Sora 2 and ChatGPT are consuming so much power that OpenAI just did another 10 gigawatt deal

OpenAI has partnered with Broadcom to design and develop custom AI chips and systems, a strategic move to secure massive computing resources and gain greater control over its hardware stack by embedding model learnings directly into the chips. This collaboration, expected to deploy in H2 2026, addresses the extreme power demands of AI, with the planned infrastructure consuming 10 gigawatts, highlighting AI's immense power demands. Broadcom's shares surged 12% on the announcement, confirming earlier speculation about a $10 billion customer.

Analysis

OpenAI has formalized a strategic partnership with Broadcom to co-design and develop custom AI chips and systems, a move anticipated to deploy in the second half of 2026. This collaboration, which involves a massive 10 gigawatts of power infrastructure, immediately propelled Broadcom shares up 12% on Monday, confirming prior speculation regarding a new $10 billion customer. The deal underscores OpenAI's aggressive push to secure computing resources for its rapidly expanding user base, including 800 million weekly ChatGPT users and the fast-growing Sora video generation app.

This partnership signifies OpenAI's intent to gain greater control over its hardware stack, moving beyond reliance on partners like Nvidia and AMD by embedding frontier model learnings directly into custom accelerators. This strategic shift aims to unlock new levels of capability and intelligence within its AI services. The custom chip development allows OpenAI to tailor hardware specifically for its advanced models, potentially optimizing performance and efficiency.

The initiative highlights the escalating power demands of the AI boom, with the planned infrastructure consuming as much electricity as 8 million US households. This aligns with a Department of Energy report projecting data centers to consume 6.7% to 12% of total US electricity by 2028, up from 4.4% in 2023. The significant energy footprint raises environmental concerns and underscores the increasing importance of energy efficiency and infrastructure development in the AI sector.

Key Decisions for Investors

• Investors should consider Broadcom (AVGO) as a key beneficiary of the custom AI chip trend, given the confirmed $10 billion OpenAI deal and its validation of Broadcom's ASIC capabilities.
• Monitor companies positioned to address the escalating energy demands of AI infrastructure, including those in power generation, data center cooling, and energy efficiency solutions, as the 10 gigawatt requirement highlights a significant market opportunity.
• While OpenAI's custom chip development indicates a strategic shift, investors should continue to evaluate the evolving roles of established AI chip providers like Nvidia (NVDA) and AMD (AMD) within the broader AI ecosystem.
• Scrutinize the ESG strategies of AI companies and data center operators, as the substantial energy consumption projected by the DOE report suggests increasing regulatory and environmental pressures on the sector.