# NVIDIA The dominant AI-chip company and the single most important chokepoint in the global artificial intelligence supply chain. Founded in 1993 by [[Jensen Huang]], Chris Malachowsky, and Curtis Priem. Originally a graphics-card company for gamers, NVIDIA pivoted to general-purpose GPU computing and now controls an estimated **80–92% of the global AI accelerator market**. Its market capitalization surpassed $3 trillion in 2024, making it at various points the most valuable public company on Earth. For the knowledge base, NVIDIA matters on three axes: as an infrastructure monopoly whose CUDA software ecosystem creates structural lock-in; as a geopolitical flashpoint at the center of U.S.–China chip export controls; and as an increasingly political actor whose CEO has cultivated a direct personal relationship with the [[Trump White House]]. ## CUDA Lock-In NVIDIA's true moat is not hardware but **CUDA** (Compute Unified Device Architecture), the proprietary programming platform launched in 2006 that is optimized exclusively for NVIDIA GPUs. Nearly two decades of developer adoption, libraries, and tooling have created an ecosystem so sticky that switching costs are prohibitive even for hyperscalers. Competitors like [[AMD]] offer alternative chips, but the CUDA software stack — not raw silicon — is what keeps customers locked in. This is the textbook case of [[Monopoly as Doctrine]] in the AI era: control the platform layer and the hardware becomes non-optional. ## AI Infrastructure Dominance NVIDIA's data-center segment generated **$51.2 billion in a single quarter** (Q3 FY2026) and **$215.9 billion in full-year FY2026 revenue**. The Blackwell architecture (2024) and its successor Rubin (early 2026) power the training runs for essentially every frontier AI model. Major customers include [[OpenAI]], Google, Meta, Amazon, and Microsoft, along with sovereign AI initiatives worldwide. The [[OpenAI-NVIDIA Strategic Partnership]] (2026) involves deploying 10GW of NVIDIA systems, with NVIDIA committing up to **$100 billion** in OpenAI investment — the single largest disclosed NVIDIA AI investment to date. NVIDIA is also a key supplier to the [[Stargate Project]], the $500B AI infrastructure initiative backed by [[SoftBank]], [[OpenAI]], and Oracle. In July 2026 NVIDIA extended the same pattern to the one frontier lab that had stayed outside it, investing **$5 billion** in [[Ilya Sutskever]]'s [[Safe Superintelligence]] and granting access to the Vera Rubin platform — see [[NVIDIA-Safe Superintelligence Investment]]. SSI has no product, no revenue, and no published research; the check is larger than everything the lab had raised before it combined. ## Defense and National Security NVIDIA has deepened its defense-tech footprint through partnerships with [[Palantir]] and participation in DoD-backed consortia: - **Palantir partnership:** NVIDIA H100 Tensor Core GPUs are embedded in [[Palantir TITAN]] nodes for AI-powered satellite imagery and drone video analysis in near real-time. Over **$100 million** in direct DoD-related sales in FY2023. - **6G defense consortium:** Joined a Pentagon-backed initiative developing sixth-generation wireless technologies for military applications, alongside [[Lockheed Martin]] and [[Northrop Grumman]]. - **AI targeting systems:** NVIDIA hardware underpins the compute layer for [[Project Maven]] and other military AI programs that [[Palantir]] operates at the application layer. ## Geopolitical Significance: China Chip Export Controls NVIDIA sits at the center of the U.S.–China technology confrontation. See [[GPU Export Controls]] for the full policy timeline. The Biden administration restricted export of high-end AI chips (A100, H100) to China in October 2022, directly targeting NVIDIA's product line. NVIDIA responded by creating China-specific throttled chips (H800, then H20) that complied with the letter of the rules while preserving market access. Chinese military procurement notices have nonetheless documented requests for H100-class servers, and *Fortune* reported in May 2026 on encrypted communications revealing ongoing smuggling of NVIDIA chips to China, Russia, and Iran through intermediary countries. In **December 2025**, the Trump administration rolled back some Biden-era controls, allowing NVIDIA to sell the more powerful **H200** chip to China in exchange for a 25% revenue stake — a deal that drew bipartisan criticism. Senators [[Elizabeth Warren]] (D) and [[Jim Banks]] (R) had earlier flagged NVIDIA's Shanghai R&D facility as a Military-Civil Fusion risk. [[Jensen Huang]] personally lobbied the reversal: in December 2025, he held closed-door meetings with [[Donald Trump]] and Republican senators, with Democrats shut out entirely. Senator Warren criticized the meetings as secret industry lobbying. Huang's approach — going directly to the President, bypassing senior officials — earned him a reputation in Trumpworld as "heavy-handed" but effective, with sources crediting him for managing his relationship with Trump better than any other major tech CEO. ## Political Influence NVIDIA's political positioning is concentrated in Huang's personal relationships rather than traditional PAC spending: - **David Sacks connection:** Huang's standing with administration officials from the tech-right is strong, particularly with [[David Sacks]], Trump's "AI and Crypto Czar." - **Trump relationship:** Direct bilateral meetings between Huang and Trump, unusual for a semiconductor CEO, reflect NVIDIA's leverage as the sole supplier of chips that underpin the administration's AI ambitions. - **Bipartisan vulnerability:** NVIDIA faces criticism from both parties — hawks who see China chip sales as a national security threat, and progressives who see CUDA lock-in as monopolistic. ## Key Tensions NVIDIA embodies several contradictions that recur across the KB: - **Monopolist selling to both sides:** NVIDIA profits from U.S. military AI systems *and* sells to Chinese entities that may feed military applications — the same dual-use tension visible in [[Palantir]] and the broader [[Defense Tech Ecosystem]]. - **Infrastructure dependency:** The entire AI boom — and with it the political economy of [[Tech Oligarchy]] — runs on NVIDIA hardware. This makes Huang arguably the most structurally powerful figure in the AI ecosystem, even more than the model builders. - **Regulatory arbitrage:** NVIDIA's pattern of creating compliant-but-capable China-specific chips is a case study in how companies exploit the gap between the intent and the letter of export controls. ## See Also - [[Jensen Huang]] - [[OpenAI]] - [[OpenAI-NVIDIA Strategic Partnership]] - [[NVIDIA-Safe Superintelligence Investment]] - [[Safe Superintelligence]] - [[Palantir]] - [[Palantir TITAN]] - [[Project Maven]] - [[AMD]] - [[GPU Export Controls]] - [[CHIPS and Science Act]] - [[Stargate Project]] - [[SoftBank]] - [[David Sacks]] - [[Defense Tech Ecosystem]] - [[Monopoly as Doctrine]] - [[Tech Oligarchy]] - [[China]] - [[Trump-Xi China Summit 2026]] - [[Surveillance State]]