Investors are shifting the AI debate from whether massive infrastructure spending is justified to which companies can turn it into durable returns, even as the Bank for International Settlements warns that heavy spending, elevated valuations and opaque financing structures could spread risk beyond stocks into corporate credit. Alphabet, Amazon, Meta and Microsoft are expected to spend a combined $740 billion on AI computing infrastructure in 2026, potentially rising to $1 trillion in 2027, while some data-centre and chip commitments remain outside reported capex and balance sheets. Recent Microsoft and Amazon results reassured markets that cloud growth is accelerating and AI capacity remains tight, supporting hyperscalers and semiconductor suppliers, but pressure is building as some companies rely more on debt, equity, leasing and sale-and-leaseback structures amid weak free cash flow and the prospect of higher long-term U.S. interest rates.