Blackstone is preparing an investment of about $9 billion in subordinated debt related to chip financing for artificial intelligence company Anthropic. The yield on that part of the deal is above 10% - a level that shows how high a risk premium investors demand for exposure to the fast-growing but highly capital-intensive sector.
The deal is part of a larger package of about $60 billion organized by Broadcom. According to information cited by Reuters, the debt will be secured by five-year lease payments to Anthropic for the chips. About $42 billion is set to be senior debt guaranteed by Broadcom, with the rest riskier and more expensive to finance.
The financing has not yet closed. Some of the subordinated debt could be offered to other investors, with terms likely to depend on interest in the deal. The Financial Times, quoted in the transaction reports, points out that companies with similar credit profiles typically pay about 7% in public bond markets. A spread of more than 10% signals doubts about future revenues, the value of chips and the ability of AI companies to service their huge debts.
Banks finance AI infrastructure
Blackstone and other private lenders are becoming an important source of capital for data centers, power plants and specialized processors. In September, a group of 10 banks secured a $22 billion loan for Crux AI, a cloud company affiliated with Blackstone and Alphabet. Blackstone has committed to an initial $5 billion equity investment, and the company plans to deploy 500 megawatts of data center capacity by 2027.
The size and structure of these deals transfer some of the risk between technology companies, banks, private equity funds and bond investors. If demand for computing power turns out to be weaker than expected or the price of chips falls rapidly, the collateral could lose value and lenders could suffer losses.
Regulators also monitor cyber risk
The pressure on banks comes not only from the lending side. U.S. regulators are examining how financial institutions use autonomous AI systems, how they limit their actions and whether their decisions can be tracked and audited. In Europe, the European Central Bank has asked lenders to prepare plans against cyber threats related to artificial intelligence.
“No one knows how capital and computing power needs will evolve as the broader AI ecosystem begins to mature,“ said Goldman Sachs CEO David Solomon.
The Anthropic funding thus becomes a concrete test for the market. It should show whether banks and private investors will continue to lend billions for AI infrastructure at a high price of risk or will start to demand tighter collateral, shorter terms and greater certainty about future revenues.
Sources: reuters.com