OpenAI will not disclose GPT-5’s energy use. It could be higher than past models

OpenAI's newly released GPT-5 model exhibits significantly higher energy consumption per query than previous versions, with expert estimates suggesting it could be up to 20 times more power-intensive and potentially consume electricity equivalent to 1.5 million US homes daily. This substantial increase in energy demand is attributed to GPT-5's enhanced complexity, larger scale, and advanced reasoning capabilities, underscoring the growing environmental and operational cost implications of cutting-edge AI development. The article also highlights a critical lack of transparency from AI companies regarding their models' resource footprints, posing challenges for comprehensive impact assessment and raising concerns for investors monitoring long-term operational sustainability.

Analysis

OpenAI's release of GPT-5 marks a significant leap in AI capabilities, but introduces substantial operational and environmental costs. University of Rhode Island research indicates an average query on GPT-5 consumes approximately 18 watt-hours, a figure that can spike to 40 watt-hours, representing a manifold increase over its predecessor, GPT-4o. This surge in energy demand, driven by the model's larger size and complex reasoning functions, could translate to an aggregate daily electricity consumption equivalent to 1.5 million U.S. homes if deployed across ChatGPT's user base. While architectural enhancements like 'mixture-of-experts' and more efficient hardware may offer some mitigation, the model's multi-modal and advanced reasoning capabilities are expected to be significantly more resource-intensive, potentially increasing consumption by five to ten times for certain tasks. This situation is compounded by a critical lack of transparency from OpenAI and its competitors regarding model parameters and resource usage, which obstructs accurate assessment of long-term operational expenditures and ESG risks. The data highlights a direct correlation between model sophistication and its cost-to-serve, presenting a material challenge to the scalability and profitability of next-generation AI.

Key Decisions for Investors

• Investors should increase scrutiny on the operational expenditure and margin profiles of companies heavily deploying frontier AI models, as the escalating energy costs associated with models like GPT-5 represent a material headwind to profitability.
• The growing energy footprint of AI is a significant ESG risk; therefore, it is prudent to favor companies in the AI ecosystem that demonstrate transparency regarding their environmental impact and invest in energy-efficient infrastructure.
• Consider potential opportunities in downstream sectors, such as manufacturers of energy-efficient semiconductors, data center cooling technologies, and power infrastructure providers, which stand to benefit from the increasing power demands of the AI industry.
• Factor the economic viability of deploying resource-intensive models into valuation, as high operational costs may limit the scalability and widespread adoption of the most advanced AI for many use cases.