Firms like OpenAI and Meta claim a compute shortage while also exploring selling compute capacity. This isn't a contradiction but a strategic evolution. They are buying all available supply to secure their own needs and then arbitraging the excess, effectively becoming smaller-scale cloud providers for AI.

To service its massive debt for GPU purchases, CoreWeave locks customers into multi-year contracts. This secures revenue to cover debt payments but means CoreWeave misses out on the higher margins available from rising spot market prices for GPU compute—a calculated trade-off between stability and profitability.

While increased CapEx signals strength for cloud providers like Microsoft and Google (who sell that capacity to others), the market treats Meta's spending as a pure cost center. Every dollar Meta spends on AI only sees a return if it improves its own products, lacking the direct revenue potential of a cloud platform.

Instead of bearing the full cost and risk of building new AI data centers, large cloud providers like Microsoft use CoreWeave for 'overflow' compute. This allows them to meet surges in customer demand without committing capital to assets that depreciate quickly and may become competitors' infrastructure in the long run.

Large tech companies are creating SPVs—separate legal entities—to build data centers. This strategy allows them to take on significant debt for AI infrastructure projects without that debt appearing on the parent company's balance sheet. This protects their pristine credit ratings, enabling them to borrow money more cheaply for other ventures.

Cash-rich hyperscalers like Meta utilize Special Purpose Vehicles (SPVs) to finance data centers. This strategy keeps billions in debt off their main balance sheets, appeasing shareholders and protecting credit ratings, but creates complex and opaque financial structures.

Meta is using off-balance-sheet "special purpose vehicles" (SPVs) to finance its AI data centers. This financial engineering obscures the true scale of its capital commitments by keeping massive debt and assets off its main balance sheet, a tactic explicitly compared to the controversial methods used by Enron.

The huge CapEx required for GPUs is fundamentally changing the business model of tech hyperscalers like Google and Meta. For the first time, they are becoming capital-intensive businesses, with spending that can outstrip operating cash flow. This shifts their financial profile from high-margin software to one more closely resembling industrial manufacturing.

Oracle's significant investment in AI infrastructure appears less risky because they've structured deals where major clients like Meta and OpenAI pay for GPUs upfront or bring their own hardware. This strategy prevents Oracle from becoming overleveraged while rapidly scaling its data center capacity.