Financial Leaders Warn AI Infrastructure Spending Could Bring More Capital Needs and Market Volatility

Quick Summary

• Blackstone, Goldman Sachs and JPMorgan leaders expect technology companies to seek more funding for data centres, chips and energy.
• S&P expects six major hyperscalers to spend more than $7 trillion on AI-related infrastructure between 2025 and 2030.
• Executives also warned that valuations may assume revenue growth that is difficult to provide.

Technology companies are likely to seek substantially more capital to build computing infrastructure, but the scale of the investment is raising concerns about valuations and the risk of a market correction. Reuters reported on November 3 that financial leaders speaking at the Hong Kong Monetary Authority’s Global Financial Leaders’ Investment Summit described strong demand for computing power, while warning that the commercial returns may not arrive as quickly as investors expect.

Figures cited in the Reuters report

Demand is still ahead of available capacity

Blackstone president Jon Gray said demand for computing power was running ahead of supply and pointed to the enormous requirements for data centres and energy. His comments reflect the infrastructure buildout being pursued by cloud providers and specialist operators. More computing capacity requires land, power connections, cooling equipment, networking and chips, and each part of the supply chain can become a constraint when several companies expand at the same time.

Reuters cited S&P estimates that six major hyperscalers could spend more than $7 trillion on data centres and other infrastructure between 2025 and 2030. The group includes Amazon, Microsoft, Alphabet, Oracle, Meta and SpaceX. That figure is a forecast of capital expenditure over several years, not a single spending commitment already approved by all six companies. It shows how large the infrastructure cycle could become if planned projects proceed.

Financing is becoming more complex

Large projects are being funded through a mix of corporate cash flow, bonds, bank loans, private credit and equity. Reuters noted a $63 billion financing package led by Blackstone and major banks to support the purchase of Broadcom AI chips and networking equipment for Anthropic. Transactions of this size show that the infrastructure boom is drawing capital from beyond the technology sector itself.

More borrowing can allow companies to build capacity faster, but it also creates repayment obligations before the facilities have generated their expected revenue. If customer demand slows or hardware becomes more efficient, operators could find themselves with expensive assets that are not fully utilised. The financial structure of each deal matters: investors need to understand who owns the equipment, who guarantees payments and what happens if the customer changes its plans.

Investors are not ignoring the downside

Goldman Sachs chief executive David Solomon said strong companies with credible growth prospects could still raise money, but warned that some investments would fail and investors should maintain a diversified view. JPMorgan vice chairman Daniel Pinto said he remained positive about the long-term direction of the technology but believed markets could be pricing in revenue and earnings growth that is too steep. Those comments point to a distinction between the technology’s potential and the price investors are willing to pay for it.

Asia-Pacific technology fundraising more than tripled to $125.8 billion in the first three quarters of 2026 from $36 billion a year earlier, according to LSEG data cited by Reuters. Semiconductor and infrastructure deals contributed to the increase. More capital is entering the sector, but that does not guarantee that every company will earn an attractive return. The next stage will test whether demand, pricing and utilisation can support the scale of investment now being planned.

The warning from financial executives is not that every infrastructure investment will fail. Their concern is that the market may be treating very different companies and business models as if they will all benefit equally from the spending cycle. Chip suppliers, cloud operators, data-centre landlords and software developers have different exposure to utilisation, pricing and financing costs. A project backed by a long-term customer contract carries a different risk from a speculative facility built ahead of demand. Investors will need to examine cash flow, debt maturity and customer concentration rather than relying on broad claims about the size of the market.

Investors will increasingly separate projects with signed customer commitments from speculative capacity built in anticipation of demand. Both can have a place in the market, but they carry different risks. Companies that cannot show how power, equipment, financing and customer contracts fit together may find that capital becomes more expensive as the infrastructure cycle matures.