The race to build artificial intelligence infrastructure is fundamentally changing the financial profile of the world’s largest technology companies, with the industry’s unprecedented spending spree eroding free cash flow, increasing leverage and introducing new balance-sheet risks, according to Moody’s Ratings.
In a research note published this week, the ratings agency warned that the shift toward AI is forcing even Silicon Valley’s most financially resilient companies to abandon the asset-light business models that underpinned decades of exceptional profitability and instead embrace capital-intensive strategies more commonly associated with utilities and industrial manufacturers.
The warning comes as hyperscalers including Microsoft, Alphabet, Amazon, Meta, Oracle and CoreWeave collectively spend hundreds of billions of dollars building AI data centers, purchasing advanced chips and securing the power infrastructure needed to support increasingly sophisticated AI models.
“Previously, these companies relied on asset-light structures centered on software, intellectual property, and scalable cloud services that required modest capital investment,” Moody’s said. “The transition from asset-light to asset-heavy models requires unprecedented levels of investment and capital raising.”
According to Moody’s, the aggressive expansion “threatens credit quality” across the six companies it tracks, although the immediate risks vary significantly depending on each firm’s financial strength.
The ratings agency projects that capital expenditures by the group will reach $785 billion in 2026 before climbing to approximately $1 trillion annually in 2027, underscoring the extraordinary scale of investment now flowing into AI infrastructure.
The forecast indicates that generative AI has overturned Silicon Valley’s traditional economic model.
For decades, software companies generated exceptional returns because their products could be replicated at virtually no cost after development. AI, however, requires a massive physical footprint consisting of specialized data centers filled with thousands of high-performance graphics processing units (GPUs), networking equipment, storage systems and expensive electricity infrastructure.
Unlike software, those assets require continuous investment, shortening replacement cycles and consuming enormous amounts of capital before meaningful revenue is generated.
That dynamic is already weighing on one of Wall Street’s most closely watched financial metrics: free cash flow.
Moody’s noted that AI infrastructure demands significant upfront investment while revenue from AI services accumulates over a much longer period, creating pressure on cash generation even for companies reporting record earnings.
As a result, hyperscalers are now turning to external financing to sustain their AI ambitions.
Direct debt across the six companies has climbed to roughly $460 billion, according to Moody’s. Companies are also raising capital through equity markets. Alphabet recently announced an $85 billion stock offering, one of the largest equity raises ever undertaken by a technology company, highlighting how even cash-rich firms are seeking additional financial flexibility to fund AI expansion.
Beyond traditional borrowing, Moody’s highlighted a rapidly growing source of financial exposure that receives far less attention from investors: off-balance-sheet obligations. Rather than owning every new AI data center outright, hyperscalers are increasingly signing long-term leases with specialized infrastructure developers.
Those arrangements allow companies to avoid recording the facilities as conventional debt, but Moody’s considers the lease commitments economically equivalent to borrowing because they create long-term contractual payment obligations.
According to the report, lease commitments across the six companies have surged to $1.2 trillion, with more than $820 billion tied to facilities that have not yet entered service and remain under construction. As those projects come online over the coming years, the associated lease payments will become recurring financial obligations regardless of fluctuations in AI demand.
The growing reliance on leased infrastructure also reflects the emergence of a new financing ecosystem around artificial intelligence. Instead of building every facility themselves, technology companies are relying on specialized developers, private equity firms, infrastructure funds and real estate investment trusts to finance, construct and operate AI campuses before leasing them back under long-term agreements.
This approach accelerates deployment but shifts a substantial portion of future financial commitments away from traditional balance-sheet debt.
Despite these concerns, Moody’s stressed that the largest hyperscalers remain among the strongest corporate borrowers globally. Microsoft, Alphabet, Amazon and Meta continue to maintain exceptionally strong balance sheets, substantial liquidity and resilient cash generation from mature businesses such as cloud computing, digital advertising and enterprise software.
Consequently, Moody’s does not believe their investment-grade credit ratings face immediate pressure.
Instead, the greatest financial vulnerability lies with companies operating closer to the lower end of the investment-grade spectrum.
Oracle, which has dramatically expanded AI infrastructure spending in an effort to compete with larger cloud providers, carries a Baa2 credit rating with a negative outlook, leaving it only two notches above speculative, or junk, status.
CoreWeave faces even greater financing challenges.
The AI cloud provider operates with a Ba3 high-yield rating and depends heavily on complex private debt structures to finance massive fleets of Nvidia GPUs, making it significantly more sensitive to changes in financing costs or shifts in investor sentiment.
Moody’s also identified what it described as a growing structural circularity within the AI economy.
Many of the largest cloud providers have invested billions of dollars in leading AI developers such as OpenAI and Anthropic. Those same AI companies then spend billions leasing computing capacity from the cloud providers that financed them, creating an ecosystem in which capital, infrastructure and revenue increasingly circulate among a relatively small group of companies.
While those arrangements have helped generate enormous AI backlogs for cloud providers, Moody’s warned they also create concentration risk because much of the industry’s future growth depends on the same customers, the same infrastructure providers and similar assumptions about long-term AI adoption.
Should enterprise demand for AI services grow more slowly than expected, or should pricing for AI computing come under pressure, those interconnected relationships could amplify financial stress across multiple companies simultaneously.
Nevertheless, Moody’s believes several factors continue to support the sector. Demand for AI computing remains robust, hyperscalers continue to report strong growth in cloud businesses, and many have secured hundreds of billions of dollars in long-term customer contracts that provide visibility into future revenue.
Those strengths help offset concerns surrounding the current investment cycle. Still, the ratings agency argues that investors should recognize that the economics of Big Tech are undergoing one of the most significant structural transformations in decades.
For years, investors rewarded technology companies for producing extraordinary cash flows with relatively modest capital requirements. The AI era is reversing that equation, requiring companies to commit unprecedented amounts of capital years before realizing full economic returns. As a result, future market leadership may depend less on which company spends the most on AI infrastructure and more on which one can demonstrate that those investments translate into sustainable earnings growth and attractive returns on invested capital.
“Investors will increasingly focus on these companies’ ability to realize an adequate return on investment,” Moody’s said.
