Microsoft Maps Gigawatt-Scale AI Training With 500 Staff to Cut OpenAI Reliance Before 2030
Updated
Updated · Fortune · May 21
Microsoft Maps Gigawatt-Scale AI Training With 500 Staff to Cut OpenAI Reliance Before 2030
2 articles · Updated · Fortune · May 21
500 Microsoft Superintelligence staff met in Miami the week of March 30 for a three-day offsite to plan gigawatt-scale AI training runs, a push Mustafa Suleyman says is critical to AI self-sufficiency.
2032 is the deadline driving that effort: Microsoft loses access to OpenAI technology then, and Suleyman says the company needs two to three years to catch up with top AI labs.
Satya Nadella joined the gathering, called AI a chance to "refound" Microsoft, and spent hours with researchers as the company reorganized Copilot and model-development teams to move faster.
The reset follows a stumble in AI: Copilot adoption lagged, Microsoft shares fell 34% over five months, and the company loosened its OpenAI exclusivity while adding Anthropic models to Azure and Copilot.
$190 billion in 2026 capital spending shows the scale of the bet, as Microsoft races to build enough compute and enterprise AI products to stay competitive without depending on any single model provider.
How will Microsoft's $190B AI bet pay off as the underlying AI models become cheap commodities?
Can Microsoft tame the powerful but insecure OpenClaw for enterprise use without neutralizing its core appeal?
Microsoft’s $100 Billion Bet: Gigawatt-Scale AI Data Centers and the Push for Self-Sufficiency by 2030
Overview
Driven by the escalating demands of developing and training advanced AI models, Microsoft is fundamentally shifting its approach to artificial intelligence. The company is making a gigawatt-scale investment in computing power, leading to an unprecedented expansion of its AI infrastructure. This massive build-out supports the growing complexity and size of AI models and is closely linked to Microsoft’s evolving partnership with OpenAI, which began in 2016. As Microsoft’s investment in OpenAI grew, so did the need for more robust infrastructure, prompting Microsoft to build multiple datacenters and pursue greater self-sufficiency in AI development.