backdrop.

Investment Implications

AI’s macro effects will be significant, but the sharper impact may come through accelerated creative destruction at the company and industry level, heightening the need for selectivity in portfolio construction. As with prior technology waves, there will be clear winners and losers. Active, bottom‑up monitoring and disciplined security selection will become even more important as AI adoption deepens.

Diversification, as ever, remains essential. Rather than a single “AI mega-theme,” we expect multiple micro-themes across value chains and end markets. Given uncertainty around ultimate winners, investors should maintain diversified exposure to high‑quality businesses across both equity and credit.

Capital discipline will also be a key differentiator. Rising ROIC expectations and heavier investment needs demand close attention to capital structure integrity. Companies with marginal balance sheets or limited capacity to fund capex from organic cash flow may be vulnerable as the cycle evolves. Even sound strategic bets can be undermined if leverage and financing costs render the capital structure unsustainable.

Finally, valuation frameworks should broaden. As business models shift from asset‑light to more capital‑intensive, simple multiples like price/earnings will be less informative. Measures that incorporate the full balance sheet, such as enterprise value to cash flow or EBITDA, can better capture capital intensity.

Leaning Into the Sequencing

Amid frequent headline risk and competing narratives, the more durable signal is a multiyear capital cycle that is already reshaping cash flows, balance sheets and competitive dynamics. For investors, the opportunity is to lean into that sequencing—backing infrastructure today, positioning for software and services monetization next and insisting on demonstrated customer outcomes over time—while staying disciplined on ROIC, capital structure and valuation.

To read our series of insights related to this topic, How AI Is Reshaping Credit Markets, please click here.