Business

Meta’s AI ambitions face scrutiny over off-balance sheet risks amid soaring sector debt

Wednesday, 26 November 2025 5:27AM UTC

Meta’s extensive AI investments and innovative yet opaque financing strategies raise concerns about long-term financial stability as the sector accelerates AI infrastructure spending amid rising debt and risk.

Meta's ambitious AI investments and financial manoeuvres have drawn fresh scrutiny as the tech giant pursues dominance in artificial intelligence amid concerns about transparency and long-term risk. The company’s stock performance in 2025 tells a complex story: while Meta shares have risen by 5.7%, this growth substantially lags behind the S&P 500’s 15.2% gain, buoyed by AI leaders like Nvidia and Alphabet. Investors are increasingly cautious as Meta’s massive AI spending, including a $27 billion AI data centre project, has yet to generate sufficient revenue growth to justify the outlay.

At the heart of the controversy is the financing structure behind Meta’s $27 billion data centre in Richland Parish, Louisiana. Rather than owning the facility outright or recording the debt on its balance sheet, Meta is engaging in off-balance sheet borrowing through a joint venture called Hyperion, created with investment firm Blue Owl Capital. Meta holds a 20% stake while Blue Owl owns 80%. In October 2025, bonds worth $27.3 billion were issued to finance the project, primarily bought by institutional investors like Pimco.

This setup uses a Variable Interest Entity (VIE) structure, an approach that raises alarm bells among observers familiar with its notorious use by Enron prior to its collapse in 2001. Back then, VIEs were used to hide debt, inflate earnings, and obscure financial risks. Meta’s application of a VIE is coupled with lease accounting designed to classify its involvement as an operating lease, which effectively keeps the debt off Meta’s balance sheet. However, this classification is contentious: for an operating lease to apply, Meta should not control key activities of the venture, yet reports suggest Meta’s decisions and expertise will be pivotal to the joint venture’s success. Moreover, while Meta asserts it lacks power to direct the venture, it also faces obligations around absorbing losses or paying guarantees, which muddles the clarity of risk exposure.

Financial analysts caution that these accounting techniques might obscure the true financial burden Meta is undertaking. A residual value guarantee in the JV agreement means Meta could be liable for capped cash payments based on the data centre's value under certain conditions, notably if it chooses not to renew the lease after the initial term. As industry capital expenditures on AI infrastructure are projected to triple, potentially reaching $5 trillion to $7 trillion by 2030, this type of borrowing practice raises broader concerns about credit risk and financial stability in the sector.

Meta is not alone in leveraging heavy debt to fuel AI data centre construction. Since September 2025, hyperscalers like Amazon, Google, Oracle, as well as bitcoin miners pivoting to AI services such as TeraWulf and Cipher Mining, have collectively issued approximately $97 billion in AI-related bonds. The surge in AI financing is prompting credit analysts to raise the cost of capital for higher-risk players. Oracle and CoreWeave, another AI cloud computing firm, are among those experiencing increased borrowing costs. CoreWeave, notably, struck a $14.2 billion deal with Meta to supply cutting-edge computing infrastructure until 2031, intended to accelerate Meta’s AI model training capabilities.

This wave of capital raises the stakes for investors who must now carefully assess credit ratings and risk profiles among AI infrastructure providers to distinguish sustainable growth from precarious debt-fuelled expansion. Credit rating agencies and market observers suggest that only projects with sound financial structuring and reasonable cost of capital will likely endure, warning against unchecked growth driven by optimistic forecasts.

While Meta’s total AI expenditure remains staggering, with investments including a 49% stake in Scale AI for $14.3 billion and promising ongoing product spend of $450 million annually, the company is simultaneously restructuring its AI workforce. Around 600 employees have been laid off from Meta’s Superintelligence Labs in a bid to streamline operations and enhance decision-making efficiency. Despite these cuts, Meta's TBD Lab, focused on developing next-generation large language models, remains unaffected.

Investors, regulators, and market watchers need to remain vigilant regarding Meta’s financial engineering and broader AI capital trends. Anticipated increases in AI spending and the resulting debt accumulation accentuate the risk that weaker players may face deteriorating credit ratings and elevated borrowing costs. The spectre of off-balance sheet liabilities reminiscent of previous corporate scandals like Enron adds an additional layer of caution to the bullish narrative around AI’s transformative potential.

Ultimately, while Meta’s AI strategy could pioneer new business tools and online experiences, the opaque financing methods and looming credit risks call for a measured approach to buying the stock. As the AI sector matures and spending intensifies towards 2030, transparency and financial prudence may well determine who emerges as the genuine winners in this high-stakes race.

📌 Reference Map:

^[1] (Google News) - Paragraphs 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
^[2] (Forbes - Peter Cohan) - Paragraphs 3, 4
^[3] (Forbes - Kirk Ogunrinde) - Paragraph 7
^[4] (Forbes - Zachary Folk) - Paragraph 8
^[5] (Forbes - Shivaram Rajgopal) - Paragraph 4
^[6] (Forbes - Peter Cohan) - Paragraph 2
^[7] (Forbes - David Jeans) - Paragraph 8

Source: Noah Wire Services

More on this

https://news.google.com/rss/articles/CBMiuwFBVV95cUxNTDZXMnMwVUxTdV9qSjdRSnJDU3U4U2xFVW1QUm5GNXYzcnYyV1NCVzhmQWdsNEw3T1V5MDgyVW1IeVhIcnltQ2lwY25qd1l0T2ZZSUtlOEtfZ3NYam9kWUh3cTVBNVkxRWdXRlVXbGQwdlNNNG1FQXZ5T21qRXZhU0J3bWYwZ2lGRWxFWHhJbUpZT3FiSkVxQ1dfVmJqZThqaTZ1MTE1dTJCT2xWdG80NjZnSkZndXBpN2Y4?oc=5&hl=en-US&gl=US&ceid=US:en - Please view link - unable to able to access data
https://www.forbes.com/sites/petercohan/2025/11/25/metas-ai-gamble-why-investors-should-think-twice-before-buying-meta/ - Meta is financing a $27 billion AI data centre in Louisiana through a joint venture with Blue Owl Capital, employing off-balance sheet borrowing and lease accounting methods reminiscent of Enron's financial practices. This strategy raises concerns about transparency and potential financial risks, as Meta does not own the data centre or carry the associated debt on its balance sheet. The deal involves Meta owning 20% and Blue Owl Capital 80% of the venture, with Meta using the facilities under operating lease agreements. The arrangement includes a residual value guarantee, obligating Meta to make a capped cash payment to the joint venture based on the data centre's value if certain conditions are met following a non-renewal or termination of the lease. This financial manoeuvre has drawn parallels to Enron's use of Variable Interest Entities to conceal debts and inflate earnings, prompting scrutiny over Meta's financial practices and their implications for investors. ([forbes.com](https://www.forbes.com/sites/petercohan/2025/11/25/metas-ai-gamble-why-investors-should-think-twice-before-buying-meta/?utm_source=openai))
https://www.forbes.com/sites/kirkogunrinde/2025/09/30/coreweave-strikes-142-billion-deal-with-meta-to-boost-ai-capability/ - CoreWeave, an AI cloud computing firm, has entered into a $14.2 billion agreement with Meta to provide computing infrastructure until 2031. This deal aims to enhance Meta's AI capabilities by granting access to Nvidia's latest computing systems, expediting the training of new AI models. The partnership reflects the escalating demand for AI resources and the substantial investments being made by tech giants to bolster their AI initiatives. ([forbes.com](https://www.forbes.com/sites/kirkogunrinde/2025/09/30/coreweave-strikes-142-billion-deal-with-meta-to-boost-ai-capability/?utm_source=openai))
https://www.forbes.com/sites/zacharyfolk/2025/10/22/meta-laying-off-about-600-staff-at-ai-superintelligence-labs-heres-whats-impacted/ - Meta is laying off approximately 600 employees from its Superintelligence Labs, including staff from the FAIR AI research lab, AI product division, and AI infrastructure division. The layoffs are part of a restructuring move aimed at streamlining decision-making processes and increasing individual impact within the company. The cuts will not affect Meta's TBD Lab, which is focused on developing the company's next generation of large language models. ([forbes.com](https://www.forbes.com/sites/zacharyfolk/2025/10/22/meta-laying-off-about-600-staff-at-ai-superintelligence-labs-heres-whats-impacted/?utm_source=openai))
https://www.forbes.com/sites/shivaramrajgopal/2025/11/16/the-strange-case-of-meta/ - Meta has announced a joint venture with Blue Owl Capital to construct a $27 billion data centre in Richland Parish, Louisiana, expected to begin operations in 2029. Meta will provide construction services and property management, while Blue Owl Capital will provide financing. The joint venture will own 80% of the project, with Meta retaining 20% ownership. Meta will use the facilities through operating lease agreements with a four-year initial term and options to extend. The deal includes a residual value guarantee, obligating Meta to make a capped cash payment to the joint venture based on the data centre's value if certain conditions are met following a non-renewal or termination of the lease. ([forbes.com](https://www.forbes.com/sites/shivaramrajgopal/2025/11/16/the-strange-case-of-meta/?utm_source=openai))
https://www.forbes.com/sites/petercohan/2025/11/03/meta-stock-up-8-why-zuckerbergs-innovation-drought-may-sink-meta/ - Despite reporting strong third-quarter earnings, Meta's stock has declined by 13% since October 30, due to increased capital expenditures without clear revenue growth from these investments. The company's focus on AI and data centres has raised concerns among investors about the return on investment and the impact on profitability. ([forbes.com](https://www.forbes.com/sites/petercohan/2025/11/03/meta-stock-up-8-why-zuckerbergs-innovation-drought-may-sink-meta/?utm_source=openai))
https://www.forbes.com/sites/davidjeans/2025/06/13/scale-ai-meta-spend/ - Meta has acquired a 49% stake in Scale AI for $14.3 billion, aiming to strengthen its AI capabilities. The deal includes a provision for Meta to spend at least $450 million annually over five years on Scale AI's products, potentially constituting more than half of Meta's annual AI budget. Alexandr Wang, Scale AI's CEO, will join Meta to lead its AI lab. ([forbes.com](https://www.forbes.com/sites/davidjeans/2025/06/13/scale-ai-meta-spend/?utm_source=openai))

Noah Fact Check Pro

The draft above was created using the information available at the time the story first emerged. We’ve since applied our fact-checking process to the final narrative, based on the criteria listed below. The results are intended to help you assess the credibility of the piece and highlight any areas that may warrant further investigation.

Freshness check

Score: 8

Notes: The narrative presents recent developments regarding Meta's $27 billion AI data centre project in Louisiana, with the earliest known publication date being October 21, 2025. The report includes updated data, such as Meta's $30 billion bond issuance announced on October 30, 2025, indicating a high freshness score. However, the inclusion of older material alongside these updates suggests a need for caution. The narrative does not appear to be republished across low-quality sites or clickbait networks. Additionally, the narrative is based on a press release from Meta, which typically warrants a high freshness score. No discrepancies in figures, dates, or quotes were identified. Overall, the freshness score is high, but the recycling of older material alongside updates warrants attention.

Quotes check

Score: 9

Notes: The narrative includes direct quotes from Meta's press release dated October 21, 2025. These quotes are consistent with the original source, indicating no reuse from earlier material. No variations in wording were found, and no online matches were identified for identical quotes elsewhere. This suggests that the quotes are original or exclusive content.

Source reliability

Score: 10

Notes: The narrative originates from Meta's official press release, a reputable organisation. The press release is accessible on Meta's investor relations website, confirming its authenticity. The press release is dated October 21, 2025, and is available at ([investor.atmeta.com](https://investor.atmeta.com/investor-news/press-release-details/2025/Meta-Announces-Joint-Venture-with-Funds-Managed-by-Blue-Owl-Capital-to-Develop-Hyperion-Data-Center/?utm_source=openai)).

Plausability check

Score: 9

Notes: The claims made in the narrative align with other reputable sources. For instance, Reuters reported on October 21, 2025, that Meta has struck a $27 billion financing deal with Blue Owl Capital to fund its biggest data centre project globally. Additionally, Reuters reported on October 30, 2025, that Meta plans to raise up to $30 billion through a six-part bond offering to support its AI infrastructure investments. These reports corroborate the narrative's claims. The narrative includes specific factual anchors, such as dates, figures, and institutional names, enhancing its credibility. The language and tone are consistent with corporate communications, and the structure is focused on the main claim without excessive or off-topic detail. Overall, the plausibility score is high.

Overall assessment

Verdict (FAIL, OPEN, PASS): PASS

Confidence (LOW, MEDIUM, HIGH): HIGH

Summary: The narrative is based on Meta's official press release dated October 21, 2025, and includes updated data from subsequent reports, indicating high freshness. The quotes are original and consistent with the press release. The source is reliable, originating from Meta's official communication. The claims are plausible and corroborated by other reputable sources. Therefore, the overall assessment is a PASS with high confidence.

Meta
Artificial Intelligence
Financial risk