The AI conversation has always really revolved around a few key players: OpenAI, Anthropic Google, Meta, Nvidia and Microsoft. That’s not to ignore smaller companies using artificial intelligence, but there’s no denying the fact that the market has been quite condensed, at least at the very top. While these positions have shuffled around a little over the last decade or so, the big question has generally been about which company could build the most powerful model, and who would come out on top.
But another conversation that’s gaining momentum, and one that could have just as much influence over the future of the industry, is centred on open-source AI.
It’s definitely not a new idea – open models have been around for years – but they’re becoming increasingly relevant as businesses, developers and governments are being forced to think very seriously about who controls AI, how much it costs and who gets to benefit from it. Even Nvidia, one of the most powerful companies in the AI industry, has made a significant bet on the open-model ecosystem.
So, why is open-source AI suddenly such a big deal, and what does its growing popularity tell us about where the industry is heading?
AI Is Becoming Too Important To Leave In the Hands of a Few Companies
One of the biggest arguments for open-source AI is accessibility, and it’s a really important one. When companies like OpenAI and Anthropic develop proprietary models, users generally access them through their platforms or APIs. That may be convenient, but it also means that businesses are relying on technology that they don’t own or control.
But open models offer another route that is potentially more attractive. In theory, developers can download available models, adapt them for particular purposes and, depending on the licence, run them on their own infrastructure. That creates opportunities for organisations that want more control over how their AI systems work, where their data goes and how much they spend, without having to do run everything they do through a big tech giant like OpenAI.
It also means innovation doesn’t have to happen exclusively inside the world’s biggest AI laboratories, allowing for more of an equal playing field, in some respects. Smaller companies, independent developers and research teams can build on existing models rather than starting from scratch, potentially reducing some of the barriers to entry.
As AI becomes embedded in more aspects of business, increasing accessibility to normal people is really important. The question is no longer simply which chatbot gives the best answer, but how organisations can build AI into their products and operations without becoming entirely dependent on someone else’s technology.
More from Artificial Intelligence
- EY, Deloitte, KPMG And PwC Are Racing To Put AI Agents On Audits While Training Staff To Doubt Them: What’s Going On?
- Can Europe Build Its Own Speech AI? Reson8 Is Giving Its Answer Away
- Has 2026 Been The Year We Finally Fell Out Of Love With AI?
- The Data Centre Dilemma: Can We Really Complain About Data Centres While Simultaneously Demanding More Of The Technology They Power?
- Why Are AI Companies So Interested In Giving Their Models Memory?
- What Are Small Language Models And How Are They Different From Large Language Models?
- Disney Is Teaching AI Its Creative Style – What Does This Mean For The Future Of Entertainment?
- Your Patient Records Could Become AI Training Data – What Do Privacy Rules Say About De-Identified Health Data?
Now We Have To Worry About the Cost of AI
Now, we can’t forget about the most obvious reason behind the popularity of open-source AI, and that’s the price tag that’s in the mix. Developing and operating advanced AIopeuters found that 59% of executives at large companies planned to increase their use of open-weight AI models over the following year. The same research found that 59% said their AI spending had significantly exceeded expectations or budgets in the previous year.
Of course, open models aren’t automatically free, and organisations still need computing power, infrastructure and technical expertise to run them. But, despite this, they can offer a way to reduce reliance on expensive external services, particularly for repetitive or specialised tasks where a smaller model may be perfectly adequate.
For businesses, that could mean using a powerful proprietary model for particularly demanding work while also relying on cheaper open alternatives for other applications. It doesn’t necessarily need to be all about choosing one over the other.
Why Does Nvidia Have a Stake In Open-Source AI?
Interestingly, some of the strongest support for open AI models comes from companies that have a considerable commercial interest in seeing the ecosystem grow. Indeed, it’s pretty counterintuitive, at least at first glance.
Nvidia is the most obvious example people point to, and even more so at the moment. While the company is best known for the chips powering much of the AI boom (and currently Jensen Heung and his alleged affair with a Chinese spy), it’s also invested heavily in open models and the infrastructure surrounding them. In September 2026, Nvidia announced its agreement to acquire Hugging Face, one of the most important platforms for sharing and developing AI models, for almost $13 billion.
And while it may not seem like an obvious trajectory, the logic is actually quite straightforward. The more organisations experimenting with, adapting and deploying AI models, the greater the demand for the computing infrastructure needed to run them. That means that open models can help expand that market beyond a relatively small number of major AI laboratories.
Nvidia’s recent leadership upheaval has also, unsurprisingly, brought renewed attention to its role in the open-model movement. Jensen Huang was a prominent advocate of wider access to AI models, and his departure from Nvidia is now making people question whether the company’s commitment to open-source would remain a priority.
And this is one part of the conversation that’s often missed: that is, open-source AI isn’t just an ideological argument. It can also be significantly commercial, with a fair bit up for grabs for relevant figures and companies who play their cards right.
Does More AI Mean Better AI?
It’s not a straightforward issue (nothing is); making AI more accessible inevitably brings its own complications. Giving developers greater freedom to modify and deploy models can encourage experimentation, competition and innovation, but in the same breath, it also means that models can be adapted for purposes that their original developers never intended. Without the same level of centralised oversight, identifying who is responsible when something goes wrong can become more complicated, and responsibility is a huge issue to be considered in the world of modern AI.
Following the same lines, there are also questions about transparency. Making a model’s weights available doesn’t necessarily mean its training data, development process or every component is publicly accessible, and that’s important when it comes to when organisations are making decisions about things like security, accountability and trust.
And while open models are becoming increasingly capable, they aren’t automatically the best choice for every application. Some businesses will still prefer the convenience, performance and support offered by proprietary services, and there’s nothing intrinsically wrong with that.
Perhaps the most interesting thing about the growing open-source conversation is that it challenges the idea that the AI industry will eventually be dominated by one or two companies offering the most powerful models. Much like with anything, there’s alot more nuance involved in the situation. Indeed, some organisations will build their businesses around proprietary models, while others will favour open alternatives that give them greater control. Many will probably use a combination of both, depending on what they’re trying to achieve.
And that’s probably a healthier direction for the industry overall. More competition, more choice and more people able to contribute to the development of AI could help prevent too much control from accumulating in too few hands. So, the big question isn’t necessarily whether open-source AI will replace ChatGPT, Claude or the other major proprietary models. It’s probably going to end up being more about whether the future of AI will be shaped by a much broader community of people and organisations than the handful of companies currently leading the charge.
In my opinion? One can only hope!
