By Lucky Nyathi | PiE Innovation | 6 September 2026
Nvidia is making some of its biggest moves beyond GPUs. The company has agreed to acquire open-source AI platform Hugging Face for $12.93 billion and invested $3.5 billion in MediaTek’s convertible bonds, strengthening its position across AI models, custom chips, edge computing and data-centre infrastructure.
Nvidia’s influence over the artificial intelligence industry is no longer simply about supplying the chips that power AI.
In the space of days, the semiconductor giant has announced two major strategic moves that could give it greater influence over both the software and hardware layers of the AI ecosystem.
The first is its proposed $12.93 billion acquisition of Hugging Face, one of the world’s largest communities for open AI models, datasets and applications.
The second is a $3.5 billion investment in MediaTek, alongside an expanded partnership that will allow MediaTek customers to develop custom AI processors capable of connecting into Nvidia-powered AI infrastructure.
Together, the deals reveal a bigger Nvidia strategy: make Nvidia’s technology part of as many layers of AI computing as possible.
Nvidia Makes a $12.93 Billion Bet on Open AI
On September 3, Nvidia announced an agreement to acquire Hugging Face for $12.93 billion.
The deal includes approximately $11.9 billion for Hugging Face shareholders and an additional equity-based retention programme of up to approximately $1 billion for employees who join Nvidia.
The transaction is expected to close in the first half of 2027, subject to regulatory approvals and other closing conditions.
For Nvidia, however, Hugging Face is about much more than buying another technology company.
Hugging Face has become a major hub for the open-source and open-weight AI movement, allowing developers and organisations to discover, customise, share and deploy AI models.
According to Nvidia, more than 18 million developers, researchers and creators use the platform, while more than 200,000 companies use Hugging Face to work with AI models, datasets and applications.
The platform hosts more than 3 million models, 500,000 datasets and 1 million applications.
That gives Nvidia access to something extremely valuable: the developers building the next generation of AI.
Hugging Face Will Remain Open — For Now
One of the most important aspects of the deal is Nvidia’s commitment to keeping Hugging Face an open platform.
Nvidia says developers will continue to be able to choose their models, frameworks, cloud providers, inference services and computing platforms.
In other words, Nvidia says Hugging Face will not become an Nvidia-only ecosystem.
That distinction matters.
The AI industry is increasingly divided between proprietary AI systems controlled by companies such as OpenAI and Anthropic and open-weight models that developers can download, modify and deploy themselves.
Nvidia’s acquisition puts one of the largest open-AI communities under the ownership of the world’s dominant AI accelerator company.
That could accelerate open AI development — but it also raises questions about how independent the ecosystem will remain over the long term.
The Security Question Behind the Hugging Face Deal
There is another reason this acquisition deserves attention from the AI security community.
Open AI ecosystems create enormous opportunities for developers, but they also introduce new security challenges.
Hugging Face recently faced an AI-related security incident involving malicious activity associated with AI agents, highlighting concerns around model repositories, developer tools, credentials and autonomous systems.
The broader lesson is important:
The more AI models, agents and applications become interconnected, the larger the attack surface becomes.
For organisations deploying open models, security is no longer simply about protecting a server.
It can involve securing:
- AI models and model repositories
- training datasets
- APIs and authentication credentials
- AI agents
- inference infrastructure
- software dependencies
- supply chains
- developer environments
- cloud infrastructure
Nvidia’s acquisition therefore puts the company in an increasingly important position at the intersection of AI development and AI security.
Nvidia Puts $3.5 Billion Into MediaTek
Just days before the Hugging Face announcement, Nvidia revealed another major move.
The company invested $3.5 billion in convertible bonds issued by Taiwanese semiconductor giant MediaTek as part of a record overseas convertible bond offering.
But the investment is about more than money.
Nvidia and MediaTek are expanding their partnership around AI infrastructure, local AI computing and automotive technology.
A key component is Nvidia’s NVLink Fusion platform.
The technology allows customers developing their own custom AI processors — sometimes called XPUs — to connect those chips into Nvidia’s rack-scale AI infrastructure.
That is significant because major cloud companies and AI developers are increasingly looking at custom silicon to reduce costs, improve efficiency and avoid relying entirely on one type of processor.
Rather than fighting that trend outright, Nvidia appears to be positioning its architecture as the infrastructure those custom chips can plug into.
Nvidia Wants to Own the AI Connection
This could prove to be one of Nvidia’s most important strategic shifts.
The company does not necessarily need every AI chip inside a data centre to carry the Nvidia name.
If custom processors from MediaTek and other manufacturers can connect into Nvidia’s infrastructure through technologies such as NVLink Fusion, Nvidia can potentially remain central to the AI data centre even when customers diversify their silicon.
That creates a powerful proposition:
The chip doesn’t have to be Nvidia. The AI infrastructure can still run through Nvidia.
Nvidia and MediaTek say their expanded partnership will cover three major areas:
AI Infrastructure
MediaTek will use NVLink Fusion to help customers develop custom AI infrastructure capable of integrating into Nvidia-connected, rack-scale AI factories.
Local AI Computing
The companies will continue working on Nvidia RTX Spark and DGX Spark products combining Nvidia GPUs with MediaTek system-on-chip technology.
Automotive AI
The partnership will also target AI-powered and software-defined vehicles, extending Nvidia’s technology into the emerging physical AI market.
From GPUs to the Full AI Stack
These deals reveal an important change in Nvidia’s business strategy.
For years, Nvidia’s biggest advantage was its GPUs and the software ecosystem built around them.
Now the company is increasingly positioning itself across the entire AI stack.
AI models → Developer ecosystem → AI software → Custom silicon → Networking → Data centres → Edge computing → Autonomous systems
The Hugging Face acquisition strengthens Nvidia’s position near the AI model and developer layer.
The MediaTek partnership strengthens its position in custom silicon, edge computing and AI infrastructure.
And Nvidia’s existing GPU, networking and software businesses sit underneath much of that ecosystem.
This is why the two announcements are more significant when viewed together.
What This Means for Africa
For Africa, the implications could extend beyond Silicon Valley and Taiwan.
African countries are increasingly looking at sovereign AI infrastructure, local data processing and the development of AI systems that understand African languages and local environments.
Open AI models could reduce some of the barriers faced by African developers because researchers and startups can build on publicly available models rather than developing everything from scratch.
At the same time, the growing demand for AI computing creates a major infrastructure challenge.
Africa needs:
- More data centres
- Reliable electricity
- High-speed fibre
- Affordable cloud computing
- AI computing capacity
- Cybersecurity expertise
- Local datasets
- African-language AI models
- Skilled developers and researchers
This is where Nvidia’s expanding ecosystem becomes relevant.
Companies such as Cassava Technologies are already investing in African AI infrastructure, data centres, cloud services and sovereign computing.
The broader race is therefore not simply about who creates the smartest AI model.
It is increasingly about who controls the infrastructure that allows AI to be built, trained, secured and deployed.
The Bigger Battle: Open AI vs Closed AI
Nvidia’s Hugging Face deal also arrives at an interesting moment in the AI industry.
The world’s largest technology companies are spending enormous amounts of money developing proprietary AI models and custom processors.
At the same time, open-weight models are becoming increasingly capable and accessible.
For developers, this creates more choice.
For Nvidia, it creates another opportunity.
If more developers choose open models, those models still require enormous amounts of computing power to train and run.
Nvidia can potentially benefit from that demand regardless of which model ultimately wins.
The model may change. The infrastructure still needs to exist.
PiE Perspective
Nvidia’s latest moves suggest that the next phase of the AI race will not be won by one technology alone.
The battle is moving from “Who has the best AI model?” to a much bigger question:
Who controls the ecosystem that allows millions of people and organisations to build with AI?
Hugging Face gives Nvidia a much stronger relationship with the open AI developer community.
MediaTek strengthens its reach into custom silicon, edge computing and automotive AI.
Its existing GPU, networking and software businesses provide the infrastructure underneath those ambitions.
That makes Nvidia’s strategy increasingly clear: don’t just sell the machines that power the AI revolution — become part of the platform on which the revolution is built.
For Africa, the lesson is equally important.
The continent should not only be thinking about using AI. It needs to be thinking about building AI infrastructure, developing local models, protecting its data and training the people who will control the technology.
The next digital divide may not simply be who has internet access.
It could be who has access to AI compute.
