Google names 15 Australia and New Zealand startups for its 2026 AI accelerator

Google has named 15 startups for the 2026 Google for Startups Accelerator: Australia & New Zealand, using the program to deepen its role in a regional AI ecosystem that is moving beyond demos into sector-specific products. The company announced the cohort on August 31, 2026, through its Australia blog, describing the accelerator as a 10-week, equity-free hybrid program for Seed and Series A startups using artificial intelligence and machine learning.

The cohort reflects how broad the AI application layer has become. Rather than concentrating on general chatbots, Google selected companies working on design automation, lab research, worker communication, maritime forecasting, dental imaging, child mental health, materials chemistry, supply-chain planning, retail inventory, agriculture, manufacturing memory, spatial data, computer vision, energy flexibility and real-estate trust infrastructure. That spread suggests AI startup formation in Australia and New Zealand is now tied closely to domain data and workflow ownership.

Among the companies named are Blunge, which is building an AI design agent for editable brand work; Bower, which captures laboratory experiments and research decisions in real time; Eyes of AI, which applies imaging AI to dental diagnosis; Mentana, which is developing an agentic autonomous supply-chain manager; Merchmix, which focuses on AI-native inventory operations; RossOps, which captures operational memory for manufacturing; and Unseen, which rebuilds verified commercial real-estate states from source documents to support auditable valuation workflows.

Google framed the announcement as part of its wider Digital Future Initiative and said applications from founders across both countries surged this year. The wording matters because accelerators are increasingly becoming distribution and trust mechanisms for AI platforms. Startups do not only need model access. They need product review, go-to-market advice, responsible deployment practices, cloud infrastructure choices and proof that an AI feature can survive contact with customers in regulated or operationally sensitive industries.

The regional angle is also significant. Australia and New Zealand have strong technology communities, but many local AI companies have to prove they can serve global markets while solving specific local problems. Maritime forecasting, energy-load flexibility, clinical imaging and urban planning all benefit from local data, local regulation and local customer relationships. A global platform company can provide technical scaffolding, but durable products usually come from teams that understand the operating environment.

For Google, the cohort is a way to keep promising AI startups close to its cloud, model and developer ecosystem at an early stage. For founders, the equity-free structure lowers the cost of participation while offering credibility and access to technical guidance. The arrangement also reflects a shift in AI competition: major technology companies are no longer only competing to release foundation models, but also to shape the startup pipelines that turn those models into industry software.

The next test will be whether the 10-week program produces measurable deployment progress rather than polished presentations. Many of the selected companies operate in fields where accuracy, accountability and integration matter more than novelty. If the accelerator helps them improve evaluation, security, data handling and customer adoption, the cohort could become a useful signal that regional AI ecosystems are maturing into practical software markets. If not, it will still show where founders believe AI demand is forming fastest across the Australian and New Zealand economy.