for early access to the latest igaming videos!
and get the latest igaming news first!
For years, the global race to dominate frontier AI has been framed as a duel between the heavyweights: the US, with OpenAI and Anthropic, and China, with DeepSeek and Baidu. But another contender is quietly assembling its arsenal from an unexpected corner of the tech world — Melbourne’s CBD — backed by billionaire Stake co-founder Ed Craven and engineered by returning systems architect Dave Lemphers.
The project, known as MainCode, aims to create something Australia has never had: a fully sovereign, trained-from-scratch large language model running entirely on local infrastructure. If it succeeds, it would reshuffle the country’s position in the global AI hierarchy and potentially influence the direction of Australia’s online technology industries.
The Quiet Homecoming Behind a Big Bet
Lemphers’ return to Melbourne after more than a decade in the US feels less like a homecoming and more like the opening move of a long-planned campaign. After building robotics companies, climbing the ranks at Microsoft, and selling AI startups in Silicon Valley and Austin, the engineer has re-emerged with a mission: to establish the first frontier-scale LLM trained entirely on Australian soil.
He’s not doing it alone. Craven — best known for co-founding Stake.com and Kick.com, two companies that reshaped the global online betting and streaming landscape — has supplied the capital and independence MainCode says is essential for sovereignty. With Craven as majority investor, the company has been built outside government funding cycles, venture capital pressure, or foreign cloud dependencies.
MainCode’s leadership adds another layer to the story. Its CEO previously served as CTO of EasyGo, the parent company behind both Stake and Kick — placing someone with deep experience in high-scale, real-time digital infrastructure at the helm of Australia’s most ambitious AI effort to date.
The team Lemphers has assembled is an academic who’s-who of machine learning talent drawn largely from Melbourne’s universities, with one key exception: an AI researcher based in Seattle. Her presence hints at an ambition that extends beyond building a domestic tool — MainCode wants to compete.
A Race Australia Has Long Avoided
Australia has historically underinvested in foundational tech infrastructure. The “brain drain” narrative is so common it’s practically a cliché: promising engineers leave for California, London, Shenzhen. Lemphers was one of them.
Now, he and Craven are trying to flip that script.
MainCode’s internal mantra — “own the main code or someone else owns you” — encapsulates a broader concern shared by many policymakers: Australia’s reliance on foreign AI systems has reached a point where data sovereignty, national security, and commercial independence can no longer be separated from technological infrastructure.
Foreign LLMs are increasingly absorbing sensitive corporate data: customer support transcripts, internal documents, compliance workflows. Many Australian companies either don’t know or don’t care that this data may cross borders or be used to improve models they don’t control.
MainCode is positioning itself as the antidote — a locally governed, fully transparent alternative that avoids the “invisible contamination” problem inherent in systems trained on opaque foreign datasets.
But Why Is the Founder of Stake.com Funding an LLM?
This is where motivations become more complex — and less certain.
Craven’s businesses have rarely stayed within the lines set by existing industries. His companies have repeatedly built custom digital infrastructure to accelerate, disrupt, and outperform competitors. It’s natural, then, that observers wonder whether a sovereign LLM could eventually connect to Craven’s existing ecosystem.
At the same time, the investment may be far more straightforward: frontier AI is becoming one of the defining technological assets of the next decade, and Craven — like many entrepreneurs — may simply see MainCode as a high-upside opportunity with national significance.
Both readings are plausible.
How Stake and Kick Might, in Theory, Use a Homegrown LLM
(If that becomes part of the strategic direction)
If MainCode’s technology ever intersects with Craven’s entertainment or iGaming ventures, several hypothetical uses emerge:
- Hyper-personalised gambling risk modelling and harm modelling
Real-time behavioural analysis across millions of betting patterns — performed entirely on Australian-owned infrastructure. - Automated content moderation for Kick.com
Faster, safer, locally aligned moderation without depending on foreign platforms. - Creator tools to rival Twitch’s backend
AI-powered overlays, automated clipping, and live translation built directly into the streaming stack. - A proprietary recommendation engine
An algorithm shaped around Kick’s priorities rather than those of a foreign LLM provider. - A closed-loop ecosystem independent of OpenAI
Experiments and product innovation that remain entirely sovereign.
These possibilities remain illustrative — not stated by MainCode, not confirmed by Craven — but the overlap between infrastructure and opportunity is hard to dismiss.
A Sovereign LLM With Two Futures
If MainCode succeeds, Australia gains more than an AI model. It gains local autonomy in a sector dominated by two superpowers. It could attract expat talent, reduce reliance on imported algorithms, and open a new technological frontier.
Yet the intersection with Craven’s background invites an ongoing question. Is this a play to strengthen Australia’s digital independence? Or simply a high-conviction investment in next-generation infrastructure? Or both?
Lemphers emphasises that the mission is to build a safer, transparent AI system aligned with Australian interests — and on its face, MainCode reflects exactly that.
But with a CEO who once oversaw the tech stack of Stake and Kick, and a majority investor known for building fast-moving, vertically integrated entertainment systems, the longer-term path remains unwritten.
Australia may soon find itself with a homegrown frontier LLM.
The open question is not just whether it works — but which vision it ultimately advances.
Disclosure
Any discussion of potential applications of MainCode’s technology within Stake, Kick, or other Craven-backed ventures is hypothetical and not based on public statements or confirmed plans by the company.
