OpenAI Is Just Another Boring, Desperate AI Startup

Source: Where’s Your Ed At Date: 2025-10-03 URL: https://www.wheresyoured.at/sora2-openai/

Summary

Ed argues OpenAI is fundamentally unremarkable without its hype: GPT-5 costs more to operate than its predecessor while delivering underwhelming improvements; hallucinations are mathematically inevitable in LLMs and thus unfixable; ChatGPT is merely a “wrapper” around foundation models. API revenue is a tiny fraction of income — OpenAI functions identically to any other AI startup, just with more capital. The company leaked plans for hardware, ads, browsers, and semiconductors not as strategy but to inflate valuation through press coverage. OpenAI spent 150% of H1 2025 revenue on R&D while burning cash on commoditized, slowing-growth products.

Implications

The “wrapper, not innovator” claim is the most falsifiable argument in this piece: if true, it means OpenAI’s moat is brand and capital access, not technical capability — and both are erodable. The 150% R&D spend figure is a direct sustainability indicator: a company spending $1.50 on research for every $1.00 in revenue cannot reach profitability through efficiency gains alone; it requires revenue to grow faster than costs, which requires either price increases (churn risk) or volume growth (requires genuine value differentiation). The valuation-through-press-cycle observation connects to the broader artificial demand thesis.