The rumored Hugging Face acquisition follows two other major open-weight deals this month: Nvidia’s $6 billion agreement with Poolside, an open-weight model builder, and Stripe’s more than $7 billion purchase of OpenRouter, the leading provider of open-weight models to enterprise customers. Together, these transactions represent over $26 billion in capital flowing into a sector built on giving technology away.
Why Nvidia Wants the Model Business
Nvidia’s push into open-weight AI reflects a strategic hedge against its dependence on hyperscalers and frontier labs. The chip giant already builds its own Nemotron family of open-weight models, but adoption has been limited. Acquiring Hugging Face would give Nvidia direct access to the largest U.S. developer community for open models — a user base it can steer toward its own chips and software standards.
The urgency is growing. OpenAI this week announced capabilities for its own inference chip, code-named Jalapeño, signaling that major model builders are moving into Nvidia’s core hardware business. If model makers become chip makers, Nvidia needs a foothold in the model-making business to protect its revenue pipeline.
Open-Weight Adoption Remains Small but Strategic
Despite the acquisition frenzy, open-weight model adoption is still in its early stages. Just 6% of companies use open-weight models, according to spending data from Ramp, and only 2% of software engineers work with them, per Jellyfish, a developer tools maker.
Nik Albarran, AI product lead at Jellyfish, said open-weight models are primarily used by companies whose products rely on high-volume, repetitive inference workloads — such as customer service chatbots. “Because these are high-volume tasks with a lot of repetition, an open-weight model can be tuned to answer the questions cheaply,” Albarran told TechCrunch.
For coding and agentic tasks, frontier models from proprietary labs still dominate. They offer easier access, better reasoning capabilities, and in some cases token subsidies that lower the effective cost. Albarran noted that the main driver for open-weight adoption today is control and configurability, not cost savings.
“There are not many companies where that is the case yet … [but] if the prices continue to go up from the frontier labs, more and more companies will be forced to at least consider it,” Albarran said. “When your AI-driven workflows are much more mature, that’s when it makes sense to invest in self-hosting models.”
Stripe Bets on Token Economics
Stripe framed its OpenRouter acquisition around the economics of AI inference. “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,” Stripe co-founder and CEO Patrick Collison said in a statement.
The payments giant’s move signals that open-weight infrastructure is becoming a strategic layer for companies that want to offer AI-powered services without locking into a single proprietary model provider. Open-router-style platforms give businesses the ability to route requests across multiple models, optimizing for cost, latency, and capability.
Fireworks Points Toward Specialized Intelligence
Lin Qiao, CEO of Fireworks — a leading open-weight model router and host often discussed as a potential acquisition target — said her company processes 40 trillion tokens per day, more than either Gemini’s or OpenAI’s APIs. Fireworks’ thesis is that as LLMs proliferate, companies will increasingly train models tailored to their specific needs.
“Every single app company should consider hiring an in-house researcher,” Qiao told TechCrunch. “They can use their product and product data to build their own model. The future is actually specialized intelligence. Literally, every single company should have their own model per use case, and that will happen automatically.”
That vision aligns with the broader shift underway. Chinese companies including Moonshot, DeepSeek, and Alibaba are building cheaper open-weight alternatives, and while adoption remains small, the cost pressure on frontier lab pricing is building.
What Happens Next
If Nvidia confirms the Hugging Face acquisition, expect a wave of defensive moves from other tech giants. Google, Microsoft, and Amazon all have deep relationships with frontier labs, but none controls a comparable open-weight developer ecosystem. Acquiring Fireworks, Together AI, or similar platforms would be the logical countermove.
Watch for two signals in the coming months. First, whether open-weight adoption ticks above single-digit percentages as enterprises mature their AI workflows and frontier lab pricing continues to climb. Second, whether Nvidia uses Hugging Face’s platform to push its own inference standards and chip architectures — a move that would reshape the competitive dynamics between model makers and hardware providers. The open-weight land grab is just beginning, and the companies that secure developer ecosystems now will shape which models run where for years to come.
— David Kim, technology desk, AXO News