Quick Take
- Australia’s AI startups secured AUD $1.3 billion in 2024 VC funding, representing 25-30% of total deals
- Nation lacks flagship large language model to compete with GPT-4 or Claude 3.5 Sonnet
- Kangaroo LLM project targets 4.2 million Australian websites for homegrown AI model
- Global AI startups captured nearly $50 billion in Q2 2025 venture capital globally
- Universities excel in AI research but focus on adaptation rather than foundational model development
Australia’s artificial intelligence sector attracted significant venture capital investment in 2024, yet the nation remains dependent on international AI models amid growing sovereignty concerns, according to industry trackers.
Australia’s AI startup ecosystem hit a major milestone last year, but the numbers tell only part of the story. Venture capital funding for AI-linked startups reached approximately AUD $1.3 billion in 2024, with AI representing around 25–30% of total deals according to industry trackers. However, Australia’s artificial intelligence ambitions face a reality check.
The country lacks a flagship large language model to compete with GPT-4 or Claude 3.5 Sonnet. Research institutions and businesses rely almost entirely on international AI models, creating what experts call a sovereignty gap that could undermine long-term competitiveness.
Kangaroo LLM Emerges as Australia’s AI Independence Play
The Kangaroo LLM represents Australia’s most ambitious push toward AI independence. Backed by a consortium including Katonic AI, Rack Corp, NEXTDC, Hitachi Vantara, and Hewlett Packard Enterprise, this open-source project aims to create the first major Australian-built language model.
The initiative targets a critical weakness that many overlook. International models consistently underperform on Australian English. Preliminary benchmark testing from UNSW researchers suggests LLMs struggle with Australian sarcasm detection, with F-scores noticeably lower than standard sentiment benchmarks.
“We’re creating an AI foundation for Australian digital independence,” the Kangaroo LLM consortium has stated.
The project has identified 4.2 million Australian websites for data collection, focusing initially on 754,000 sites. Their VegeMighty Dataset will feed through the Great Barrier Reef Pipeline to train models that understand local humour, slang, and legal frameworks.
International Giants Dominate Despite Local Limitations
International giants currently rule Australia’s AI landscape. Claude 3.5 Sonnet became available in AWS’s Sydney region in early 2025 and powers applications from customer service to scientific research. GPT-4 and LLaMA 2 dominate university labs and corporate innovation centres.
Australian researchers have applied LLMs, including cloud-hosted tools like Claude, to environmental data projects such as whale acoustic detection, showing accuracy improvements over traditional methods. In detecting minke whales—a significant jump from traditional methods at 76.5%. Yet this success highlights Australia’s reliance on foreign AI infrastructure.
Data sovereignty concerns plague enterprise adoption. Privacy law reforms in 2024 introduced new AI transparency requirements, forcing organisations to evaluate their model selection and deployment strategies carefully.
Venture Capital Shifts Toward Infrastructure-Heavy Investments
Australia’s venture capital scene reflects global AI investment patterns. After experiencing a record boom in 2021, funding dropped by more than half during the global cooldown. However, AI startups in deep tech and climate solutions saw capital raising lift in Q1 2025 to the highest level since 2022.
Globally, AI startups captured nearly half of all venture capital in Q2 2025, with estimates placing the figure close to $50 billion. This shift toward fewer but larger deals creates challenges for smaller fund managers who struggle to compete for mega-rounds.
“The clock is ticking for funds that raised capital during the boom years,” notes a recent Pitcher Partners analysis. “There’s tens of billions in dry powder needing deployment by the end of 2025.”
Australian investors now face greater choice, with global private market managers increasingly seeking access through placement agents rather than maintaining a direct local presence. This trend brings co-investment opportunities in high-profile AI funding rounds across machine learning and automation.
Research Excellence Battles Infrastructure Constraints
Australian universities punch above their weight in AI research, but focus on evaluation, fairness, and domain adaptation rather than foundational model development. This strategic positioning reflects resource constraints but also creates opportunities.
Macquarie University demonstrates this approach by fine-tuning BERT variants for medical applications, achieving top scores in international competitions. CSIRO Data61 leads in agent-based systems and privacy-preserving AI, establishing Australia’s strength in applied research.
The CommBank Centre for Foundational AI, launched in late 2024 through a University of Adelaide partnership, represents significant industry investment in financial AI applications. However, the focus remains on adaptation rather than building new architectures.
Implementation Delays Challenge Ambitious Timeline
Despite ambitious goals, Kangaroo LLM faces significant hurdles. Initially slated for an October 2024 launch, the project remains in data collection phase as of August 2025. Legal and privacy concerns delayed website crawling, and no model weights, benchmarks, or production deployments have been published.
The project operates as a nonprofit organisation with approximately 100 volunteers providing over 10 full-time equivalent labour hours. Corporate clients and potential government grants provide funding, but no significant public or private investment has been announced.
Strategic Implications for Australian Business
Successful AI adoption in Australia requires striking a balance between global capabilities and local compliance. Organisations must navigate data sovereignty requirements while leveraging superior international models to gain competitive advantage.
The government’s risk-based AI policy framework mandates transparency and accountability for high-risk applications. This regulatory landscape affects model selection and deployment strategies, particularly for financial services and healthcare applications.
Investment opportunities span from application-layer companies to infrastructure plays. The shift toward mega-rounds favours established players, but specialised sectors like medical AI and privacy-preserving systems offer niches for focused investment.
Kangaroo LLM’s eventual success could reshape Australia’s AI sovereignty. However, the current reality demands pragmatic approaches to integrating international models while building local capabilities.
