The progress of AI is currently facing potential slowdown due to escalating training costs for frontier models, which exceed $100 million and have hit diminishing returns. New models require significantly more computational resources but are yielding unimpressive performance enhancements. Consequently, a shift towards an “inference scaling paradigm” is emerging, where AI firms rely on extensive compute to generate answers. OpenAI’s new model, o3-pro, exemplifies this, taking around ten minutes per query and costing $200 monthly, yet offers limited improvements. OpenAI is using promotional trials to attract users and inflate its user metrics, backed by venture capital funding, which could boost its valuation to around $300 billion. However, with AI facing competition and diminishing returns, the expected annual revenue might not materialize, likening the industry to a bubble. Future profitability depends on acquiring lucrative government contracts and retaining user subscriptions as the landscape changes.
Source link
Experience o3-pro in ChatGPT for just $1!

Leave a Comment
Leave a Comment