2024 saw a shift in how AI in Australia is being perceived, funded and built. Now representing a quarter of venture capital funding, it is becoming clearer where founders might have opportunity to create generational businesses, and equally where investors can deliver outsized returns.
The evolution of the stack
The AI stack is evolving at a breakneck pace since the public launch of ChatGPT in November 2022, which saw decades of behind-the-scenes R&D pushed into the mainstream. The following two years were marked with a Cambrian explosion of AI development, notably leading to a convergence of model providers, fertile ground for startups at the application layer, and increasing competition amongst Big Tech.
The foundation layer is becoming increasingly oligopolistic and mostly dominated by Big Tech who drove many mega deals (Microsoft x OpenAI, Amazon x Anthropic). These model providers are aggressively innovating and pushing the limits of existing transformer architectures and scaling laws. (Arguably) 2024’s biggest development was OpenAI’s o1 launch which showcased reasoning capabilities. This new frontier allows models to “think” before responding, moving beyond just pattern recognition. It’s hard not to be optimistic about the use cases this unlocks - shifting compute allocation from pre-training to inference allows founders to tackle more complex problems like those in STEM fields.
The application layer saw substantial activity in the past year, representing 35% of deal count - investors pivoted from skeptical "wrapper" dismissals to recognising genuine value creation. The AI ecosystem is maturing, investment frameworks are being reshaped, and investors are questioning whether it makes sense to be assessing these businesses like vertical SaaS. The market and founder thesis must always hold true, but the customer utility is fundamentally different and begins to diverge on finer points like pricing - how will charging on a per-seat evolve as software replaces those seats?
Australians do software very well - we don’t expect the quality to change, but founders will need to adapt to the way customers purchase, engage and extract value.
2024, the year of the agent
AI agents took centre stage this year, showcasing capabilities that far surpassed traditional chatbots, which still placed the onus of the task on the user. These agents demonstrated an ability to perceive, plan, and act with minimal human oversight - effectively replacing both software and labour costs.
2024 was a big year for designing and prototyping - we’re excited to see the impact these agents will have in production, particularly as they’re supported by emerging reasoning and multi-modal capabilities. Take fraud detection as an example, concurrently analysing different data formats (text, tables, images) allows for context-rich interpretation and enhances the accuracy of identifying complex and evolving fraud patterns.
Primary questions emerging from this boom revolve around how this technology will be adopted, what defines a superior product, and whether competition drives the value of these agents upwards, or into a race to the bottom.
Have we reached peak (anti) hype?
The initial exuberance experienced post-ChatGPT's launch was met with a testing period of fast pivots, product discovery challenges, and difficulties for some in securing downstream capital. Investors are becoming more discerning around revenue traction, driven by ‘pilot-happy’ behaviours from enterprises with an ‘AI mandate’, renewal cycles will be crucial to determine whether real problems are being solved.
In 2024, we saw a more balanced approach, tempering enthusiasm with caution. Now, the bar has been raised; being 'AI-first' is no longer enough. With AI becoming central to most modern tech stacks, investors will look to scrutinise a company’s competitive edge. Going into 2025, founders that employ proprietary architectures, GTM excellence and product elegance will be best placed to outcompete the crowd.
We’re also likely to see increased M&A activity going into 2025. In Australia, exceptional talent pools flocked to AI startups, presenting attractive acquisition targets for enterprises late to the AI game, or acquihires for competitors. Coupled with talks of offshore deregulation, pending domestic merger reforms and a (prospective) easing rate environment, we may see more early-stage deals transact faster, but without the long term value creation required by venture capital. Time will tell, and if this is the case we hope durable AI companies servicing high value customers becomes the primary focus of Australian founders.
Insight used was drawn from data gathered as part of our fourth State of Australian Startup Funding Report, amongst other sources.
Subscribe to our newsletter for updates delivered directly to your inbox.
%20(1).jpg)
