Nvidia–OpenAI $100B megadeal stalls as $20B equity plan emerges

The Nvidia–OpenAI megadeal was supposed to be the capstone of the current AI compute cycle: 10 gigawatts of power and up to $100 billion in tightly coupled capex to turn GPUs into an industrial utility for one customer. The fact that this arrangement is now “on ice,” with both sides probing a much smaller equity stake instead, is a strong signal that even the biggest players are re‑evaluating how far they want to vertically integrate compute and model ownership.([the-decoder.com](https://the-decoder.com/nvidia-is-reportedly-investing-20-billion-in-openai-says-it-will-proceed-one-step-at-a-time/))

For OpenAI, a downsized deal would reduce single‑vendor dependence and keep room for AMD, custom chips and cloud partners like Microsoft, Google and Oracle. For Nvidia, shifting from a circular $100 billion capex commitment to a more modest equity position limits balance‑sheet risk while preserving OpenAI as a marquee customer. Strategically, this is the first visible crack in the narrative that hyperscale AI will be built via a few massive bilateral alliances; instead, we’re drifting toward a more multipolar compute market where leading labs actively arbitrage between vendors and financing structures.([wsj.com](https://www.wsj.com/tech/ai/the-100-billion-megadeal-between-openai-and-nvidia-is-on-ice-aa3025e3?gaa_at=eafs&gaa_n=AWEtsqc4eHHFdb1Aaaw2Gr7jLvPloEr-vvPl8GrbT-YUmj0cUeZU6a4XM2ci&gaa_sig=4Tqkng7i_cMfQC6mSHihpUkNDvgKWfYXQQje-X-Sf71YENnZAFc831aH5Lf9MnpzacaQX4UyZgWx7ySa6OdaOw%3D%3D&gaa_ts=6983554a&utm_source=openai))

From an AGI race perspective, the outcome is ambiguous. Less tightly coupled capex may slow the ramp‑up of a single 10‑GW “AI factory,” but it also diversifies the ecosystem and may spread frontier‑scale compute across more actors, making the landscape more competitive—and potentially harder to govern.