Nvidia has agreed to acquire Hugging Face for $12.9 billion, The Information reported on Wednesday, in a deal that would give the world’s leading AI chipmaker control of one of the most widely used platforms for developing, sharing, and deploying open-source artificial intelligence models.

The report, citing a person familiar with the matter, said negotiations began after Hugging Face received acquisition interest from another potential buyer. Business Insider separately reported that Nvidia had been in talks to acquire the startup.

The transaction, at completion, would represent a major expansion of Nvidia’s reach beyond semiconductors and deeper into the software and model layers of the AI industry.

Hugging Face has become a central hub for developers and researchers working with open-source AI. Its platform allows users to share models, datasets and development tools, making it an important distribution and collaboration layer for the rapidly expanding open-source AI ecosystem.

The acquisition would therefore give Nvidia access to a community and platform that sits much closer to AI model development than the chipmaker’s traditional hardware business.

The potential deal also fits Nvidia’s broader strategy of building an integrated AI technology stack.

Siddy Jobe, a fund manager at Eonopolis Exponential Technologies funds, said Nvidia’s approach to open-source AI made Hugging Face a natural target.

“I think Nvidia is very much a community, a platform-based company, and in that respect, I think Hugging Face fits perfectly within that,” Jobe told CNBC’s “Squawk Box Europe.”

He said Nvidia has identified foundational models as one of several layers of the AI ecosystem it wants to participate in.

“It is clear that Nvidia wants to be integrated in the entire stack vertically, going from energy to foundational models and also to applications,” Jobe said.

That move would mark a significant evolution from Nvidia’s position as primarily a supplier of GPUs and networking equipment. The company has increasingly invested in the infrastructure surrounding its chips, including software, AI startups, cloud infrastructure and model developers. A Hugging Face acquisition would extend that strategy into an open-source platform used by a broad developer community.

Nvidia’s shares rose about 4% in after-hours trading following its blockbuster earnings report on Wednesday, underscoring the extraordinary financial momentum generated by demand for AI infrastructure. The company has also pursued large transactions and strategic investments. Its recent deals include a reported $20 billion licensing agreement with AI chip startup Groq.

A $12.9 billion acquisition of Hugging Face would be considerably larger than many of Nvidia’s recent investments and would signal that the company is prepared to deploy substantial capital to secure strategic positions across the AI stack.

The potential acquisition also raises questions about the future of open-source AI.

Hugging Face has positioned itself as a major advocate and infrastructure provider for open models, while Nvidia has repeatedly sought to support AI developers regardless of whether they use proprietary or open-source models. Bringing Hugging Face under Nvidia’s ownership could give the chipmaker greater influence over how open models are distributed, optimized, and integrated with Nvidia’s hardware and software ecosystem.

The commercial logic for Nvidia is that more AI models and applications ultimately require computing infrastructure, and Nvidia dominates the market for the high-end accelerators used to train and run many of those systems. Owning a major model-development platform could allow Nvidia to deepen that relationship with developers at an earlier stage of the AI development process.

The proposed deal is also expected to give Nvidia a direct connection to a vast pool of AI developers experimenting with models from multiple companies and research communities. That could help Nvidia maintain relevance as AI computing becomes increasingly diversified and as developers explore alternatives to the large proprietary models offered by companies such as OpenAI, Anthropic and Google.

Hugging Face has also recently become involved in a high-profile AI cybersecurity incident, adding another dimension to the proposed transaction. The platform was targeted in a hacking incident that raised concerns about the security implications of increasingly capable AI and cybersecurity systems.

Hugging Face CEO Clément Delangue, a prominent advocate of open-source AI, attributed the incident to engineering mistakes and said his company used an Nvidia version of a Chinese open model to address the attack.

“AI cybersecurity is going to become a huge market in the U.S. and in the world,” Delangue told CNBC earlier this month.

“In this market, probably open models will be kings,” he added.

The episode reveals both the opportunity and risk surrounding the open-source AI ecosystem. Open models can accelerate innovation by giving developers broad access to powerful technology, but their availability can also create new security challenges as increasingly capable models are used for offensive and defensive cyber operations.

However, acquiring Hugging Face would bring those open-source models, developers, and associated tools closer to Nvidia’s own technology stack. That means that well beyond Nvidia’s semiconductor business, the deal would potentially give the company a stronger position across several layers of the AI economy, from computing infrastructure to software, model development and the developer community that turns models into applications.

But the growing concern is whether Nvidia can maintain Hugging Face’s appeal as a relatively open platform after bringing it under corporate ownership.

Hugging Face’s value has been built partly on its role as a neutral meeting point for developers working across different AI models and hardware platforms. Nvidia will have to preserve that ecosystem while finding ways to connect it more closely to its own products.

If the reported $12.9 billion transaction is completed, Nvidia would be making a much larger bet that control of the AI ecosystem cannot be secured through chips alone. The company would instead be positioning itself across the stack, seeking to capture value from the models, software, and developer networks that determine how those chips are ultimately used.

OpenAI CEO Sam Altman has acknowledged that public sentiment toward artificial intelligence and the data centers powering the technology has deteriorated, as growing opposition threatens to complicate the industry’s rapid expansion across the United States.

“Clearly, people hate data centers right now, at least,” Altman told Time in interviews for a cover story published Wednesday. “People are pretty negative on AI.”

Altman’s comments offer a candid assessment of a problem that has become increasingly difficult for AI companies to ignore. The industry is investing hundreds of billions of dollars in computing infrastructure. Still, data centers are facing opposition from communities concerned about electricity demand, water consumption, land use, noise, and the potential impact on utility costs.

The backlash has significantly impacted OpenAI and its technology partners because the development of capable AI models requires enormous amounts of computing power. The company is pursuing a large-scale expansion of data-center capacity, making access to electricity and infrastructure a strategic constraint alongside chips and capital.

Altman has previously acknowledged the tension surrounding new facilities. During an interview last month, he compared opposition to data centers to concerns about living near a nuclear power plant.

“I understand emotionally why people don’t want data centers in their backyard in the same way that I don’t really want a nuclear power plant next to my house, even though I know it’s a super safe thing,” he said.

He was even more direct in March, telling an audience at BlackRock’s U.S. Infrastructure Summit in Washington that “AI is not very popular in the US right now.”

The political response is becoming more consequential. Governors in Pennsylvania and Texas have recently sought to restrict or slow future data-center development, even after previously promoting the facilities as drivers of investment and economic growth.

Texas Gov. Greg Abbott, a Republican, once described the state as the “epicenter of AI development.” More recently, he said that data-center operators had contributed to the backlash by failing to adequately engage with state and local authorities.

“They pop up in places that nobody ever heard of before, until they started the construction,” Abbott said Sunday on ABC’s “This Week.” “And so, they had not been working in collaboration with the state, they’d not been in collaboration with local governments, and so they basically dug their own grave for the problem that’s been caused for them and that’s why they got the backlash they deserve.”

The political shift matters because AI infrastructure depends on favorable local permitting, access to large quantities of electricity, and cooperation from utilities and communities. Resistance could increase the time and cost required to build new facilities, potentially slowing the expansion of computing capacity.

OpenAI and other companies have begun responding to some of the concerns. OpenAI has highlighted efforts to address water and power consumption and has announced community initiatives around some of its data-center projects, including Codex credits for eligible students in Ohio, Georgia and Michigan.

But the criticism has expanded beyond environmental and infrastructure concerns into a broader debate about who benefits from the AI boom.

Chamath Palihapitiya, co-host of the “All-In Podcast,” described the situation as a “powder keg” and warned that resistance could intensify. He said that data centers had become a symbol of the perceived concentration of AI’s economic gains among a relatively small group of technology companies and investors.

That perception could become a major issue as AI companies seek public and government support for massive infrastructure investments. The industry has promoted data centers as sources of construction activity, jobs, tax revenue, and technological competitiveness. Local communities, however, are now focused on the immediate costs, particularly pressure on power grids and questions over whether the economic benefits justify those costs.

The backlash has even entered popular culture. Super Bowl champion Jason Kelce appeared in an advertisement for Garage Beer and Liquid Death that mocked data centers by encouraging people to send their urine to them, a reference to concerns over the water required to operate and cool large computing facilities.

Altman has also faced a more personal manifestation of the hostility surrounding the AI industry. In April, an attacker threw a Molotov cocktail at his San Francisco mansion and later threatened to burn down OpenAI before being arrested.

After the attack, Altman said much of the criticism directed at the industry came from legitimate concerns about the consequences of increasingly powerful technology.

“A lot of the criticism of our industry comes from sincere concern about the incredibly high stakes of this technology,” he wrote on his personal blog.

The criticism now presents a strategic problem for the industry. AI companies need to persuade governments and communities to approve unprecedented amounts of infrastructure while demonstrating that the economic benefits will be broadly distributed and that the costs to electricity systems, water resources and local communities can be managed.

President Donald Trump, who has strongly backed the U.S. AI industry and warned that excessive regulation could undermine America’s position in the global technology race, has also acknowledged the industry’s communications problem.

“I would say that maybe it could use a little public relations help,” Trump told reporters at the White House.