The AI Dumpster Fire
Notes from https://youtu.be/zDtvrme-L-0?is=U3IvdLnC3Rnfd37-
With nearly $3 trillion in AI investment on the line, taxpayers shouldn't be worrying about the return on our investment—we should be worrying about the return of it. Everyday citizens will pay dearly for this inferno.
Today’s tech billionaires are playing a familiar game: privatizing profits while socializing costs and losses. Despite sky-high market valuations, internal warnings are leaking out. Reports reveal that OpenAI’s Chief Financial Officer cautioned CEO Sam Altman about the company's aggressive spending plans. While Altman publicly praises the transformative potential of AI, internal messaging to investors paints a far more precarious financial reality.
Recent web-traffic tracking data from analysts at Similarweb (via Momentic) shows OpenAI capturing 53.9% of worldwide chatbot visits, followed by Google Gemini at 27.9% and Anthropic's Claude at 9.2%, with smaller players splitting the remainder. Yet, consumer willingness to pay for subscriptions remains stubbornly low, as users routinely migrate to free alternatives.
Even massive promotional pushes—such as Samsung's $6 billion push to drive users toward $20/month tiers—have yielded lackluster returns. Corporate adoption tells a similar story: only 25% of enterprise AI initiatives deliver their expected ROI, and a mere 16% have successfully scaled across organizations. Where productivity gains do exist, they remain narrow and tightly concentrated in specific niches like basic code generation and automated customer support.
At the same time, AI supply and capacity are wildly mismatched. Tech giants are leasing out computing power simply to monetize their staggering capital expenditures rather than serving actual end-user demand. Meta has explored cloud leasing talks with entities like Anthropic, while xAI has rented out infrastructure to offset capital burn.
To curb ballooning expenses, developers are actively turning away from costly American frontier models. Microsoft is testing lower-cost alternatives, while Cursor admitted to building its frontier coding model on top of domestic Chinese options like Kimi. Consequently, Chinese models surged from roughly 1% of token usage on OpenRouter in late 2024 to over 60% by early 2026.
Meanwhile, once-solid strategic partnerships are fracturing. Microsoft is no longer the exclusive cloud provider for OpenAI and will no longer pay them a revenue share—signaling a fundamental rift in their once-tight alliance.
This skepticism extends far beyond industry observers; it has reached the highest levels of federal regulation. A leaked report submitted by Treasury analysts to Secretary Bessent warned of systemic risks to the broader economy, noting that AI firms are far more deeply entrenched in the financial system than dot-com companies were in the late 1990s. With over $750 billion flowing into data center buildouts, capital markets, and infrastructure, federal regulators are quietly sounding the alarm while executives urge the public to keep an "open mind."
The Treasury report highlights three primary systemic triggers:
-Missed Productivity Goals: Enterprises pressured into rapid AI adoption are seeing minimal efficiency gains rather than promised transformations.
-Backlash & Infrastructure Delays: Public outrage over energy-guzzling infrastructure has delayed or blocked over $64 billion in data center projects nationwide.
-Tightening Capital: OpenAI isn't just burning cash; it is exhausting favorable financing options, forcing a reliance on private equity firms demanding returns as high as 17.5%.
To keep the industry afloat, proposals have surfaced suggesting government-backed stakes in leading AI firms. If Washington uses taxpayer dollars to prop up or bail out failing business models, it will become another textbook case of corporate welfare. If this bubble bursts, tech executives will walk away with secured fortunes while everyday citizens bear the brunt through higher electricity bills, inflated prices, and economic instability.
This AI dumpster fire is no longer a question of if—it is a question of when.
Comments