According to CoinMeta, Dylan Patel, the founder of SemiAnalysis, predicted in the latest podcast that by 2028, OpenAI and Anthropic will together account for 70% to 80% of the global increase in AI computing power, with the total computing power scale potentially exceeding 100GW. Currently, these two companies already account for about 30% of the global annual increase in computing power, and this proportion is still rising rapidly. Patel pointed out that the business models of cutting-edge AI laboratories are changing, and the efficiency of computing power output has significantly increased. Anthropic currently generates revenue of about $50 million per megawatt of computing power, which may further rise to $100 million. Patel estimates that from 2024 to 2029, global AI related capital expenditures will reach about $11 trillion, of which over $5 trillion will need to be financed through debt. Since the potential return on investment for AI infrastructure is much higher than that of traditional industries, tech giants may accept higher financing costs, which could drive up overall credit interest rates and put pressure on the valuation of traditional assets and high-debt economies. In addition, Patel believes that the newly added computing power in the future may not necessarily be mainly used externally, but may instead flow more towards internal research and development in AI laboratories and self-improvement of models.