HSBC CEO warns the AI boom won’t make every data center a good bet
HSBC CEO Georges Elhedery advises caution on data center investments, emphasizing the need to scrutinize tenant agreements and contract pricing amid a boom fueled by AI.
Intelligence analysis by Gemini 2.5 Flash Lite
While acknowledging AI's transformative potential, HSBC CEO Georges Elhedery urges selective investment in data centers, warning that not all projects will be profitable due to varying contract terms and tenant quality.
Imagine building a giant computer house for AI. Everyone wants to build them because AI is super cool! But just like building any house, it costs a lot of money. The boss of a big bank says we need to be careful and check who is renting these computer houses and how much they are paying, because not all deals are good deals.
Analysis
The AI Infrastructure Gold Rush
The artificial intelligence revolution is undeniably a major enabler for re-engineering businesses and boosting productivity, as described by HSBC CEO Georges Elhedery. This transformative potential is driving a massive surge in demand for the physical infrastructure that powers AI: data centers. McKinsey estimates that global spending on these facilities could reach a staggering $7 trillion by 2030. This projection underscores the immense opportunity, but also the significant capital required to build and maintain these complex operations. Big Tech companies, in particular, are aggressively tapping into private credit markets and issuing substantial amounts of debt to fund the rapid expansion of their data center campuses.
Financing Pressures and Yield Insensitivity
However, this frenzied build-out is creating significant tension in financial markets. The cost of capital has soared, and corporations are accumulating massive debt piles to finance AI development and data center construction. Mohamed El-Erian, chief economic adviser at Allianz, points to increased competition for financing, with governments also borrowing heavily and refinancing existing debt at higher rates. Adding to this pressure, hyperscalers are reportedly raising hundreds of billions, potentially a trillion this year, in bond financing. Crucially, these tech giants are described as 'yield insensitive' and 'spread insensitive' because they are betting on AI's truly transformational nature, a characteristic that can mask underlying financial risks for lenders and investors if the AI revolution doesn't materialize as expected or if their financing strategies become unsustainable.
The Need for Selective Investment
Elhedery's core message is one of discernment. He stresses that not all data center investments are created equal and that a selective approach is paramount. Investors and financiers must meticulously examine the quality of tenancy agreements – who are the long-term lessees, and what are the terms of their contracts? Equally important is the pricing of these contracts. Some deals may offer robust returns and strong support, while others could prove more challenging to sustain, especially in an environment of rising interest rates and increased competition. HSBC, a major global bank with significant assets and a presence in data center operations, is positioning itself as a leader in financing transition ecosystems, including data centers and AI. Yet, even as it embraces the AI journey, its CEO's caution serves as a critical reminder that the underlying economics of infrastructure projects, even those supporting revolutionary technology, require rigorous financial scrutiny.
Key points
- HSBC CEO advises selective investment in data centers despite the AI boom.
- Data center deals must be evaluated based on tenant quality and contract pricing.
- Global data center spending is projected to reach $7 trillion by 2030.
- Hyperscalers' significant bond issuance adds to financing competition and debt levels.
- Rising interest rates and inflation concerns are increasing the cost of capital.
If the AI boom continues to drive sustained demand for data processing and storage, well-structured data center projects with high-quality, long-term tenant agreements could yield significant returns. This could lead to a robust expansion of digital infrastructure, supporting further technological innovation and economic growth.
The race to build data centers, coupled with high borrowing costs and potentially yield-insensitive hyperscalers, could lead to overcapacity or financially strained projects. If AI adoption or its profitability falters, some data centers may struggle to secure tenants or profitable contracts, leading to financial losses for investors and lenders.
