X users discuss AI capital expenditure trends, including a decentralized spatial data network project using smartphones as edge nodes and debate over tech company AI spending strategies. Topics range from DePIN economic models to market technical analysis amid Fed meetings and inflation concerns.
A social media discussion explores the AI paradox: despite AI model costs dropping 285x, major hyperscalers are increasing data center capital expenditure 4x from $90.1bn (2020) to $377.8bn (2025), suggesting they see massive untapped demand as AI becomes cheaper. A separate post warns that rising Treasury yields near 5% pose a financing risk to AI infrastructure investments, as corporations compete with government borrowing and higher discount rates threaten project returns.
A social media discussion debates the AI paradox: while AI model costs have plummeted 285x (from $20 to $0.07 per million tokens), major tech companies are quadrupling capital expenditure on AI infrastructure from $90.1bn (2020) to $377.8bn (2025). The post argues hyperscalers recognize that cheaper AI will unlock massive new use cases across customer service, coding, research, and enterprise workflows, driving continued infrastructure investment rather than signaling a bubble.
AI model pricing has collapsed 285.7x while major tech companies increased capital expenditure from $90.1bn (2020) to $377.8bn (2025), a 4x increase. As AI intelligence costs approach zero, hyperscalers are investing heavily in data centers and infrastructure to capture new markets where AI applications were previously too expensive.