A Leap of Faith Now Needs the Bond Market…

Many view the rise of generative artificial intelligence (AI) as a key driving force behind the next industrial revolution, where the integration of AI, robotics, and increased interconnectivity permeates all industries, driving transformational changes in how humans live and work. While only time will tell the outcome of this bold experiment, current investors—and this generation as a whole—are facing seismic changes that will come with the broad proliferation of AI, and moving forward will require a significant “leap of faith.”

According to JP Morgan Research, USD 5 trillion of capital is needed by 2030 to fund massive infrastructure investments in AI-related semiconductors and data centers, which have demanding power requirements. Unlike the buildout of railroads in the 19th century that lasted decades after construction, and the durability of internet infrastructure put in place at the end of last century, AI requires constant maintenance, as the hardest working semiconductors in a data center may only last three to five years based on use. And yes, semiconductor-related technology will continue to advance, but this reality only feeds a taxing maintenance spend for AI, as there will be incentives to use only the data centers with the latest and most advanced semiconductors. All of this arguably creates a much less cyclical “super cycle” for today’s leading semiconductor companies.

Fortunately, as any good commercial lender appreciates, even with the uncertainty that comes with the current “leap of faith” in AI, the integrity of the project in the U.S. has been bolstered by the sheer scale of how much this expensive buildout has been funded from existing cash flows by its leading “hyperscaling” companies. But even these traditionally asset-light companies that generate large levels of free cash flows and have multi-trillion-dollar market caps have limits as to how much they can deploy if they are to maintain and grow their equity valuations from current levels. They, and AI infrastructure players in general, will need to borrow money in the investment-grade credit market to see this grand project through to a more mature maintenance-level state.

It is against the backdrop described above that the T. Rowe Price Fixed Income Division held its November Policy Week meeting, where the following expectations emerged for the foreseeable future:

• A resilient global economy, as the global AI arms race continues and fiscal policy dominates,
• One more rate cut from the Fed in December, but no cuts in the first half of next year as the jobs market remains rangebound while higher U.S. inflation potentially peaks at 3.5% next fall,
• Away from fed funds rate and shorter maturity U.S. Treasuries, higher U.S. rates and a steepening  Treasury yield curve as the U.S. aggregate debt pile continues to grow,
• Caution on investment-grade (IG) debt against the backdrop of tight spreads and the rapid expansion of expected AI-related IG debt coming to market, and
• A weaker U.S. dollar that could be supportive of emerging market debt.

Beyond these considerations, the Fixed Income Division also deliberated on the quickly rising scale of complex IG debt that will be needed to fund the historically large AI infrastructure arms race referenced above that has just begun. Just as shopping mall or golf course developers must finance their projects with borrowed money and manage the associated interest expenses while waiting for the asset to eventually generate cash flows to repay the debt, the current AI buildout will increasingly rely on structured debt to fund its massive infrastructure investments, and these debt offerings may be highly complex given the relationships between the companies involved in the rapidly accelerating AI industry.

While uncertainty abounds with now more and increasingly complex AI debt-deal structures that will combine joint venture project finance, we believe there will be opportunity for asset managers who can analyze and understand the highly complex deal structures that may be present as companies issue debt in the IG corporate market to fund infrastructure spending. To this end, a yield premium has emerged based on the difference in spreads between the technology component of the Bloomberg U.S. Investment Grade Corporate Bond Index and the full Bloomberg U.S. Investment Grade Corporate Index.

Bottom Line—Just as the T. Rowe Price Fixed Income Division avoided the esoteric and complex deal structures that ultimately helped trigger the global financial crisis of 2008, today’s fixed income platform remains predicated on fundamental research, enjoys close collaboration for greater perspective with T. Rowe Price’s dedicated technology equity team, and is prepared to navigate the uncertainty and opportunity that accompany today’s historic AI buildout.