The most important marriage in AI is getting a $20 billion renewal vow.
After a week of rumors that the deal had stalled, Nvidia (NVDA) is reportedly nearing a definitive agreement to invest roughly $20 billion in OpenAI. This is the anchor check for a historic funding round targeting $100 billion in total capital at a staggering $830 billion valuation.
💰 THE CAP TABLE: The round is shaping up to be a titan-level syndicate:
- Nvidia: ~$20 Billion (Strategic Anchor)
- Amazon: In talks for up to $50 Billion
- SoftBank: Discussing $30 Billion
- Total Valuation: ~$830 Billion
⚡ THE “FRENEMY” DYNAMIC: This deal comes at a critical friction point.
- The Tension: Reuters reports that OpenAI has grown unsatisfied with Nvidia’s chips for inference workloads, actively exploring alternatives (AMD, custom silicon) to reduce costs and latency.
- The Response: By writing a $20B check, Jensen Huang isn’t just buying equity; he is buying loyalty. He is ensuring that even as OpenAI looks around, they stay tethered to the CUDA ecosystem for their most critical training runs.
🎤 CEO CLARITY: Jensen Huang has moved quickly to kill the “breakup” narrative.
- “We are going to make a huge investment… probably the largest investment we’ve ever made.”
- He explicitly denied rumors of a rift, calling them “nonsense” and reaffirming that he “loves working with Sam.”
💡 ANALYST TAKEAWAY: This is the ultimate “Circular Economy” trade. Nvidia invests billions into OpenAI -> OpenAI uses that cash to buy Nvidia H100/Blackwell GPUs -> Nvidia books it as record revenue. It’s a brilliant, if expensive, flywheel. But with OpenAI secretly testing competitor chips for inference, this investment serves as a golden handcuff—ensuring the biggest buyer in history doesn’t drift too far from the supplier that made them.
👇 Semi & VC Pros: Is Nvidia buying revenue by investing in its own customers, or is this necessary ecosystem defense against Google/Amazon custom silicon?