NVIDIA is reportedly considering the acquisition of Hugging Face, the renowned open-weight AI model platform, for a staggering $13 billion. Dubbed the 'GitHub of the AI era,' Hugging Face has garnered significant attention. Earlier, NVIDIA had already made a substantial investment of $6 billion to acquire Poolside. And just two weeks prior, Stripe secured OpenRouter for a sum exceeding $7 billion. This surge in capital inflow into the open-weight model sector underscores the profound transformations underway in the AI industry landscape.

NVIDIA's strategic maneuver is aimed at diminishing its dependence on major cloud service providers and leading AI labs, while simultaneously venturing into the burgeoning model market. Concurrently, prominent model manufacturers like OpenAI are fervently developing their proprietary inference chips.

Although open-weight models are not yet the mainstream choice, their adoption rate is steadily climbing. Presently, around 6% of companies and 2% of software engineers have embraced these models. They are particularly well-suited for businesses that necessitate extensive repetitive inference, such as customer service chatbots.

However, leading closed-source models still maintain an edge in programming and agent tasks. Companies primarily opt for open models to gain enhanced control and configurability. Should leading AI labs persist in raising prices, an increasing number of companies might be compelled to transition to open models.

Firms like Fireworks are bullish on the diversification of models, believing that the future will be dominated by specialized intelligence. They argue that each company will require proprietary models tailored to diverse scenarios. At present, the AI industry landscape is still evolving, and the allure of open technologies is gradually gaining momentum.