On August 28, according to reports from NUPIAO, Nvidia has reached an acquisition agreement with Hugging Face, the world’s largest open-source AI platform, with a total transaction value of $12.9 billion. If the deal goes through, it would rank among the highest-value M&A deals in the global AI sector for 2026.
Among the early investors in this blockbuster deal, NBA superstar Kevin Durant is poised to reap massive returns thanks to his early bets in the seed and Series A rounds. As an early individual investor in Hugging Face, Durant’s investment journey began back in 2018.
Public records show that Durant, through Thirty Five Ventures—the venture capital firm he co-founded with business partner Rich Kleiman—participated in Hugging Face’s 2018 seed round with a $100,000 investment. When the company raised its Series A in 2019, Durant doubled down with an additional $150,000, bringing his total investment to $250,000.
At the time, Hugging Face was still in the early stages of pivoting from a consumer app to a developer platform, and its business model had yet to be fully validated. This was a true angel-stage risk bet. According to estimates, Durant’s investment would now yield paper returns exceeding $60 million—a return multiple of more than 240x, even surpassing his $43.9 million annual salary with the Houston Rockets for the 2026-2027 season. Since Hugging Face hasn’t disclosed its detailed cap table, Durant’s exact ownership stake and final payout remain unconfirmed by official sources.
Hugging Face was founded in 2016 by French entrepreneurs Clément Delangue, Julien Chaumont, and Thomas Wolf, with headquarters in New York City. The company initially focused on chatbot products for teenagers. In 2018, the core team made a pivotal strategic pivot by fully open-sourcing its Transformer library. From there, the company gradually grew into the world’s largest open-source AI model hosting and collaboration platform, widely dubbed the “GitHub of AI.”
As of 2026, the Hugging Face platform hosts over 3 million pre-trained AI models and nearly 1 million open-source datasets, serving more than 13 million developers globally. Major open-source models including Meta’s Llama series, Mistral’s open models, and China’s Qwen series all use Hugging Face as their primary distribution channel. On the monetization front, the company generates revenue through enterprise-grade model deployment services, premium account subscriptions, and customized AI solutions. In 2026, its annualized recurring revenue hit approximately $150 million—several times higher than just two years ago.
Looking at the deal timeline, the financial ties between Nvidia and Hugging Face were already in place. In August 2023, Hugging Face closed a $235 million Series D round at a $4.5 billion valuation, with Nvidia joining Google, Amazon, Intel, Salesforce, and other tech giants as a strategic shareholder.
At that point, Nvidia had already begun building its open-source AI ecosystem, but it participated purely as a financial investor without gaining control. In late 2025, Nvidia proposed a $500 million investment at a valuation of roughly $7 billion, but Hugging Face turned it down—the company wanted to maintain platform neutrality and didn’t want any single shareholder to wield outsized influence. However, in mid-August this year, word leaked that Hugging Face had hired investment banks to explore a full sale. Just one week later, the two sides quickly hammered out an acquisition agreement. In terms of valuation, the $12.9 billion price tag represents a nearly 187% increase over the $4.5 billion Series D valuation from 2023.
For Nvidia, the payment burden of this deal is manageable. According to its fiscal Q2 2026 earnings report, the company holds more than $60 billion in cash and short-term investments, meaning the $12.9 billion acquisition price accounts for only about one-fifth of its cash reserves.
Nvidia’s willingness to pay a hefty premium comes down to its core strategic goal of building a full-stack AI ecosystem. Right now, competition in the AI industry is shifting from pure compute hardware toward the software ecosystem layer. Closed-source AI giants like OpenAI and Anthropic have all launched in-house AI chip projects to reduce their dependence on Nvidia GPUs. OpenAI’s self-developed “Jalapeño” chip, for instance, is already in testing and is claimed to rival Nvidia’s flagship Blackwell chip in performance.
Against this backdrop, the open-source AI ecosystem has become Nvidia’s inevitable hedge against the closed-source camp. The more widely open-source models are adopted and the larger the developer community grows, the harder it becomes for any single player to shake Nvidia’s GPU market demand. As the core distribution hub for open-source AI models, Hugging Face holds the gateway to global developers and serves as a barometer for tech trends. Taking control of this platform means Nvidia can extend upward from hardware supply into the model distribution layer, further reinforcing the moat around its CUDA ecosystem.
Additionally, Nvidia has recently poured significant capital into building its own Nemotron series of open-source models. Acquiring Hugging Face will allow the company to deeply integrate its proprietary models with the platform’s traffic. Just days before this deal was announced, payment giant Stripe agreed to acquire the AI model routing platform OpenRouter for over $8 billion. These two deals landing back-to-back underscore that AI “middleware” infrastructure has become the core battleground for tech giants’ strategic acquisitions.
