- Markets - Around The World
- Posts
- OpenAI’s o1-Pro is the Most Expensive AI model in the Industry
OpenAI’s o1-Pro is the Most Expensive AI model in the Industry
& AI's Moore's Law
American Companies are Highly Valued over Peers
The top nine American companies by market capitalization are each valued at over $1 trillion, showcasing the overwhelming dominance of U.S. tech firms in the global market. This stark contrast is evident when comparing the top 10 companies from the U.S., EU/UK, and China:
U.S. Top 10: Combined market cap of $20.4 trillion
EU/UK Top 10: Combined market cap of $2.8 trillion
China Top 10: Combined market cap of $2.6 trillion
Apple leads the pack with a $3.4 trillion valuation, followed closely by Nvidia at $3.3 trillion and Microsoft at $3.1 trillion. Even the 10th-ranked U.S. company, Walmart ($826.8B), surpasses China's most valuable company, Tencent ($513.7B), by over $300 billion.
Several factors seem to contribute to this American dominance:
Profit-making ability: U.S. tech giants excel at generating substantial profits.
Scalability: Software-based companies like Microsoft and Meta can scale operations more easily than traditional industries.
Future growth expectations: Investors valuing the potential for exponential growth, reflected in higher price-to-earnings ratios (e.g., Apple at 38x vs. LVMH at 28x).
Supportive ecosystem: Top universities, a culture of innovation, and business-friendly policies (lower corporate taxes, flexible labor laws) reinforce this dominance.
OpenAI's New O1-Pro Model Costs 1,000 Times More Than DeepSeek
OpenAI has released pricing for its new o1-pro reasoning model, setting a new industry high for AI model costs. The o1-pro model is priced at $150 per million input tokens and $600 per million output tokens. This is significantly more expensive than other leading models:
It's 10 times more expensive than OpenAI's standard o1 model
It's twice as expensive as GPT-4.5
It's 100 times more expensive than Google's Gemini 2.0 Flash for input tokens
OpenAI justifies this high price by stating that o1-pro "uses more compute than o1 to provide consistently better responses". The model is designed for complex reasoning tasks and offers features like a 200,000-token context window, image input support, and structured outputs. The company's focus on "reasoning" models like o1-pro leads to much higher computing costs. For example, solving individual ARC-AGI puzzles (Abstraction and Reasoning Corpus for Artificial General Intelligence is an AI benchmark designed to test a system's ability to infer and generalize abstract rules from minimal examples) costs $3,400 each. OpenAI is reportedly considering charging $20,000 per month for "PhD-level agents" and is seeking funding at a $340 billion valuation. Despite bringing in $3.7 billion in revenue in 2024, the company still reported a loss of about $5 billion.
AI's Exponential Leap: Doubling Capabilities Every Seven Months
Moore's Law is the observation that the number of transistors in a dense integrated circuit doubles about every two years, leading to exponential growth in computing power. This principle has driven advancements in computer technology for decades. Recent research from METR (Measuring AI Ability to Complete Long Tasks) reveals a similar exponential growth pattern in AI capabilities:
AI's ability to complete complex tasks has doubled approximately every seven months since 2019.
Current frontier AI models like Claude 3.7 Sonnet have a 50% time horizon of around 50 minutes for task completion.
This rapid advancement is driven by improvements in logical reasoning, tool use, error correction, and self-awareness in task execution.
If this trend continues, AI systems could potentially handle month-long human-equivalent tasks with 50% reliability by 2029. This acceleration in AI capabilities has significant implications for industries relying on automation, potentially transforming knowledge work across various sectors.