Pomegra
---
---
The Original $100 Billion Deal Collapses Under Its Own Weight
In September 2025, Nvidia and OpenAI unveiled what was positioned as the most consequential deal in artificial intelligence history: a letter of intent under which Nvidia would invest up to $100 billion in OpenAI's infrastructure, funding the deployment of 10 gigawatts of Nvidia GPU systems — a scale of computing power equivalent to the output of roughly 10 nuclear reactors. Nvidia CEO Jensen Huang described it at the time as matching the company's entire GPU shipment volume for the year.
Five months later, that deal has never closed. No definitive agreement was signed, no capital transferred, and the ambitions embedded in the original announcement have formally given way to a fundamentally different structure. As of February 20, 2026, Nvidia is finalizing a $30 billion direct equity investment in OpenAI as part of a broader fundraising round in which OpenAI is seeking more than $100 billion in total capital, at a valuation of approximately $830 billion.
---
A Restructured Deal Built Around an Equity Stake
The new arrangement marks a significant strategic shift. Rather than a conditional infrastructure commitment tied to gigawatt-by-gigawatt deployment milestones, Nvidia's $30 billion now takes the form of a direct equity position in one of its single largest customers. The investment is expected to be finalized as early as this weekend, with SoftBank Group and Amazon also positioned to participate in the broader round.
OpenAI plans to deploy much of the fresh capital toward the purchase of Nvidia's H100 and next-generation GPU systems, which remain the backbone of large-scale AI model training and deployment. The circular dynamic — whereby Nvidia's investment fuels GPU procurement that flows back to Nvidia — has drawn scrutiny from market analysts, though both companies have consistently defended the strategic logic of the arrangement.
---
How the Original Deal Unraveled
The collapse of the $100 billion framework was not a single event but a months-long erosion. By late January 2026, the Wall Street Journal reported the deal was "on ice," citing internal doubts within Nvidia and private criticism from Jensen Huang regarding what he described as a lack of operational discipline in OpenAI's business approach. Huang also reportedly raised concerns about OpenAI's competitive exposure to Google and Anthropic — claims he later publicly dismissed as "nonsense."
At a Taiwan press event in late January, Huang directly walked back the $100 billion figure, stating it was "never a commitment" and that Nvidia would "invest one step at a time." He added that Nvidia's participation in OpenAI's forthcoming round would still represent the single largest investment Nvidia has ever made.
Compounding the friction, Reuters reported in early February that OpenAI had been quietly exploring alternative chip suppliers, citing dissatisfaction with Nvidia GPU performance for AI inference workloads — particularly for Codex, OpenAI's AI-assisted code generation product. OpenAI subsequently struck a $10 billion deal with Cerebras for low-latency inference capacity, an AMD partnership covering 6 gigawatts of GPU supply, and a collaboration with Broadcom to develop a proprietary AI chip.
---
Sam Altman Moves to Contain the Fallout
As reports of the stalled megadeal circulated in early February, OpenAI CEO Sam Altman moved quickly to publicly reaffirm the partnership. "We love working with NVIDIA and they make the best AI chips in the world," Altman posted on X. "We hope to be a gigantic customer for a very long time." The statement was widely read as an attempt to stabilize market sentiment following a roughly 1.1% decline in Nvidia's share price in the immediate aftermath of the Wall Street Journal report.
---
Market Implications: Consolidation at the Top of the AI Stack
The revised deal reflects a broader pattern of deepening financial interdependence among the leading players in artificial intelligence infrastructure. Chip suppliers, cloud providers, and foundation model developers are increasingly bound not just by commercial relationships but by cross-equity stakes and co-investment structures.
For Nvidia, a $30 billion commitment at an $830 billion OpenAI valuation locks in a strategic anchor position in the world's most valuable AI company, while maintaining significant financial flexibility compared with the open-ended exposure of the original $100 billion framework. For OpenAI, access to Nvidia's capital — alongside contributions from SoftBank and Amazon — supports the largest private capital raise in recorded history, providing a multi-year runway for infrastructure buildout, model development, and international expansion.
---
Outlook: The AI Infrastructure Race Enters a New Phase
The transition from letter of intent to structured equity signals a maturation in AI deal-making — one in which headline-grabbing announcement figures give way to more precisely negotiated instruments with defined valuations, governance rights, and investment tranches. OpenAI's parallel moves to diversify its chip supply chain, through Cerebras, AMD, and custom silicon with Broadcom, further indicate that the company is building a more resilient and competitive infrastructure posture regardless of its relationship with any single supplier.
The $30 billion investment, once finalized, will stand as a landmark transaction at the intersection of semiconductor dominance and AI model leadership, cementing the Nvidia-OpenAI relationship as the defining commercial alliance of the current AI cycle.
---
Mentioned tickers: `NVDA, AMZN, AMD, GOOGL`
