Rising Risks in Artificial Intelligence as Chipmakers Fund Their Own Customers

2026-07-31 companies

New York, Thursday, 30 July 2026.
Central banks warn that artificial intelligence chipmakers funding their own buyers creates major financial stability risks, as tech giants commit over $145 billion to loss-making specialized cloud providers.

Circular Financing Risks Emerge in AI Infrastructure

Central banking authorities have identified circular financing models in the artificial intelligence sector as a top global financial stability risk, alongside sovereign debt fragility and a potential AI capital expenditure bust [2]. This financing structure involves AI labs or neocloud providers receiving funding from chipmakers or hyperscalers in exchange for multi-year purchase commitments, creating concentrated dependencies [2]. Specialized neocloud vendors such as CoreWeave and Nebius have adopted these arrangements to fund costly GPU acquisitions from Nvidia (NVDA), driving soaring sales and backlog valuations despite lacking the balance sheet durability of traditional Big Tech firms [1]. Microsoft and Meta have committed approximately $122.2 billion to neocloud providers, a figure derived from Microsoft’s $60 billion and Meta’s $62.2 billion in agreements, which is equivalent to roughly 90% of the trailing twelve-month revenue of AWS [1][2].

Nvidia’s Expanding Footprint

Nvidia has intensified its infrastructure presence through significant lease agreements and financing backstops announced earlier this week. On July 27, 2026, Nvidia reportedly signed a lease agreement for Hut 8’s 1 GW Beacon Point data center campus in Texas, with a 15-year base term valued at $19.6 billion, potentially reaching $50.2 billion if all renewal options are exercised [3]. Additionally, Nvidia announced a $250 billion financing backstop for OpenAI on Monday, July 27, 2026, to support a 10-GW datacenter project in Piketon, Ohio [4]. These moves extend Nvidia’s influence beyond semiconductor manufacturing to the full-stack AI operation sector, reshaping the competitive landscape of AI infrastructure provision [3].

Market Reaction and Credit Concerns

Investor sentiment shifted markedly in late July 2026, with Nvidia’s 5-year Credit Default Swap (CDS) experiencing its largest intraday increase since active trading began in November 2025, rising approximately 14 basis points to hit a record peak of approximately 82 basispoints annually on July 24, 2026 [3]. Concurrently, the Nasdaq Index fell 5% during the week of July 26, 2026, and is down 10% from June 2026 highs, reflecting growing skepticism regarding circular financing models [4]. Market analysts have expressed concerns that Nvidia providing financing or equity to companies that subsequently purchase Nvidia chips could distort commercial incentives and amplify losses if AI demand weakens [3]. Credit insurance costs for AI infrastructure deals are rising, indicating increased systemic risk, while enterprise demand is shifting toward open-source AI models over high-cost U.S. frontier models [4].

Cash Flow Disparities Among Neoclouds

Financial data reveals significant strain within neocloud balance sheets despite revenue growth. CoreWeave reported Q1 2026 revenue of $2.08 billion, representing a 112% year-over-year increase, but free cash flow was negative $4.71 billion due to $7.7 billion in capital expenditure against $2.98 billion in operating cash flow [1]. The company’s cash balance fell 28.3% quarter-over-quarter to $2.27 billion, highlighting the liquidity challenges inherent in scaling AI infrastructure [1]. Interest expense is also rising, with Q1 2026 payments totaling $536 million, which represents 25.8% of the $2.08 billion revenue [1].

Analyst Perspectives and Sector Consolidation

Contrasting views exist regarding the sustainability of these financial structures, with UBS arguing that fears over circular financing are misplaced as chip supply chains remain the primary entities capturing cash flow [6]. UBS projects Nvidia will generate approximately $900 billion in free cash flow through December 31, 2028, suggesting suppliers are underwriting the AI infrastructure buildout [6]. Meanwhile, consolidation continues in the neocloud sector, evidenced by British AI neocloud Nscale announcing the acquisition of software startup Anyscale for $1.65 billion on July 29, 2026 [5]. This acquisition aims to integrate workload management and scaling services into Nscale’s vertically integrated AI compute stack, following Nscale’s $2 billion Series C round in March 2026 which achieved a $14.6 billion valuation [5].

Sources

Neocloud Financing AI Infrastructure Risks