“Loans Tighten, Tech Stocks Plunge”: Wall Street Reluctant to Lend Even With Google Guarantees, Casting Clouds Over Anthropic IPO Amid AI Funding Chill

Wall Street’s caution on funding constrains AI investment expansion
Equity financing conditions worsen amid long-term payment obligations
Anthropic faces major fundraising test ahead of listing

Wall Street, which had been pouring money into artificial intelligence (AI) investment, is raising lending standards, putting the brakes on data center expansion. Some banks are scrutinizing repayment prospects so closely that they are reluctant to extend new loans even for deals guaranteed by Big Tech companies such as Google. The AI industry, which has increased capital spending while taking on long-term rent and guarantee obligations, urgently needs additional funding, but plunging tech stocks have also worsened conditions for raising money through equity markets. Anthropic, which had planned to finance investment through a large initial public offering (IPO), also faces concerns that falling investor valuations will make it difficult to maintain its target offering price.

Wall Street Tightens AI Loan Reviews

According to Reuters on the 10th (all dates hereafter are local time), major U.S. banks and credit investors have recently tightened their scrutiny of AI-related loans and become more cautious about providing additional funding. Some banks are reportedly reluctant to extend new loans even for transactions backed by guarantees from Anthropic or Google. Their reviews include testing cash flows under recession scenarios and examining data center commissioning deadlines, service obligations, and conditions under which major customers can reduce or terminate contracts. In other words, the process of verifying whether rent and computing-service revenue—the sources of loan repayment—will arrive as contracted has become considerably more stringent.

The financial sector’s conservative stance is also evident in collateral valuations. Some lenders have demanded expanded guarantees for Nvidia’s USD 500 billion AI financing initiative, saying it is difficult to assess the long-term collateral value of its chips. Loans secured by graphics processing units (GPUs) are typically assessed using a depreciation period of 3–4 years, as there is insufficient data to estimate their profitability and resale prices beyond that period. With uncertainty over whether selling the collateral alone would recover the loans, financial institutions are also emphasizing long-term usage contracts with creditworthy customers as a source of repayment. Indeed, the stability of payments Meta is contractually obligated to make was a key factor behind the investment-grade rating assigned to CoreWeave’s USD 8.5 billion GPU-backed loan. This means that lenders, increasingly cautious about assessing chips’ residual value, are also closely examining customers’ creditworthiness and contractual payment obligations.

Alongside raising the bar for new loans, banks are continuing efforts to share the risks of loans they have already underwritten with outside investors. Concentrated lending to particular companies and the AI sector has reduced their capacity to provide additional funding. According to the Financial Times (FT), major banks spent more than 6 months distributing USD 38 billion in construction loans for data center projects leased by Oracle in Texas and Wisconsin to other investors, with some even pursuing sales of the loans at a discount. They are also reportedly considering arrangements in which they retain the loans but pay private credit funds, insurers and others to assume a portion of the risk of losses.

AI Funding Proves Difficult Even With Google Guarantees

Banks’ efforts to reduce lending risks are also weighing on an AI investment model that has expanded on the strength of Big Tech guarantees. One example is a transaction in which Google backs Anthropic’s data center rent payments. According to an agreement disclosed on the 4th by AI data center infrastructure company TeraWulf, Google agreed to provide credit enhancement for Anthropic’s 20-year, USD 19 billion lease of a data center in Kentucky. Under the arrangement, if Anthropic fails to pay rent or becomes insolvent, Google would pay the outstanding rent and take over as tenant, or make a termination payment under certain conditions. The developer plans to use this backing to secure construction financing, while Google would assume contractual obligations if Anthropic’s ability to pay deteriorates. Some banks’ reluctance to extend additional loans even for deals backed by Google’s credit is seen as an effort to manage their total lending exposure to the AI sector, including the companies providing guarantees.

The potential obligations accumulating at companies providing guarantees are also substantial. According to Google parent Alphabet’s third-quarter filing, credit support commitments backing data center rent and certain electricity costs totaled USD 75.7 billion as of the end of September, including obligations to provide support in the future. Of that amount, USD 48 billion consisted of commitments already signed with third-party data center operators, reflecting an accumulation of contracts that could turn AI companies’ payment difficulties into financial burdens for their guarantors. In particular, because rent payments across multiple projects depend on the revenue of a small number of AI companies, guarantees extended separately across individual contracts carry a significant risk of becoming payable at the same time.

Yet Anthropic, the company that will actually use the facilities, recorded an operating loss of USD 567 million due to computing costs and other expenses, despite third-quarter revenue rising 47% from the previous quarter. Continued operating losses, even as the company increases long-term spending commitments in anticipation of revenue growth, are fueling concerns that facility expansion dependent on borrowing and guarantees could lead to overinvestment. Moreover, if utilization of computing resources secured through long-term contracts falls short of expectations, reliance on external funding to meet committed payments will increase further. In its securities filing, Anthropic also explicitly identified the risk that constraints on infrastructure partners’ external financing or credit support could delay or reduce the supply of contracted computing resources or lead to price adjustments. In other words, if financing difficulties delay facility delivery, they could also hamper revenue generation through service expansion and push back the timeline for recovering investment costs.

Table 1. Guarantee and Rent Obligations in Big Tech AI Investment

Big Tech AI Guarantees Reach Up to USD 300 Billion

Warnings that guarantee-backed investment expansion is increasing Big Tech’s potential losses began emerging months ago. According to the FT, residual value guarantees that technology companies had provided or committed to provide through financing programs over the preceding year totaled up to USD 300 billion. Under such arrangements, a special-purpose vehicle (SPV) borrows to acquire data centers and semiconductors, while a technology company covers the difference between the assets’ disposal price and an agreed minimum value. The guarantee structure used for Meta’s data centers was also applied to financing for AI chips supplied by Broadcom to Anthropic. This has raised concerns that weak AI demand coupled with falling equipment values could trigger guarantee obligations across multiple projects simultaneously.

As guarantee obligations grow, long-term payment commitments that do not appear in financial statements are also accumulating. In February, global credit rating agency Moody’s warned that renewal periods and residual value guarantees in data center leases were not adequately reflected in reported lease liabilities. If renewal is judged not to be reasonably certain, rent for extension periods is excluded from liability calculations, while recognition of residual value guarantees also varies depending on the likelihood of payment. At the time, Moody’s calculated that major large cloud companies had USD 969 billion in future rent commitments, of which USD 662 billion was not reflected on balance sheets because the leases had not yet commenced. As facilities enter operation in stages, related liabilities and rent payments will increase, meaning deteriorating AI business profitability could pressure both investment capacity and cash flow.

AI Fundraising Conditions Worsen

Big Tech’s funding burden does not end with guarantees and rent. Huge outlays on their own AI capital investment are also driving increased external fundraising. In June, Alphabet conducted a public offering of common stock, a private placement to Berkshire Hathaway affiliates, and an issuance of mandatory convertible preferred stock. Net proceeds listed in its third-quarter filing were USD 20.5 billion, USD 10 billion and USD 19 billion, respectively, and the company included capital spending on AI infrastructure and expansion of global computing capacity among the uses of the funds. It also established a program to sell up to USD 40 billion in shares into the market from time to time, raising USD 12.4 billion by the end of September. With Big Tech both backing customers’ long-term contracts and tapping equity markets to finance its own investment, investors’ capacity to buy shares is also influencing its funding management.

Recent share-price declines, however, are adding pressure to efforts to raise additional funds through stock issuance. Alphabet shares fell about 19% during October and plunged more than 12% in a single day on the 29th of last month, immediately after its earnings announcement. Raising the same amount at a lower share price requires issuing more shares, increasing dilution for existing shareholders. In particular, because the at-the-market sale program depends on market prices at the time of actual sales, a prolonged stock market downturn makes it harder to determine the size and timing of additional issuance.

Downward Pressure Mounts on Anthropic’s IPO Price

Tighter loan reviews by financial institutions and plunging tech stocks are also dealing a direct blow to Anthropic’s listing as it prepares to raise substantial funds. In a securities filing submitted on the 10th, Anthropic set a target offering price of USD 70–78 per share and said it would offer 825 million shares, including shares sold by existing shareholders. At the top of the range, the offering would total USD 64.4 billion, with a fully diluted valuation of USD 1.4 trillion accounting for stock options and other instruments. Although that is below the USD 2 trillion some investors had previously hoped for, investors remain wary of its massive infrastructure spending commitments. Some prospective investors reportedly questioned even the previously discussed USD 1.5 trillion valuation, citing enormous infrastructure spending commitments and volatility in AI-related stocks.

Investors have also moved to lower their valuations of Anthropic during their investment reviews. Potential investors discussed valuations of around USD 1.2 trillion–1.3 trillion, with some considering figures as low as USD 1.1 trillion. Investors have revised their desired purchase prices following the sharp decline in AI-related stocks, suggesting that demands for a lower price could persist during the offering process. A lower offering price would reduce the funds the company raises from issuing the same number of new shares, making it harder to finance long-term investment. Based on the currently proposed offering volume, every USD 1 reduction in the per-share price would cut total offering proceeds by USD 825 million. With financial institutions tightening lending, if the listing also raises less than expected, the company would have to seek financing again to meet infrastructure spending commitments it has already made.