OpenAI Scales Compute Ambition to 30GW by 2030 After Securing Over 8GW

Details

• OpenAI announced on April 22, 2026, that it has identified over 8GW of compute capacity toward its January 2025 commitment of 10GW.
• The company now plans to reach 30GW of compute by 2030 to meet accelerating demand for intelligent systems.
• This milestone reflects rapid scaling needs in AI infrastructure amid surging global data center growth.
• Recent industry deals align with OpenAI's progress, such as Nscale's acquisition of the 8GW Monarch Compute Campus in West Virginia, certified as the US's first AI microgrid\[1]\[2][6].
• Nscale signed an LOI with Microsoft for 1.35GW of AI compute using 430,000 NVIDIA Vera Rubin GPUs, with first phase online by 2027\[3]\[6].
• Global trends show India's data center capacity hitting 8GW by 2030 with $30B investment, while US pipeline could triple to 102GW\[4]\[5].
• Rival Anthropic is pursuing at least 10GW of data center capacity with ex-Google executives[7].

Impact

OpenAI's jump from a 10GW near-term goal to 30GW by 2030 positions it to match or exceed rivals like Anthropic, which is targeting at least 10GW, amid a US data center pipeline that could triple to 102GW. This aggressive scaling pressures competitors including Microsoft-backed Nscale's 8GW campus and intensifies the global race for AI power, where India's capacity is set to hit 8GW by 2030. The move aligns with surging demand but highlights infrastructure bottlenecks, potentially widening OpenAI's lead in frontier AI deployment while spurring investments in energy-efficient microgrids and next-gen GPUs like NVIDIA Vera Rubin.