The Nobel Prize–winning physicist Richard Feynman famously observed that no one really understands quantum mechanics1. The economics of artificial intelligence may be heading towards a similar distinction. Contrary to popular belief, investors can be a passionate bunch, and few topics in recent memory have sparked more debate than financing the AI buildout.
Artificial intelligence has extraordinary potential, but there is considerably less agreement about how much revenue it will ultimately generate, who will capture the profits, and whether the returns will justify the trillions of dollars being invested. For credit investors, these questions are becoming increasingly important as the world’s largest technology companies turn to debt markets to help finance an unprecedented expansion in computing infrastructure, creating new opportunities and risks for bondholders.
The long-term potential applications of AI are almost limitless. Companies cite benefits ranging from accelerating drug discovery, to the use of robotics in manufacturing and industrial automation, and digital sales assistants that tailor recommendations to customers. The clearest commercial opportunities today are in software and enterprise productivity. Here, use cases include coding assistants for software development, automating enterprise tasks such as legal analysis, fraud detection and medical diagnostics, as well as handling customer enquiries and sales processes.
- Investment in artificial intelligence (AI) infrastructure is beginning to exceed internally generated cash flow available to fund it. Our analysis indicates an external financing requirement of approximately USD 1.1 trillion for the five major US technology companies (Alphabet, Amazon, Microsoft, Meta and Oracle), creating an expanding opportunity for credit investors.
- Computing infrastructure providers appear well positioned to capture AI revenues, but attractive returns depend on demand, pricing discipline, and the scale and timing of capital investment. Strong starting balance sheets provide support, although borrowing will increase leverage.
- Senior unsecured hyperscaler bonds, data centre and AI chipbacked financing, and utility hybrids offer opportunities to capture additional income. Security selection and fundamental research are essential to assess whether investors are adequately compensated for the risks as financing structures become more complex.