HSBC CEO Says AI Infrastructure Is Promising, but Data Center Investments Require Selectivity

Table of Contents

• You might want to know


• Main Topic


• Key Insights Table


• Afterwards...

You might want to know

• Why does HSBC’s CEO see artificial intelligence as transformative while cautioning investors about data center projects?


• How could the race to finance AI infrastructure affect borrowers and broader financial markets?

Main Topic

Artificial intelligence is increasingly influencing how companies operate, and HSBC CEO Georges Elhedery views it as a technology with the potential to reshape the bank’s work. In an interview with CNBC’s Dan Murphy on Tuesday, Elhedery described AI as “a major enabler to re-engineer and reimagine all our businesses, all our activities.” He said the technology could give colleagues greater capacity through productivity gains and help the bank re-engineer its operations.

Elhedery characterized AI adoption as a journey rather than a completed transformation. His remarks point to a broad expectation that AI tools can change how employees handle tasks, how businesses organize their processes, and how institutions deliver services. Yet enthusiasm for the technology does not, in his view, justify investing indiscriminately in every asset built to support it.

That distinction is important for data centers, which house the servers and related infrastructure used for AI and cloud computing. Demand for these facilities has expanded alongside interest in AI, but projects differ in their commercial terms and financial strength. Elhedery advised investors to assess individual opportunities rather than treat the sector as a single, uniformly attractive investment.

“One has to be selective,” he said. Investors, he argued, should examine the quality of tenancy agreements and the pricing of the contracts supporting a project. Some agreements may provide a stronger foundation than others, while certain developments could face more difficult conditions. The central point is that demand for AI infrastructure does not automatically make every data center a sound investment.

Contract quality can shape the reliability of a facility’s income. A data center may require substantial upfront investment, while its returns depend on tenants continuing to use the space and meeting agreed payment terms. Pricing also matters: if a contract does not adequately reflect a project’s costs, risks, or market conditions, strong sector-wide demand may not be enough to ensure that the individual investment performs well.

HSBC has a significant connection to this market as both a financier and an operator of data centers. S&P Global ranked the bank as the world’s seventh-largest and Europe’s biggest lender last year, with assets worth close to $3 trillion. Headquartered in London, HSBC is also positioning itself to serve businesses involved in growing economic sectors.

In its 2025 annual report, HSBC said it intends to become “the leading bank for fast-growing transition ecosystems.” It noted that its customer base includes businesses in areas such as data centers and the AI sector. That ambition places the bank within the broader development of AI infrastructure while also making the quality of financing decisions and customer projects particularly relevant.

The scale of the expected build-out is substantial. McKinsey estimates that global spending on data centers could reach $7 trillion by 2030. Big Tech companies are increasingly turning to private credit and debt to fund new facilities, adding borrowing to the already large investment commitments associated with developing the infrastructure.

These financing patterns are becoming a source of tension in financial markets. Companies are competing to secure capital as they race to develop AI capabilities and data center capacity. At the same time, the cost of capital is rising, and corporations are adding to significant debt loads. The resulting pressure is not limited to technology businesses: it also intersects with borrowing by governments and the wider supply of bonds.

Mohamed El-Erian, chief economic adviser of Allianz and former CEO of PIMCO, discussed this competition at LSEG and Eurasia Group’s GZERO Summit in London earlier this month. He described government borrowing and refinancing at higher interest rates alongside rapidly growing financing needs among hyperscalers as a source of pressure in bond markets.

El-Erian said hyperscalers raised $200 billion in bond financing last year and were likely to raise a trillion in bond financing this year. He characterized these companies as “yield insensitive” and “spread insensitive” because they are investing on the expectation that AI will be genuinely transformational. In this context, the terms refer to the companies’ willingness to borrow even when financing costs or credit spreads may be less favorable than they would otherwise prefer.

That appetite for funding reflects how strongly some companies believe AI could change their businesses. It also raises questions about how much debt can be absorbed, how borrowing costs may evolve, and whether future income from AI-related services will support the scale of investment. The infrastructure may serve a strategically important purpose, but the financial outcome will depend on the relationship between costs, revenues, contracts, and long-term demand.

Bond markets are already facing several pressures. Bond yields issued by various major economies have hit multi-decade highs in 2026 amid concerns about elevated government deficits and rising debt levels, alongside fears of persistent inflation and more hawkish central banks. When governments and companies seek large amounts of financing at the same time, the competition can influence the price and availability of capital across markets.

JPMorgan Chase estimated earlier this year that $4.1 trillion in AI-related debt will be issued by the end of the decade. That projection reflects the expected borrowing of data center and other AI-adjacent companies as they spend heavily to build capacity. It is a forecast, not a guarantee of the amount ultimately issued, but it illustrates the potential scale of the financing challenge.

Together, these developments suggest two realities can coexist. AI may become a major driver of productivity and business change, while the infrastructure supporting it presents varied investment risks. The technology’s potential does not remove the need to evaluate whether a particular facility has dependable tenants, appropriately priced contracts, and a financing structure that can withstand changing market conditions.

For investors and lenders, a careful assessment therefore extends beyond headline demand. Relevant questions include who has committed to use the facility, how the tenancy agreements allocate risk, whether contract prices remain viable, and how the project is funded. These considerations help distinguish a project that can support its obligations from one that may struggle despite operating in a fast-growing sector.

Key Insights Table

Afterwards...

The next stage of AI development will require closer attention to both technological performance and financial resilience. Researchers and businesses can explore how AI systems deliver measurable productivity gains, how those gains are distributed across workers and industries, and what kinds of computing infrastructure are genuinely needed to support them. A clearer understanding of demand could help companies plan capacity without relying solely on broad forecasts or competitive pressure.

Infrastructure planning also calls for better analysis of energy use, grid capacity, cooling requirements, and the useful life of specialized hardware. These factors influence operating costs and the long-term viability of data centers, just as tenant commitments and contract terms influence revenue. More transparent reporting on these areas could help investors compare projects and assess risks more consistently.

Financial institutions, regulators, and market participants may also need to examine how large-scale AI borrowing interacts with other sources of demand for capital. Stress testing, careful disclosure, and monitoring the concentration of credit exposures could improve understanding of potential vulnerabilities. Such work would not determine whether AI succeeds, but it could help identify where financial risks are accumulating.

Ultimately, AI’s long-term value will depend not only on the technology’s capabilities but on whether organizations can use it responsibly and sustainably. The data center boom may create important infrastructure, yet careful evaluation remains essential: a promising sector can still contain projects with very different prospects. Exploring the links between AI adoption, sound contracts, energy systems, and durable financing will help society distinguish lasting progress from investment enthusiasm alone.