Why Nvidia Is Backing Open Source And What It Means For The Rest Of Us
Why Nvidia's Open Secure AI Alliance is a commercial play against the big model labs — and why it points Australian businesses towards cheaper, more flexible and more sovereign AI options.
By Andrew Lai · First published by iTWire. Republished with permission.
GUEST OPINION: Jensen Huang has run Nvidia for more than 30 years and never once posted on Twitter, now X, in the two decades the platform has existed. Late last month he finally joined, and he did not do it to share a photo of his lunch. His first post was a public letter arguing for open AI models followed up with a world changing announcement that has drawn the battlelines for the future of AI.
Nvidia launched the Open Secure AI Alliance, a group of close to 40 technology companies committed to building open-source tools that help organisations defend against AI-powered cyberattacks. The membership list is a roll call of the industry, including Microsoft, IBM and SpaceX. The stated purpose is cybersecurity, which happens to be a genuinely good reason. As I wrote recently after the first AI-run ransomware attack, defenders need open tools they can actually inspect and adapt. But the significance of this goes well beyond security.
The most telling thing about the alliance is who is not in it. OpenAI, Anthropic and Google are all absent, and so is Meta. Between them, these companies spend well in excess of $US100 billion a year on Nvidia chips, so this is not a falling out. These are frenemies, dependent on each other and competing at the same time, and Huang has just drawn a line with the big model labs on the other side of it.
To understand why, you have to be clear about how Nvidia actually makes its money. It does not primarily sell AI models. It sells the chips and infrastructure that every AI model runs on. Whether a business uses a model from OpenAI, one from Google, a Chinese open-weight model or something it built in-house, it still needs enormous amounts of computing power, and Nvidia wants that computing to happen on Nvidia hardware. From where Huang sits, a large and varied field of AI companies is far better for business than a market carved up between three or four giants.
That is the strategy behind the open-source push. The wording in Nvidia's own announcement gives it away, framing the choice as one between security tools locked inside a few opaque systems and tools built on open models that defenders can inspect and modify. Read it again with a commercial eye and it is an argument against a future where a handful of labs become the gatekeepers of advanced AI. Nvidia would much rather deal with a diverse crowd of smaller AI companies than a couple of superpowers big enough to set the market's terms.
And those superpowers are starting to build their own chips. The more a single lab designs its own silicon, the less it needs Nvidia, which is a direct threat to the core business. That threat gets sharper as these companies head towards the public markets. Anthropic filed confidentially for a listing on the first of June, reportedly targeting late this year, and OpenAI is weighing its own float. It is no coincidence that Huang has chosen this moment, in the months before those debuts, to rally the rest of the industry around open alternatives.
None of this requires you to see Huang as either hero or villain. He is doing what any sharp operator would do, which is protect his position by widening the field. The interesting part is what his self-interest sets in motion for everyone else.
For the rest of us, this is bigger than a squabble between tech giants. When a group of the world’s most valuable companies in the world throws their weight behind open source, two things follow. The first is a wave of open products and tools from the broader industry, because that is now where the momentum and the money are pointing. The second is more consequential for businesses. Companies and particularly large enterprises who have spent years treating open-source software with suspicion for decades, are going to start taking open source alternatives much more seriously.
That matters for Australian businesses in particular. As per the recent speech by Prime Minister Anthony Albanese, sovereign capability is the theme of the day and open source models and software are an enabler. Open source enables self-hosted infrastructure, full control over the location of data, and most importantly, an ability to diversify AI models.
Once the domain of tech nerds, the Nvidia led declaration on open source signals a strategic shift in the AI era. Fortunately for Australian business, it means not only cheaper and more flexible AI and software options, but a pathway to improved sovereign capability.