Navigating Global Uncertainties: Fed Policy Paths and the AI Capex Cycle

In a recent global tour across Europe and Asia, Serena Tang, Global Head of Cross-Asset Strategy Research at Morgan Stanley, engaged with investors to distill the prevailing market sentiments. The discourse was overwhelmingly dominated by two critical pillars: the trajectory of the Federal Reserve’s monetary policy and the structural evolution of the Artificial Intelligence (AI) capital expenditure (CapEx) cycle.

The Fed’s Monetary Policy: Deciphering the Hawkish Signal

A significant portion of investor concern centered on the Federal Reserve’s June FOMC meeting. Many market participants interpreted the meeting—the first under Chair Kevin Warsh—as "unambiguously hawkish." This sentiment was primarily driven by the Fed’s "summary of economic projections" (SEP), which signaled an additional hike in 2026 and a delay in anticipated rate cuts, suggesting a "more restrictive policy" stance.

However, Morgan Stanley’s economists advise caution against reading too deeply into these projections. They argue that the SEP forecasts are conditioned on "elevated near-term inflation" and likely fail to account for the "disinflation from a straight reopening." Morgan Stanley’s internal outlook remains contrarian to the hawkish market interpretation, anticipating a "lower path for core inflation." This expectation is fueled by a "reversal in travel-related inflation" and "tariff payback," reinforcing their base case that the Fed will remain "on hold through 2026."

The AI Paradigm: Chipflation and Funding Dynamics

Beyond central bank policy, the AI sector remains a focal point of global investor attention. While the "secular story of ongoing AI capex cycle" is widely accepted, recent conversations have shifted toward the macro-economic implications of "chipflation" and the micro-economic challenges of funding such massive infrastructure.

Addressing the "Chipflation" Debate

"Chipflation" has emerged as a buzzword, sparking debates on whether rising costs could "derail the AI capex cycle." Morgan Stanley’s analysis provides a more nuanced perspective. While acknowledging that "memory prices is up sixfold over the past year," the firm believes this phenomenon acts as a mechanism to "reprice and ration AI infrastructure" rather than a catalyst for cycle collapse. The demand for AI is scaling rapidly across three distinct layers: "more memory per chip, more chips per system and more systems per cluster," with hyperscalers maintaining their priority status in the "allocation queue."

The Mechanics of AI Funding

As hyperscalers expand their data center construction, the demand for capital has surged. Data indicates that a "majority of corporate bond issuance" this quarter is earmarked for these infrastructure projects. To sustain this, hyperscalers are diversifying their funding sources, evidenced by the issuance of roughly "25 billion of debt in other currencies" such as the Euro, Swiss franc, and JPY.

The scale of this funding is set to grow significantly. Morgan Stanley’s credit strategy team forecasts an additional "600 billion of AI-related global issuance in 2026," contributing to an expectation of "1 trillion of net issuance" for US Investment Grade (IG) corporate bonds. This massive supply is a primary reason why the firm suggests that the asset class may "underperform this year." Furthermore, with equity analysts projecting hyperscaler cash CapEx to exceed "1 trillion in 2027," the acceleration of debt issuance appears inevitable.

Conclusion: A Constructive Outlook

Despite the prevailing uncertainties regarding the Fed’s restrictive stance and the pressures of AI-related CapEx, Morgan Stanley maintains a "constructive" outlook on risk assets. The firm’s base case rests on the belief that disinflationary trends will keep the Fed on the sidelines and that the fundamental AI investment cycle remains "intact." Investors are encouraged to look past the short-term noise of market debates and focus on the underlying structural strength of these two dominant themes.