I don't believe the magnificent +70% revenue growth fiscal year 2028 announced by Nvidia, sorry.

Nvidia's official projection of 70% revenue growth for fiscal 2028 masks a fundamentally unsustainable spending cycle that cannot mathematically scale. Driven by hyperscalers, total AI capital expenditures are projected by Goldman Sachs and JPMorgan to reach $725 billion to $800 billion in 2026 before surging past $1.0 trillion to $1.2 trillion in 2027. However, this expansion relies heavily on vendor-backed "circular financing," where equipment makers subsidize their own buyers. A prime example is Nvidia's $100 billion investment in OpenAI combined with its $105 billion backing for OpenAI's Ohio data center lease. This dynamic forces heavily indebted clients to make massive infrastructure commitments that cannot be sustained by operating cash flows alone. According to Morgan Stanley, hyperscalers' off-balance-sheet commitments and lease guarantees now top $3.1 trillion, while total debt liabilities have surged to $770 billion. Meanwhile, physical infrastructure bottlenecks—including power generation, cooling capacity, and memory component inflation—are beginning to compress profitability. Nvidia has already adjusted its gross margin outlook down toward 72% due to rising HBM memory costs. The market is now rapidly approaching a critical tipping point—the inevitable "STOP" trigger. If end-user software revenues fail to justify these trillion-dollar hardware outlays, this interconnected financial web will unspool quickly. Well before fiscal 2028, power grid constraints and the collapse of vendor-financed loops will force a sharp and painful rationalization of the entire AI hardware sector.