OpenAI Needs $400B In The Next 12 Months
2 days ago
- #Financial Recklessness
- #Data Centers
- #OpenAI
- OpenAI, NVIDIA, Broadcom, and AMD are making ambitious claims about building massive data center capacities, with OpenAI alone aiming for 250GW by 2033, costing around $10 trillion.
- The author criticizes the feasibility of these plans, citing unrealistic timelines, insufficient funding, and lack of infrastructure (e.g., transformers, electrical-grade steel, talent).
- OpenAI's financial demands are staggering: it needs ~$400 billion in the next 12 months to meet its 2026 goals, far exceeding global venture capital raised in 2024 ($368 billion).
- Sam Altman's promises are labeled as lies, with the author arguing that OpenAI lacks the capacity, revenue, or demand to justify such investments.
- The media is criticized for uncritically reporting these plans as plausible, despite their implausibility.
- OpenAI's spending is highlighted as reckless, with $11.7 billion burned in R&D over 1.5 years without commensurate product releases.
- The article questions the purpose of OpenAI's expansion, noting that even if ChatGPT usage grows, it doesn’t justify $10 trillion in data centers.
- Oracle’s deal with OpenAI is scrutinized, with doubts about Oracle’s ability to deliver 4.5GW of capacity by the promised timeline.
- The author concludes that OpenAI’s plans are a dangerous financial bubble, unsustainable for the global economy.