Tech giants race to acquire open-source AI platforms as dominance concerns grow

The tech industry is pouring unprecedented capital into open-source AI infrastructure. Nvidia's reported $13 billion acquisition of Hugging Face represents the latest in a series of major deals targeting platforms built for developers working with open-weight models—the AI equivalent of shared GitHub repositories rather than proprietary systems locked behind corporate walls.

A new acquisition wave reshapes the AI landscape#

This buying spree extends beyond Nvidia's Hugging Face bid. The semiconductor giant previously agreed to a $6 billion deal with Poolside, an open-weight model builder whose workforce would largely transfer to Nvidia. Meanwhile, payments platform Stripe acquired OpenRouter, a leading distributor of open models to enterprises, for over $7 billion just weeks earlier.

The pattern points to deliberate strategic positioning: major players are moving aggressively to own or control the infrastructure supporting alternatives to closed-source AI systems.

Why giants are hedging against frontier labs#

For companies like Nvidia, the calculus is straightforward. As frontier AI labs—OpenAI and Google among them—develop their own inference hardware (including OpenAI's recently announced Jalapeño chip), traditional chip suppliers face eroding reliance. By controlling major open-model platforms, Nvidia gains direct access to vast developer communities and the ability to steer them toward its own computing standards and Nemotron family of models.

Stripe frames the opportunity differently. "Tokens are the central currency for companies building with AI, and it's clear that the real-world economic potential will depend on making good use of scarce compute resources," founder and CEO Patrick Collison said of the OpenRouter acquisition.

The current state of adoption remains modest#

Despite the investment frenzy, open-weight model adoption remains limited. According to spending data analysis by Ramp, only 6% of companies currently use open models—a figure that drops to 2% among software engineers surveyed by Jellyfish. However, the trajectory is distinctly upward.

Nik Albarran, AI product lead at Jellyfish, explained that open models currently thrive in narrowly defined scenarios with repetitive workloads. Customer service chatbots exemplify this use case: high-volume, predictable tasks where tuning a lighter model for cost efficiency makes economic sense. Frontier models continue to dominate coding and agentic tasks that demand varied reasoning across novel problems.