Why Jensen Huang Spent $12.9 Billion on a Free AI Website

Why did Jensen Huang drop a staggering $12.9 billion to acquire a seemingly free website? NVIDIA’s latest blockbuster acquisition — the full purchase of Hugging Face — has left many industry observers confused. The platform generates merely $150 million in annual revenue, yet NVIDIA paid over 80 times its sales multiple to take full ownership.

Hugging Face is no tech giant with proprietary hardware or high-margin enterprise services. At its core, it is an open-source AI community and repository hosting over one million public AI models. All developers worldwide can download these complete model weights and datasets for free. On the surface, it looks like a public-interest platform serving the entire AI industry — so why would the world’s dominant computing powerhouse splurge on what seems like a charitable public good?

This is not reckless spending. It is a shrewd, long-term strategic move that secures NVIDIA’s monopoly power for the next decade, rooted in a classic tech business principle: the commoditization of complementary goods.

Software thinker Joel Spolsky first summarized this rule in 2002. To maximize profits from your core product, you must turn all its complementary goods into cheap, even free, commodities. The more accessible and affordable the complements become, the more indispensable your core business grows.

History bears this out. Microsoft once actively supported low-cost third-party PC hardware manufacturers, driving down global hardware prices. Cheaper PCs meant more users worldwide, which in turn boosted sales of Microsoft’s high-margin Windows operating system. Similarly, IBM embraced open PC compatibility standards to sell far more high-profit hardware components at scale.

Applying this logic to today’s AI landscape makes NVIDIA’s strategy crystal clear. The company’s core profit engine is its high-end AI chips and the exclusive CUDA ecosystem, which underpins its 75%+ gross margin and unrivaled dominance in global AI computing.

The critical complementary goods for NVIDIA’s expensive GPUs are AI models, algorithm frameworks, and massive parameter datasets that run on its hardware.

Imagine a future where global AI capability is monopolized by a handful of closed-source giants like OpenAI and Anthropic. These top-tier model providers would hold overwhelming bargaining power. They could leverage their massive chip procurement volume to force down NVIDIA’s pricing, or even develop custom in-house chips to completely bypass and eliminate NVIDIA’s role in the industry supply chain.

Hugging Face flips this dynamic entirely. It democratizes state-of-the-art AI by making cutting-edge open-source models accessible to every developer, researcher, and startup worldwide. Any individual or enterprise looking to fine-tune, deploy, and run these free open-source models locally ultimately needs one essential resource: NVIDIA GPU computing power.

In short, the more prosperous, widespread, and free the open-source AI ecosystem becomes, the greater the global demand for NVIDIA’s physical computing chips.

This raises a natural question: if a thriving open-source ecosystem benefits NVIDIA so greatly, why not let Hugging Face grow independently? Why invest $12.9 billion to take full control?

The answer lies in existential risk. Hugging Face previously rejected a $500 million investment offer from NVIDIA at a $7 billion valuation. NVIDIA’s decision to nearly double the valuation for a full acquisition reveals deep strategic urgency.

Hosting over one million open-source models and 200,000 datasets, Hugging Face has become the de facto center of the global open-source AI ecosystem. It defines the daily workflow of millions of developers across the world. Whoever controls this model distribution gateway controls the underlying rules of AI development.

If a competitor acquired Hugging Face, or if the platform prioritized optimization for rival chip architectures, NVIDIA’s decades-long moat would suffer an irreversible crack. By bringing Hugging Face fully in-house, NVIDIA embeds its inference optimization tools and hardware infrastructure as the default factory setting for the entire open-source AI world.

Rival chips, no matter how impressive their raw computing parameters, will be relegated to second-tier status in the open-source ecosystem from the source of distribution — locked out of mainstream developer adoption entirely.

This echoes Microsoft’s iconic $7.5 billion acquisition of GitHub in 2018. The deal was widely mocked as overpriced at the time, yet eight years later, it stands as one of Microsoft’s most transformative investments, securing its control over global developer traffic and spawning profitable products like GitHub Copilot.

For NVIDIA, a firm with tens of billions in annual net profit, the $12.9 billion purchase is not merely a financial investment. It is a high-value insurance policy, locking up dominance over global AI computing distribution for the next decade.

The fiercest battles in tech and business are never fought on spec sheets. Top-tier industry players build unbreakable tollgate monopolies — then fund and build free, open, accessible highways that lead every user directly to their tollgate.